├── .Rbuildignore ├── .gitattributes ├── .github ├── .gitignore ├── CONTRIBUTING.md ├── issue_template.md ├── pull_request_template.md └── workflows │ └── R-CMD-check.yaml ├── .gitignore ├── CONTRIBUTING.md ├── CRAN-SUBMISSION ├── DESCRIPTION ├── LICENSE ├── NAMESPACE ├── NEWS.md ├── R ├── biomartr-tools.R ├── connection-tools.R ├── db-get-tools.R ├── db-setup-tools.R ├── download-tools.R ├── entrez-tools.R ├── extract-tools.R ├── filepath-tools.R ├── gb-get-tools.R ├── gb-setup-tools.R ├── log-tools.R ├── mock-tools.R ├── print-tools.R ├── rentrez-wrappers.R ├── restez.R ├── status-tools.R ├── test-tools.R └── wrappers.R ├── README.Rmd ├── README.md ├── _pkgdown.yml ├── codemeta.json ├── cran-comments.md ├── data └── record.rda ├── examples ├── count_db_ids.R ├── db_delete.R ├── demo_db_create.R ├── entrez_fetch.R ├── gb_definition_get.R ├── gb_extract.R ├── gb_fasta_get.R ├── gb_organism_get.R ├── gb_record_get.R ├── gb_sequence_get.R ├── gb_version_get.R ├── is_in_db.R ├── list_db_ids.R ├── restez_path_get.R ├── restez_ready.R └── restez_status.R ├── inst └── CITATION ├── logo.png ├── man ├── add_rcrd_log.Rd ├── cat_line.Rd ├── char.Rd ├── check_connection.Rd ├── cleanup.Rd ├── connected.Rd ├── connection_get.Rd ├── count_db_ids.Rd ├── db_create.Rd ├── db_delete.Rd ├── db_download.Rd ├── db_download_intern.Rd ├── db_sqlngths_get.Rd ├── db_sqlngths_log.Rd ├── demo_db_create.Rd ├── dir_size.Rd ├── dwnld_path_get.Rd ├── dwnld_rcrd_log.Rd ├── entrez_fasta_get.Rd ├── entrez_fetch.Rd ├── entrez_gb_get.Rd ├── extract_accession.Rd ├── extract_by_patterns.Rd ├── extract_clean_sequence.Rd ├── extract_definition.Rd ├── extract_features.Rd ├── extract_inforecpart.Rd ├── extract_keywords.Rd ├── extract_locus.Rd ├── extract_organism.Rd ├── extract_seqrecpart.Rd ├── extract_sequence.Rd ├── extract_version.Rd ├── file_download.Rd ├── filename_log.Rd ├── flatfile_read.Rd ├── gb_build.Rd ├── gb_definition_get.Rd ├── gb_df_create.Rd ├── gb_df_generate.Rd ├── gb_extract.Rd ├── gb_fasta_get.Rd ├── gb_organism_get.Rd ├── gb_record_get.Rd ├── gb_sequence_get.Rd ├── gb_sql_add.Rd ├── gb_sql_query.Rd ├── gb_version_get.Rd ├── gbrelease_check.Rd ├── gbrelease_get.Rd ├── gbrelease_log.Rd ├── has_data.Rd ├── identify_downloadable_files.Rd ├── is_in_db.Rd ├── last_add_get.Rd ├── last_dwnld_get.Rd ├── last_entry_get.Rd ├── latest_genbank_release.Rd ├── latest_genbank_release_notes.Rd ├── list_db_ids.Rd ├── message_missing.Rd ├── mock_def.Rd ├── mock_gb_df_generate.Rd ├── mock_org.Rd ├── mock_rec.Rd ├── mock_seq.Rd ├── ncbi_acc_get.Rd ├── predict_datasizes.Rd ├── print.status.Rd ├── readme_log.Rd ├── record.Rd ├── restez.Rd ├── restez_connect.Rd ├── restez_disconnect.Rd ├── restez_path_check.Rd ├── restez_path_get.Rd ├── restez_path_set.Rd ├── restez_path_unset.Rd ├── restez_ready.Rd ├── restez_rl.Rd ├── restez_status.Rd ├── search_gz.Rd ├── seshinfo_log.Rd ├── setup.Rd ├── slctn_get.Rd ├── slctn_log.Rd ├── sql_path_get.Rd ├── stat.Rd ├── status_class.Rd └── testdatadir_get.Rd ├── other ├── demonstration_1.R ├── dev.R ├── generate_records_data.R ├── make_sure_things_work.R ├── phylotar_demo.R ├── random_file_tester.R ├── rodent_build_times.rds ├── rodent_db.R ├── spell_check.R └── viral_lookup.R ├── paper ├── outline.png ├── paper.bib ├── paper.html └── paper.md ├── pkgdown └── favicon │ ├── apple-touch-icon-120x120.png │ ├── apple-touch-icon-152x152.png │ ├── apple-touch-icon-180x180.png │ ├── apple-touch-icon-60x60.png │ ├── apple-touch-icon-76x76.png │ ├── apple-touch-icon.png │ ├── favicon-16x16.png │ ├── favicon-32x32.png │ └── favicon.ico ├── revdep ├── .gitignore ├── README.md ├── cran.md ├── email.yml ├── failures.md └── problems.md ├── tests ├── test-all.R └── testthat │ ├── data │ ├── records.RData │ └── release_notes_gb224.RData │ ├── test-biomartr-tools.R │ ├── test-connection-tools.R │ ├── test-db-tools.R │ ├── test-entrez-tools.R │ ├── test-extract-tools.R │ ├── test-filepath-tools.R │ ├── test-gb-setup-tools.R │ ├── test-get-tools.R │ ├── test-log-tools.R │ ├── test-mock-tools.R │ ├── test-print-tools.R │ ├── test-rentrez-wrappers.R │ ├── test-status-tools.R │ └── test-test-tools.R └── vignettes ├── 1_rodents.Rmd ├── 1_rodents.Rmd.orig ├── 2_search_and_fetch.Rmd ├── 2_search_and_fetch.Rmd.orig ├── 3_parsing.Rmd ├── 3_parsing.Rmd.orig ├── 4_phylotar.Rmd ├── 4_phylotar.Rmd.orig ├── 5_tips_and_tricks.Rmd ├── 5_tips_and_tricks.Rmd.orig ├── precompile.R └── restez.Rmd /.Rbuildignore: -------------------------------------------------------------------------------- 1 | ^codemeta\.json$ 2 | ^restez_test$ 3 | ^paper$ 4 | ^other$ 5 | ^rodents_copy$ 6 | ^rodents$ 7 | ^vignettes$ 8 | ^examples$ 9 | ^.*\.Rproj$ 10 | ^\.Rproj\.user$ 11 | ^README\.Rmd$ 12 | ^README-.*\.png$ 13 | ^docs$ 14 | ^README\.html$ 15 | ^CONTRIBUTING\.md$ 16 | ^logo\.png$ 17 | ^_pkgdown\.yml$ 18 | ^\.gitattributes$ 19 | ^pkgdown$ 20 | ^cran-comments\.md$ 21 | ^\.github$ 22 | ^CRAN-SUBMISSION$ 23 | ^revdep$ 24 | -------------------------------------------------------------------------------- /.gitattributes: -------------------------------------------------------------------------------- 1 | *.html linguist-documentation=true 2 | 3 | -------------------------------------------------------------------------------- /.github/.gitignore: -------------------------------------------------------------------------------- 1 | *.html 2 | -------------------------------------------------------------------------------- /.github/CONTRIBUTING.md: -------------------------------------------------------------------------------- 1 | # CONTRIBUTING # 2 | 3 | ### Fixing typos 4 | 5 | Small typos or grammatical errors in documentation may be edited directly using 6 | the GitHub web interface, so long as the changes are made in the _source_ file. 7 | 8 | * YES: you edit a roxygen comment in a `.R` file below `R/`. 9 | * NO: you edit an `.Rd` file below `man/`. 10 | 11 | ### Prerequisites 12 | 13 | Before you make a substantial pull request, you should always file an issue and 14 | make sure someone from the team agrees that it’s a problem. If you’ve found a 15 | bug, create an associated issue and illustrate the bug with a minimal 16 | [reprex](https://www.tidyverse.org/help/#reprex). 17 | 18 | ### Pull request process 19 | 20 | * We recommend that you create a Git branch for each pull request (PR). 21 | * Look at the Travis and AppVeyor build status before and after making changes. 22 | The `README` should contain badges for any continuous integration services used 23 | by the package. 24 | * We recommend the tidyverse [style guide](http://style.tidyverse.org). 25 | You can use the [styler](https://CRAN.R-project.org/package=styler) package to 26 | apply these styles, but please don't restyle code that has nothing to do with 27 | your PR. 28 | * We use [roxygen2](https://cran.r-project.org/package=roxygen2). 29 | * We use [testthat](https://cran.r-project.org/package=testthat). Contributions 30 | with test cases included are easier to accept. 31 | * For user-facing changes, add a bullet to the top of `NEWS.md` below the 32 | current development version header describing the changes made followed by your 33 | GitHub username, and links to relevant issue(s)/PR(s). 34 | 35 | ### Code of Conduct 36 | 37 | Please note that the restez project is released with a 38 | [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By contributing to this 39 | project you agree to abide by its terms. 40 | 41 | ### See rOpenSci [contributing guide](https://ropensci.github.io/dev_guide/contributingguide.html) 42 | for further details. 43 | 44 | ### Discussion forum 45 | 46 | Check out our [discussion forum](https://discuss.ropensci.org) if you think your issue requires a longer form discussion. 47 | 48 | ### Prefer to Email? 49 | 50 | Email the person listed as maintainer in the `DESCRIPTION` file of this repo. 51 | 52 | Though note that private discussions over email don't help others - of course email is totally warranted if it's a sensitive problem of any kind. 53 | 54 | ### Thanks for contributing! 55 | 56 | This contributing guide is adapted from the tidyverse contributing guide available at https://raw.githubusercontent.com/r-lib/usethis/master/inst/templates/tidy-contributing.md 57 | 58 | 59 | -------------------------------------------------------------------------------- /.github/issue_template.md: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 |
Session Info 6 | 7 | ```r 8 | 9 | ``` 10 |
11 | 12 | 13 | -------------------------------------------------------------------------------- /.github/pull_request_template.md: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | ## Description 8 | 9 | 10 | ## Related Issue 11 | 14 | 15 | ## Example 16 | 18 | 19 | 21 | 22 | 23 | -------------------------------------------------------------------------------- /.github/workflows/R-CMD-check.yaml: -------------------------------------------------------------------------------- 1 | # Workflow derived from https://github.com/r-lib/actions/tree/v2/examples 2 | # Need help debugging build failures? Start at https://github.com/r-lib/actions#where-to-find-help 3 | on: 4 | push: 5 | branches: [main, master] 6 | pull_request: 7 | branches: [main, master] 8 | 9 | name: R-CMD-check 10 | 11 | jobs: 12 | R-CMD-check: 13 | runs-on: ${{ matrix.config.os }} 14 | 15 | name: ${{ matrix.config.os }} (${{ matrix.config.r }}) 16 | 17 | strategy: 18 | fail-fast: false 19 | matrix: 20 | config: 21 | - {os: macOS-latest, r: 'release'} 22 | - {os: windows-latest, r: 'release'} 23 | - {os: ubuntu-latest, r: 'devel', http-user-agent: 'release'} 24 | - {os: ubuntu-latest, r: 'release'} 25 | - {os: ubuntu-latest, r: 'oldrel-1'} 26 | 27 | env: 28 | GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }} 29 | R_KEEP_PKG_SOURCE: yes 30 | 31 | steps: 32 | - uses: actions/checkout@v2 33 | 34 | - uses: r-lib/actions/setup-pandoc@v2 35 | 36 | - uses: r-lib/actions/setup-r@v2 37 | with: 38 | r-version: ${{ matrix.config.r }} 39 | http-user-agent: ${{ matrix.config.http-user-agent }} 40 | use-public-rspm: true 41 | 42 | - uses: r-lib/actions/setup-r-dependencies@v2 43 | with: 44 | extra-packages: any::rcmdcheck, any::XML 45 | needs: check 46 | 47 | - uses: r-lib/actions/check-r-package@v2 48 | with: 49 | upload-snapshots: true 50 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | restez/ 2 | # data 3 | *.seq 4 | restez_database 5 | .Rprofile 6 | README.html 7 | rodents/ 8 | # Working files 9 | working* 10 | # History files 11 | .Rhistory 12 | .Rapp.history 13 | # Session Data files 14 | .RData 15 | restez.Rproj 16 | # Example code in package build process 17 | *-Ex.R 18 | # Output files from R CMD build 19 | /*.tar.gz 20 | # Output files from R CMD check 21 | /*.Rcheck/ 22 | # RStudio files 23 | .Rproj.user/ 24 | # produced vignettes 25 | vignettes/*.html 26 | vignettes/*.pdf 27 | # OAuth2 token, see https://github.com/hadley/httr/releases/tag/v0.3 28 | .httr-oauth 29 | # knitr and R markdown default cache directories 30 | /*_cache/ 31 | /cache/ 32 | # Temporary files created by R markdown 33 | *.utf8.md 34 | *.knit.md 35 | .Rproj.user 36 | inst/doc 37 | -------------------------------------------------------------------------------- /CRAN-SUBMISSION: -------------------------------------------------------------------------------- 1 | Version: 2.1.5 2 | Date: 2025-03-06 23:44:12 UTC 3 | SHA: 1c3d9896f0da0000a5c396dc90c0ad43bdf2c628 4 | -------------------------------------------------------------------------------- /DESCRIPTION: -------------------------------------------------------------------------------- 1 | Package: restez 2 | Type: Package 3 | Title: Create and Query a Local Copy of 'GenBank' in R 4 | Version: 2.1.5.9000 5 | Authors@R: c( 6 | person( 7 | given = "Joel H.", 8 | family = "Nitta", 9 | role = c("aut", "cre"), 10 | email = "joelnitta@gmail.com", 11 | comment = c(ORCID = "0000-0003-4719-7472")), 12 | person( 13 | given = "Dom", 14 | family = "Bennett", 15 | role = "aut", 16 | email = "dominic.john.bennett@gmail.com", 17 | comment = c(ORCID = "0000-0003-2722-1359")) 18 | ) 19 | Maintainer: Joel H. Nitta 20 | Description: Download large sections of 21 | 'GenBank' and generate a local 22 | SQL-based database. A user can then query this database using 'restez' 23 | functions or through 'rentrez' 24 | wrappers. 25 | URL: https://github.com/ropensci/restez, 26 | https://docs.ropensci.org/restez/ 27 | BugReports: https://github.com/ropensci/restez/issues 28 | License: MIT + file LICENSE 29 | Encoding: UTF-8 30 | LazyData: true 31 | Depends: 32 | R (>= 3.3.0) 33 | Imports: 34 | utils, 35 | rentrez, 36 | DBI (>= 1.0.0), 37 | curl, 38 | cli, 39 | crayon, 40 | stringi, 41 | duckdb, 42 | fs, 43 | assertthat, 44 | ape 45 | Suggests: 46 | sessioninfo, 47 | testthat, 48 | knitr, 49 | R.utils, 50 | rmarkdown, 51 | mockery 52 | RoxygenNote: 7.3.2 53 | Roxygen: list(markdown = TRUE) 54 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | YEAR: 2022 2 | COPYRIGHT HOLDER: Antonelli Lab and Joel Nitta -------------------------------------------------------------------------------- /NAMESPACE: -------------------------------------------------------------------------------- 1 | # Generated by roxygen2: do not edit by hand 2 | 3 | S3method(print,status) 4 | export(count_db_ids) 5 | export(db_create) 6 | export(db_delete) 7 | export(db_download) 8 | export(demo_db_create) 9 | export(entrez_fetch) 10 | export(gb_definition_get) 11 | export(gb_extract) 12 | export(gb_fasta_get) 13 | export(gb_organism_get) 14 | export(gb_record_get) 15 | export(gb_sequence_get) 16 | export(gb_version_get) 17 | export(is_in_db) 18 | export(list_db_ids) 19 | export(ncbi_acc_get) 20 | export(restez_connect) 21 | export(restez_disconnect) 22 | export(restez_path_get) 23 | export(restez_path_set) 24 | export(restez_path_unset) 25 | export(restez_ready) 26 | export(restez_status) 27 | -------------------------------------------------------------------------------- /R/biomartr-tools.R: -------------------------------------------------------------------------------- 1 | #' @title Helper function to test if a stable internet connection 2 | #' can be established. 3 | #' @description All retrieval functions need a stable 4 | #' internet connection to work properly. This internal function pings 5 | #' the google homepage and throws an error if it cannot be reached. 6 | #' @author Hajk-Georg Drost 7 | #' @family private 8 | # originally connected.to.internet 9 | check_connection <- function() { 10 | check_url <- "https://www.ncbi.nlm.nih.gov/" 11 | if (url_exists(check_url)) { 12 | TRUE 13 | } else { 14 | msg <- paste0("Unable to connect to ", char(check_url), 15 | "Are you connected to the internet?") 16 | stop(msg, call. = FALSE) 17 | } 18 | } 19 | 20 | url_exists <- function(url){ 21 | h <- curl::new_handle(nobody = TRUE) 22 | tryCatch({ 23 | req <- curl::curl_fetch_memory(url, handle = h) 24 | return(req$status_code < 400) 25 | }, error = function(e){FALSE}) 26 | } 27 | -------------------------------------------------------------------------------- /R/connection-tools.R: -------------------------------------------------------------------------------- 1 | #' @name restez_ready 2 | #' @title Is restez ready? 3 | #' @family setup 4 | #' @description Returns TRUE if a restez SQL database is available. 5 | #' Use restez_status() for more information. 6 | #' @return Logical 7 | #' @export 8 | #' @example examples/restez_ready.R 9 | restez_ready <- function() { 10 | fp <- sql_path_get() 11 | inherits(fp, 'character') && length(fp) == 1 && file.exists(fp) 12 | } 13 | 14 | #' @name connected 15 | #' @title Is restez connected? 16 | #' @family private 17 | #' @description Returns TRUE if a restez SQL database has been connected. 18 | #' @return Logical 19 | connected <- function() { 20 | res <- FALSE 21 | connection <- getOption('restez_connection') 22 | if (is.null(connection)) return(FALSE) 23 | if (inherits(x = connection, what = 'duckdb_connection')) { 24 | res <- TRUE 25 | } 26 | res 27 | } 28 | 29 | #' @name has_data 30 | #' @title Does the connected database have data? 31 | #' @family private 32 | #' @description Returns TRUE if a restez SQL database has data. 33 | #' @return Logical 34 | has_data <- function() { 35 | tryCatch(expr = { 36 | suppressWarnings(list_db_ids(n = 1)) 37 | TRUE 38 | }, error = function(e) FALSE) 39 | } 40 | 41 | #' @name restez_connect 42 | #' @title Connect to the restez database 43 | #' @family private 44 | #' @description Sets a connection to the local database. 45 | #' @param read_only Logical; should the connection be made in read-only 46 | #' mode? Read-only mode is required for multiple R processes to access 47 | #' the database simultaneously. Default FALSE. 48 | #' @return NULL 49 | #' @export 50 | restez_connect <- function(read_only = FALSE) { 51 | restez_path_check() 52 | connection <- DBI::dbConnect( 53 | drv = duckdb::duckdb(), 54 | dbdir = sql_path_get(), 55 | read_only = read_only) 56 | options('restez_connection' = connection) 57 | invisible(NULL) 58 | } 59 | 60 | #' @name restez_disconnect 61 | #' @title Disconnect from restez database 62 | #' @family private 63 | #' @description Safely disconnect from the restez connection 64 | #' @return NULL 65 | #' @export 66 | restez_disconnect <- function() { 67 | if (connected()) { 68 | connection <- getOption('restez_connection') 69 | DBI::dbDisconnect(conn = connection, shutdown = TRUE) 70 | } 71 | options('restez_connection' = NULL) 72 | invisible(NULL) 73 | } 74 | 75 | #' @name connection_get 76 | #' @title Retrieve restez connection 77 | #' @family private 78 | #' @description Safely acquire the restez connection. Raises error if no 79 | #' connection set. 80 | #' @return connection 81 | connection_get <- function() { 82 | if (!connected()) { 83 | stop('No restez connection. Did you run `restez_connect`?', call. = FALSE) 84 | } 85 | getOption('restez_connection') 86 | } 87 | -------------------------------------------------------------------------------- /R/db-get-tools.R: -------------------------------------------------------------------------------- 1 | #' @name list_db_ids 2 | #' @title List database IDs 3 | #' @family database 4 | #' @description Return a vector of all IDs in 5 | #' a database. 6 | #' @details Warning: can return very large vectors 7 | #' for large databases. 8 | #' @param db character, database name 9 | #' @param n Maximum number of IDs to return, if NULL returns all 10 | #' @return vector of characters 11 | #' @export 12 | #' @example examples/list_db_ids.R 13 | list_db_ids <- function(db = 'nucleotide', n=100) { 14 | on.exit(restez_disconnect()) 15 | # first close any connection if one exists 16 | restez_disconnect() 17 | restez_connect(read_only = TRUE) 18 | connection <- connection_get() 19 | if (db == 'nucleotide') { 20 | sttmnt <- "SELECT accession FROM nucleotide" 21 | if (!is.null(n)) { 22 | sttmnt <- paste(sttmnt, 'LIMIT', as.integer(n)) 23 | } 24 | res <- DBI::dbGetQuery(conn = connection, statement = sttmnt) 25 | } 26 | if (!is.null(n)) { 27 | msg <- paste0('Number of ids returned was limited to [', n, '].\n', 28 | 'Set `n=NULL` to return all ids.') 29 | warning(msg) 30 | } 31 | restez_disconnect() 32 | res[[1]] 33 | } 34 | 35 | #' @name is_in_db 36 | #' @title Is in db 37 | #' @family database 38 | #' @description Determine whether an id(s) 39 | #' is/are present in a database. 40 | #' @param id character, sequence accession ID(s) 41 | #' @param db character, database name 42 | #' @return named vector of booleans 43 | #' @export 44 | #' @example examples/is_in_db.R 45 | is_in_db <- function(id, db = 'nucleotide') { 46 | accssns <- sub(pattern = '\\.[0-9]+', replacement = '', 47 | x = id) 48 | db_res <- gb_sql_query(nm = 'version', id = id) 49 | res <- accssns %in% db_res[['accession']] 50 | names(res) <- id 51 | res 52 | } 53 | 54 | #' @name count_db_ids 55 | #' @title Return the number of ids 56 | #' @description Return the number of ids in a user's restez database. 57 | #' @details Requires an open connection. If no connection or db 0 is returned. 58 | #' @param db character, database name 59 | #' @return integer 60 | #' @family database 61 | #' @export 62 | #' @example examples/count_db_ids.R 63 | count_db_ids <- function(db = 'nucleotide') { 64 | on.exit(restez_disconnect()) 65 | if (!restez_ready()) { 66 | warning('No database detected. Did you run `db_create()`?') 67 | return(0L) 68 | } 69 | # first close any connection if one exists 70 | restez_disconnect() 71 | restez_connect(read_only = TRUE) 72 | connection <- connection_get() 73 | res <- DBI::dbGetQuery(connection, "SELECT count(*) FROM nucleotide") 74 | restez_disconnect() 75 | as.integer(res[[1]]) 76 | } 77 | -------------------------------------------------------------------------------- /R/entrez-tools.R: -------------------------------------------------------------------------------- 1 | #' @name message_missing 2 | #' @title Produce message of missing IDs 3 | #' @description Sends message to console stating number of missing IDs. 4 | #' @param n Number of missing IDs 5 | #' @return NULL 6 | #' @family private 7 | message_missing <- function(n) { 8 | msg <- paste0('[', n, '] id(s) are unavailable locally, searching online.') 9 | message(msg) 10 | } 11 | 12 | #' Wrapper for rentrez::entrez_fetch(), used for mocking during tests 13 | #' @noRd 14 | entrez_fetch_wrap <- function(...) { 15 | rentrez::entrez_fetch(...) 16 | } 17 | 18 | #' @name entrez_fasta_get 19 | #' @title Get Entrez fasta 20 | #' @description Return fasta format as expected from 21 | #' an Entrez call. If not all IDs are returned, will 22 | #' run rentrez::entrez_fetch. 23 | #' @param id vector, unique ID(s) for record(s) 24 | #' @param ... arguments passed on to rentrez 25 | #' @return character string containing the file created 26 | #' @family private 27 | entrez_fasta_get <- function(id, ...) { 28 | id <- sub(pattern = '\\.[0-9]+', replacement = '', x = id) 29 | fastas <- gb_fasta_get(id = id) 30 | if (length(fastas) > 0) { 31 | res <- paste(fastas, collapse = '') 32 | mssng <- id[!id %in% names(fastas)] 33 | } else { 34 | mssng <- id 35 | res <- '' 36 | } 37 | if (length(mssng) > 0) { 38 | message_missing(length(mssng)) 39 | rentrez_fastas <- entrez_fetch_wrap(id = mssng, ...) 40 | res <- paste0(res, rentrez_fastas) 41 | } 42 | res 43 | } 44 | 45 | #' @name entrez_gb_get 46 | #' @title Get Entrez GenBank record 47 | #' @description Return gb and gbwithparts format as expected from 48 | #' an Entrez call. If not all IDs are returned, will 49 | #' run rentrez::entrez_fetch. 50 | #' @param id vector, unique ID(s) for record(s) 51 | #' @param ... arguments passed on to rentrez 52 | #' @return character string containing the file created 53 | #' @family private 54 | entrez_gb_get <- function(id, ...) { 55 | id <- sub(pattern = '\\.[0-9]+', replacement = '', x = id) 56 | recs <- gb_record_get(id = id) 57 | if (length(recs) > 0) { 58 | res <- paste(recs, collapse = '\n\n') 59 | res <- paste0(res, '\n\n') 60 | mssng <- id[!id %in% names(recs)] 61 | } else { 62 | mssng <- id 63 | res <- '' 64 | } 65 | if (length(mssng) > 0) { 66 | message_missing(length(mssng)) 67 | rentrez_recs <- entrez_fetch_wrap(id = mssng, ...) 68 | res <- paste0(res, rentrez_recs) 69 | } 70 | res 71 | } 72 | -------------------------------------------------------------------------------- /R/print-tools.R: -------------------------------------------------------------------------------- 1 | # Framework copied from pkgdown 2 | #' @name stat 3 | #' @title Print blue 4 | #' @description Print to console blue text to indicate a number/statistic. 5 | #' @param ... Any number of text arguments to print, character 6 | #' @return coloured character encoding, character 7 | #' @family private 8 | stat <- function(...) { 9 | crayon::blue(...) 10 | } 11 | 12 | #' @name char 13 | #' @title Print green 14 | #' @description Print to console green text to indicate a name/filepath/text 15 | #' @param x Text to print, character 16 | #' @return coloured character encoding, character 17 | #' @family private 18 | char <- function(x) { 19 | crayon::green(encodeString(x, quote = "'")) 20 | } 21 | 22 | #' @name cat_line 23 | #' @title Cat lines 24 | #' @description Helper function for printing lines to console. Automatically 25 | #' formats lines by adding newlines. 26 | #' @param ... Text to print, character 27 | #' @return NULL 28 | #' @family private 29 | cat_line <- function(...) { 30 | cat(paste0(..., "\n"), sep = "") 31 | } 32 | -------------------------------------------------------------------------------- /R/rentrez-wrappers.R: -------------------------------------------------------------------------------- 1 | #' @name entrez_fetch 2 | #' @title Entrez fetch 3 | #' @family entrez 4 | #' @description Wrapper for rentrez::entrez_fetch. 5 | #' @details Attempts to first search local database with user-specified 6 | #' parameters, if the record is missing in the database, the function then 7 | #' calls rentrez::entrez_fetch to search GenBank remotely. 8 | #' 9 | #' `rettype='fasta'` and `rettype='gb'` are respectively equivalent to 10 | #' [gb_fasta_get()] and [gb_record_get()]. 11 | #' 12 | #' @note It is advisable to call restez and rentrez functions with '::' notation 13 | #' rather than library() calls to avoid namespace issues. e.g. 14 | #' restez::entrez_fetch(). 15 | #' 16 | #' @section Supported return types and modes: 17 | #' XML retmode is not supported. Rettypes 'seqid', 'ft', 'acc' and 'uilist' 18 | #' are also not supported. 19 | #' @param db character, name of the database 20 | #' @param id vector, unique ID(s) for record(s) 21 | #' @param rettype character, data format 22 | #' @param retmode character, data mode 23 | #' @param ... Arguments to be passed on to rentrez 24 | #' @seealso [rentrez::entrez_fetch()] 25 | #' @return character string containing the file created 26 | #' @example examples/entrez_fetch.R 27 | #' @export 28 | entrez_fetch <- function(db, id=NULL, rettype, retmode="", ...) { 29 | # NCBI table https://tinyurl.com/yb5e7q9b 30 | if (db %in% c('nucleotide', 'nuccore')) { 31 | if (rettype == 'fasta' & retmode != 'xml') { 32 | return(entrez_fasta_get(db = db, id = id, rettype = rettype, 33 | retmode = retmode, ...)) 34 | } 35 | if (rettype == 'gb' & retmode != 'xml') { 36 | return(entrez_gb_get(db = db, id = id, rettype = rettype, 37 | retmode = retmode, ...)) 38 | } 39 | if (rettype == 'gbwithparts' & retmode != 'xml') { 40 | # TODO: I have detected no difference between gb and gbwithparts 41 | return(entrez_gb_get(db = db, id = id, rettype = rettype, 42 | retmode = retmode, ...)) 43 | } 44 | # TODO 45 | # if (rettype == 'ft' & rettype != 'xml') { 46 | # return(entrez_ft_get(id = id)) 47 | # } 48 | } 49 | message('Invalid args for restez, searching online ....') 50 | rentrez::entrez_fetch(db = db, id = id, rettype = rettype, 51 | retmode = retmode, ...) 52 | } 53 | -------------------------------------------------------------------------------- /R/restez.R: -------------------------------------------------------------------------------- 1 | #' restez: Create and Query a Local Copy of GenBank in R 2 | #' 3 | #' The restez package comes with five families of functions: 4 | #' setup, database, get, entrez and internal/private. 5 | #' 6 | #' @section Setup functions: 7 | #' These functions allow a user to set the filepath for where the GenBank files 8 | #' should be stored, create connections and verify these settings. 9 | #' 10 | #' @section Database functions: 11 | #' These functions download specific parts of GenBank and create the local 12 | #' SQL-like database. 13 | #' 14 | #' @section GenBank functions: 15 | #' These functions allow a user to query the local SQL-like database. A 16 | #' user can use an NCBI accession ID to retrieve sequences or whole GenBank 17 | #' records. 18 | #' 19 | #' @section Entrez functions: 20 | #' The entrez functions are wrappers to the `entrez_*` functions in the 21 | #' rentrez package. e.g the restez's entrez_fetch will first try to search the 22 | #' local database, if it fails it will then call rentrez's 23 | #' [rentrez::entrez_fetch()] with the same arguments. 24 | #' 25 | #' @section Private/internal functions: 26 | #' These functions work behind the scenes to make everything work. If you're 27 | #' curious you can read their documentation using the form 28 | #' `?restez:::functionname`. 29 | #' 30 | #' @name restez 31 | #' @keywords internal 32 | "_PACKAGE" 33 | -------------------------------------------------------------------------------- /R/test-tools.R: -------------------------------------------------------------------------------- 1 | #' @name cleanup 2 | #' @title Clean up test data 3 | #' @description Removes all temporary test data created. 4 | #' @return NULL 5 | #' @family private 6 | cleanup <- function() { 7 | restez_path_unset() 8 | if (dir.exists('test_db_fldr')) { 9 | unlink('test_db_fldr', recursive = TRUE) 10 | } 11 | if (file.exists('test_records.txt')) { 12 | file.remove('test_records.txt') 13 | } 14 | } 15 | 16 | #' @name setup 17 | #' @title Set up test common test data 18 | #' @description Creates temporary test folders. 19 | #' @return NULL 20 | #' @family private 21 | setup <- function() { 22 | dir.create('test_db_fldr') 23 | restez_path_set(filepath = 'test_db_fldr') 24 | } 25 | 26 | #' @name testdatadir_get 27 | #' @title Get test data directory 28 | #' @description Get the folder containing test data. 29 | #' @return NULL 30 | #' @family private 31 | testdatadir_get <- function() { 32 | wd <- getwd() 33 | if (grepl('testthat', wd)) { 34 | data_d <- file.path('data') 35 | } else { 36 | # for running test at package level 37 | data_d <- file.path('tests', 'testthat', 'data') 38 | } 39 | data_d 40 | } 41 | -------------------------------------------------------------------------------- /R/wrappers.R: -------------------------------------------------------------------------------- 1 | #' @name restez_rl 2 | #' @title Restez readline 3 | #' @description Wrapper for base readline. 4 | #' @param prompt character, display text 5 | #' @return character 6 | #' @family private 7 | restez_rl <- function(prompt) { 8 | base::readline(prompt) 9 | } 10 | -------------------------------------------------------------------------------- /cran-comments.md: -------------------------------------------------------------------------------- 1 | # restez v2.1.5 2 | 3 | This is a bug fix to address changed behavior of a dependency 4 | (https://github.com/ropensci/restez/issues/64) that is currently causing 5 | build failures. It also removes use of with_mock() 6 | (https://github.com/ropensci/restez/issues/63) 7 | 8 | ## Test environments 9 | 10 | * local OS X install, R 4.4.1 11 | * Windows Server 2022 x64 (winbuilder) 12 | 13 | ## R CMD check results 14 | 15 | There were no ERRORs, WARNINGs, or NOTEs. 16 | 17 | ## Reverse dependencies 18 | 19 | There are no reverse dependencies currently on CRAN. 20 | -------------------------------------------------------------------------------- /data/record.rda: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/data/record.rda -------------------------------------------------------------------------------- /examples/count_db_ids.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(filepath = tempdir()) 3 | demo_db_create(n = 5) 4 | (count_db_ids()) 5 | 6 | # delete demo after example 7 | db_delete(everything = TRUE) 8 | -------------------------------------------------------------------------------- /examples/db_delete.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | fp <- tempdir() 3 | restez_path_set(filepath = fp) 4 | demo_db_create(n = 10) 5 | db_delete(everything = FALSE) 6 | # Will not run: gb_sequence_get(id = 'demo_1') 7 | # only the SQL database is deleted 8 | db_delete(everything = TRUE) 9 | # Now returns NULL 10 | (restez_path_get()) 11 | -------------------------------------------------------------------------------- /examples/demo_db_create.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | # set the restez path to a temporary dir 3 | restez_path_set(filepath = tempdir()) 4 | # create demo database 5 | demo_db_create(n = 5) 6 | # in the demo, IDs are 'demo_1', 'demo_2' ... 7 | (gb_sequence_get(id = 'demo_1')) 8 | 9 | # Delete a demo database after an example 10 | db_delete(everything = TRUE) 11 | -------------------------------------------------------------------------------- /examples/entrez_fetch.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(tempdir()) 3 | demo_db_create(n = 5) 4 | # return fasta record 5 | fasta_res <- entrez_fetch(db = 'nucleotide', 6 | id = c('demo_1', 'demo_2'), 7 | rettype = 'fasta') 8 | cat(fasta_res) 9 | # return whole GB record in text format 10 | gb_res <- entrez_fetch(db = 'nucleotide', 11 | id = c('demo_1', 'demo_2'), 12 | rettype = 'gb') 13 | cat(gb_res) 14 | # NOT RUN 15 | # whereas these request would go through rentrez 16 | # fasta_res <- entrez_fetch(db = 'nucleotide', 17 | # id = c('S71333', 'S71334'), 18 | # rettype = 'fasta') 19 | # gb_res <- entrez_fetch(db = 'nucleotide', 20 | # id = c('S71333', 'S71334'), 21 | # rettype = 'gb') 22 | 23 | # delete demo after example 24 | db_delete(everything = TRUE) 25 | -------------------------------------------------------------------------------- /examples/gb_definition_get.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(filepath = tempdir()) 3 | demo_db_create(n = 5) 4 | (def <- gb_definition_get(id = 'demo_1')) 5 | (defs <- gb_definition_get(id = c('demo_1', 'demo_2'))) 6 | 7 | 8 | # delete demo after example 9 | db_delete(everything = TRUE) 10 | -------------------------------------------------------------------------------- /examples/gb_extract.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | data('record') 3 | (gb_extract(record = record, what = 'locus')) 4 | -------------------------------------------------------------------------------- /examples/gb_fasta_get.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(filepath = tempdir()) 3 | demo_db_create(n = 5) 4 | (fasta <- gb_fasta_get(id = 'demo_1')) 5 | (fastas <- gb_fasta_get(id = c('demo_1', 'demo_2'))) 6 | 7 | 8 | # delete demo after example 9 | db_delete(everything = TRUE) 10 | -------------------------------------------------------------------------------- /examples/gb_organism_get.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(filepath = tempdir()) 3 | demo_db_create(n = 5) 4 | (org <- gb_organism_get(id = 'demo_1')) 5 | (orgs <- gb_organism_get(id = c('demo_1', 'demo_2'))) 6 | 7 | 8 | # delete demo after example 9 | db_delete(everything = TRUE) 10 | -------------------------------------------------------------------------------- /examples/gb_record_get.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(filepath = tempdir()) 3 | demo_db_create(n = 5) 4 | (rec <- gb_record_get(id = 'demo_1')) 5 | (recs <- gb_record_get(id = c('demo_1', 'demo_2'))) 6 | 7 | 8 | # delete demo after example 9 | db_delete(everything = TRUE) 10 | -------------------------------------------------------------------------------- /examples/gb_sequence_get.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(filepath = tempdir()) 3 | demo_db_create(n = 5) 4 | (seq <- gb_sequence_get(id = 'demo_1')) 5 | (seqs <- gb_sequence_get(id = c('demo_1', 'demo_2'))) 6 | (fasta_dnabin <- gb_sequence_get(id = 'demo_1', dnabin = TRUE)) 7 | 8 | # delete demo after example 9 | db_delete(everything = TRUE) 10 | 11 | -------------------------------------------------------------------------------- /examples/gb_version_get.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(filepath = tempdir()) 3 | demo_db_create(n = 5) 4 | (ver <- gb_version_get(id = 'demo_1')) 5 | (vers <- gb_version_get(id = c('demo_1', 'demo_2'))) 6 | 7 | 8 | # delete demo after example 9 | db_delete(everything = TRUE) 10 | 11 | 12 | -------------------------------------------------------------------------------- /examples/is_in_db.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | # set the restez path to a temporary dir 3 | restez_path_set(filepath = tempdir()) 4 | # create demo database 5 | demo_db_create(n = 5) 6 | # in the demo, IDs are 'demo_1', 'demo_2' ... 7 | ids <- c('thisisnotanid', 'demo_1', 'demo_2') 8 | (is_in_db(id = ids)) 9 | 10 | 11 | # delete demo after example 12 | db_delete(everything = TRUE) 13 | -------------------------------------------------------------------------------- /examples/list_db_ids.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | restez_path_set(filepath = tempdir()) 3 | demo_db_create(n = 5) 4 | # Warning: not recommended for real databases 5 | # with potentially millions of IDs 6 | all_ids <- list_db_ids() 7 | 8 | 9 | # What shall we do with these IDs? 10 | # ... how about make a mock fasta file 11 | seqs <- gb_sequence_get(id = all_ids) 12 | defs <- gb_definition_get(id = all_ids) 13 | # paste together 14 | fasta_seqs <- paste0('>', defs, '\n', seqs) 15 | fasta_file <- paste0(fasta_seqs, collapse = '\n') 16 | cat(fasta_file) 17 | 18 | 19 | # delete after example 20 | db_delete(everything = TRUE) 21 | -------------------------------------------------------------------------------- /examples/restez_path_get.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | # set a restez path with a tempdir 3 | restez_path_set(filepath = tempdir()) 4 | # check what the set path is 5 | (restez_path_get()) 6 | -------------------------------------------------------------------------------- /examples/restez_ready.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | fp <- tempdir() 3 | restez_path_set(filepath = fp) 4 | demo_db_create(n = 5) 5 | (restez_ready()) 6 | db_delete(everything = TRUE) 7 | (restez_ready()) 8 | -------------------------------------------------------------------------------- /examples/restez_status.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | fp <- tempdir() 3 | restez_path_set(filepath = fp) 4 | demo_db_create(n = 5) 5 | restez_status() 6 | db_delete(everything = TRUE) 7 | # Errors: 8 | # restez_status() 9 | -------------------------------------------------------------------------------- /inst/CITATION: -------------------------------------------------------------------------------- 1 | bibentry( 2 | bibtype = "Article", 3 | author = c( 4 | person( 5 | given = "Dominic J.", 6 | family = "Bennett", 7 | role = "aut" 8 | ), 9 | person( 10 | given = "Hannes", 11 | family = "Hettling", 12 | role = "aut" 13 | ), 14 | person( 15 | given = "Daniele", 16 | family = "Silvestro", 17 | role = "aut" 18 | ), 19 | person( 20 | given = "Rutger", 21 | family = "Vos", 22 | role = "aut" 23 | ), 24 | person( 25 | given = "Alexandre", 26 | family = "Antonelli", 27 | role = "aut" 28 | ) 29 | ), 30 | title = "restez: Create and Query a Local Copy of GenBank in R", 31 | journal = "The Journal of Open Source Software", 32 | year = 2018, 33 | number = 31, 34 | pages = 1102, 35 | volume = 3, 36 | doi = "10.21105/joss.01102" 37 | ) 38 | -------------------------------------------------------------------------------- /logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/logo.png -------------------------------------------------------------------------------- /man/cat_line.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/print-tools.R 3 | \name{cat_line} 4 | \alias{cat_line} 5 | \title{Cat lines} 6 | \usage{ 7 | cat_line(...) 8 | } 9 | \arguments{ 10 | \item{...}{Text to print, character} 11 | } 12 | \description{ 13 | Helper function for printing lines to console. Automatically 14 | formats lines by adding newlines. 15 | } 16 | \seealso{ 17 | Other private: 18 | \code{\link{add_rcrd_log}()}, 19 | \code{\link{char}()}, 20 | \code{\link{check_connection}()}, 21 | \code{\link{cleanup}()}, 22 | \code{\link{connected}()}, 23 | \code{\link{connection_get}()}, 24 | \code{\link{db_download_intern}()}, 25 | \code{\link{db_sqlngths_get}()}, 26 | \code{\link{db_sqlngths_log}()}, 27 | \code{\link{dir_size}()}, 28 | \code{\link{dwnld_path_get}()}, 29 | \code{\link{dwnld_rcrd_log}()}, 30 | \code{\link{entrez_fasta_get}()}, 31 | \code{\link{entrez_gb_get}()}, 32 | \code{\link{extract_accession}()}, 33 | \code{\link{extract_by_patterns}()}, 34 | \code{\link{extract_clean_sequence}()}, 35 | \code{\link{extract_definition}()}, 36 | \code{\link{extract_features}()}, 37 | \code{\link{extract_inforecpart}()}, 38 | \code{\link{extract_keywords}()}, 39 | \code{\link{extract_locus}()}, 40 | \code{\link{extract_organism}()}, 41 | \code{\link{extract_seqrecpart}()}, 42 | \code{\link{extract_sequence}()}, 43 | \code{\link{extract_version}()}, 44 | \code{\link{file_download}()}, 45 | \code{\link{filename_log}()}, 46 | \code{\link{flatfile_read}()}, 47 | \code{\link{gb_build}()}, 48 | \code{\link{gb_df_create}()}, 49 | \code{\link{gb_df_generate}()}, 50 | \code{\link{gb_sql_add}()}, 51 | \code{\link{gb_sql_query}()}, 52 | \code{\link{gbrelease_check}()}, 53 | \code{\link{gbrelease_get}()}, 54 | \code{\link{gbrelease_log}()}, 55 | \code{\link{has_data}()}, 56 | \code{\link{identify_downloadable_files}()}, 57 | \code{\link{last_add_get}()}, 58 | \code{\link{last_dwnld_get}()}, 59 | \code{\link{last_entry_get}()}, 60 | \code{\link{latest_genbank_release}()}, 61 | \code{\link{latest_genbank_release_notes}()}, 62 | \code{\link{message_missing}()}, 63 | \code{\link{mock_def}()}, 64 | \code{\link{mock_gb_df_generate}()}, 65 | \code{\link{mock_org}()}, 66 | \code{\link{mock_rec}()}, 67 | \code{\link{mock_seq}()}, 68 | \code{\link{predict_datasizes}()}, 69 | \code{\link{readme_log}()}, 70 | \code{\link{restez_connect}()}, 71 | \code{\link{restez_disconnect}()}, 72 | \code{\link{restez_path_check}()}, 73 | \code{\link{restez_rl}()}, 74 | \code{\link{search_gz}()}, 75 | \code{\link{seshinfo_log}()}, 76 | \code{\link{setup}()}, 77 | \code{\link{slctn_get}()}, 78 | \code{\link{slctn_log}()}, 79 | \code{\link{sql_path_get}()}, 80 | \code{\link{stat}()}, 81 | \code{\link{status_class}()}, 82 | \code{\link{testdatadir_get}()} 83 | } 84 | \concept{private} 85 | -------------------------------------------------------------------------------- /man/char.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/print-tools.R 3 | \name{char} 4 | \alias{char} 5 | \title{Print green} 6 | \usage{ 7 | char(x) 8 | } 9 | \arguments{ 10 | \item{x}{Text to print, character} 11 | } 12 | \value{ 13 | coloured character encoding, character 14 | } 15 | \description{ 16 | Print to console green text to indicate a name/filepath/text 17 | } 18 | \seealso{ 19 | Other private: 20 | \code{\link{add_rcrd_log}()}, 21 | \code{\link{cat_line}()}, 22 | \code{\link{check_connection}()}, 23 | \code{\link{cleanup}()}, 24 | \code{\link{connected}()}, 25 | \code{\link{connection_get}()}, 26 | \code{\link{db_download_intern}()}, 27 | \code{\link{db_sqlngths_get}()}, 28 | \code{\link{db_sqlngths_log}()}, 29 | \code{\link{dir_size}()}, 30 | \code{\link{dwnld_path_get}()}, 31 | \code{\link{dwnld_rcrd_log}()}, 32 | \code{\link{entrez_fasta_get}()}, 33 | \code{\link{entrez_gb_get}()}, 34 | \code{\link{extract_accession}()}, 35 | \code{\link{extract_by_patterns}()}, 36 | \code{\link{extract_clean_sequence}()}, 37 | \code{\link{extract_definition}()}, 38 | \code{\link{extract_features}()}, 39 | \code{\link{extract_inforecpart}()}, 40 | \code{\link{extract_keywords}()}, 41 | \code{\link{extract_locus}()}, 42 | \code{\link{extract_organism}()}, 43 | \code{\link{extract_seqrecpart}()}, 44 | \code{\link{extract_sequence}()}, 45 | \code{\link{extract_version}()}, 46 | \code{\link{file_download}()}, 47 | \code{\link{filename_log}()}, 48 | \code{\link{flatfile_read}()}, 49 | \code{\link{gb_build}()}, 50 | \code{\link{gb_df_create}()}, 51 | \code{\link{gb_df_generate}()}, 52 | \code{\link{gb_sql_add}()}, 53 | \code{\link{gb_sql_query}()}, 54 | \code{\link{gbrelease_check}()}, 55 | \code{\link{gbrelease_get}()}, 56 | \code{\link{gbrelease_log}()}, 57 | \code{\link{has_data}()}, 58 | \code{\link{identify_downloadable_files}()}, 59 | \code{\link{last_add_get}()}, 60 | \code{\link{last_dwnld_get}()}, 61 | \code{\link{last_entry_get}()}, 62 | \code{\link{latest_genbank_release}()}, 63 | \code{\link{latest_genbank_release_notes}()}, 64 | \code{\link{message_missing}()}, 65 | \code{\link{mock_def}()}, 66 | \code{\link{mock_gb_df_generate}()}, 67 | \code{\link{mock_org}()}, 68 | \code{\link{mock_rec}()}, 69 | \code{\link{mock_seq}()}, 70 | \code{\link{predict_datasizes}()}, 71 | \code{\link{readme_log}()}, 72 | \code{\link{restez_connect}()}, 73 | \code{\link{restez_disconnect}()}, 74 | \code{\link{restez_path_check}()}, 75 | \code{\link{restez_rl}()}, 76 | \code{\link{search_gz}()}, 77 | \code{\link{seshinfo_log}()}, 78 | \code{\link{setup}()}, 79 | \code{\link{slctn_get}()}, 80 | \code{\link{slctn_log}()}, 81 | \code{\link{sql_path_get}()}, 82 | \code{\link{stat}()}, 83 | \code{\link{status_class}()}, 84 | \code{\link{testdatadir_get}()} 85 | } 86 | \concept{private} 87 | -------------------------------------------------------------------------------- /man/cleanup.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/test-tools.R 3 | \name{cleanup} 4 | \alias{cleanup} 5 | \title{Clean up test data} 6 | \usage{ 7 | cleanup() 8 | } 9 | \description{ 10 | Removes all temporary test data created. 11 | } 12 | \seealso{ 13 | Other private: 14 | \code{\link{add_rcrd_log}()}, 15 | \code{\link{cat_line}()}, 16 | \code{\link{char}()}, 17 | \code{\link{check_connection}()}, 18 | \code{\link{connected}()}, 19 | \code{\link{connection_get}()}, 20 | \code{\link{db_download_intern}()}, 21 | \code{\link{db_sqlngths_get}()}, 22 | \code{\link{db_sqlngths_log}()}, 23 | \code{\link{dir_size}()}, 24 | \code{\link{dwnld_path_get}()}, 25 | \code{\link{dwnld_rcrd_log}()}, 26 | \code{\link{entrez_fasta_get}()}, 27 | \code{\link{entrez_gb_get}()}, 28 | \code{\link{extract_accession}()}, 29 | \code{\link{extract_by_patterns}()}, 30 | \code{\link{extract_clean_sequence}()}, 31 | \code{\link{extract_definition}()}, 32 | \code{\link{extract_features}()}, 33 | \code{\link{extract_inforecpart}()}, 34 | \code{\link{extract_keywords}()}, 35 | \code{\link{extract_locus}()}, 36 | \code{\link{extract_organism}()}, 37 | \code{\link{extract_seqrecpart}()}, 38 | \code{\link{extract_sequence}()}, 39 | \code{\link{extract_version}()}, 40 | \code{\link{file_download}()}, 41 | \code{\link{filename_log}()}, 42 | \code{\link{flatfile_read}()}, 43 | \code{\link{gb_build}()}, 44 | \code{\link{gb_df_create}()}, 45 | \code{\link{gb_df_generate}()}, 46 | \code{\link{gb_sql_add}()}, 47 | \code{\link{gb_sql_query}()}, 48 | \code{\link{gbrelease_check}()}, 49 | \code{\link{gbrelease_get}()}, 50 | \code{\link{gbrelease_log}()}, 51 | \code{\link{has_data}()}, 52 | \code{\link{identify_downloadable_files}()}, 53 | \code{\link{last_add_get}()}, 54 | \code{\link{last_dwnld_get}()}, 55 | \code{\link{last_entry_get}()}, 56 | \code{\link{latest_genbank_release}()}, 57 | \code{\link{latest_genbank_release_notes}()}, 58 | \code{\link{message_missing}()}, 59 | \code{\link{mock_def}()}, 60 | \code{\link{mock_gb_df_generate}()}, 61 | \code{\link{mock_org}()}, 62 | \code{\link{mock_rec}()}, 63 | \code{\link{mock_seq}()}, 64 | \code{\link{predict_datasizes}()}, 65 | \code{\link{readme_log}()}, 66 | \code{\link{restez_connect}()}, 67 | \code{\link{restez_disconnect}()}, 68 | \code{\link{restez_path_check}()}, 69 | \code{\link{restez_rl}()}, 70 | \code{\link{search_gz}()}, 71 | \code{\link{seshinfo_log}()}, 72 | \code{\link{setup}()}, 73 | \code{\link{slctn_get}()}, 74 | \code{\link{slctn_log}()}, 75 | \code{\link{sql_path_get}()}, 76 | \code{\link{stat}()}, 77 | \code{\link{status_class}()}, 78 | \code{\link{testdatadir_get}()} 79 | } 80 | \concept{private} 81 | -------------------------------------------------------------------------------- /man/connected.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/connection-tools.R 3 | \name{connected} 4 | \alias{connected} 5 | \title{Is restez connected?} 6 | \usage{ 7 | connected() 8 | } 9 | \value{ 10 | Logical 11 | } 12 | \description{ 13 | Returns TRUE if a restez SQL database has been connected. 14 | } 15 | \seealso{ 16 | Other private: 17 | \code{\link{add_rcrd_log}()}, 18 | \code{\link{cat_line}()}, 19 | \code{\link{char}()}, 20 | \code{\link{check_connection}()}, 21 | \code{\link{cleanup}()}, 22 | \code{\link{connection_get}()}, 23 | \code{\link{db_download_intern}()}, 24 | \code{\link{db_sqlngths_get}()}, 25 | \code{\link{db_sqlngths_log}()}, 26 | \code{\link{dir_size}()}, 27 | \code{\link{dwnld_path_get}()}, 28 | \code{\link{dwnld_rcrd_log}()}, 29 | \code{\link{entrez_fasta_get}()}, 30 | \code{\link{entrez_gb_get}()}, 31 | \code{\link{extract_accession}()}, 32 | \code{\link{extract_by_patterns}()}, 33 | \code{\link{extract_clean_sequence}()}, 34 | \code{\link{extract_definition}()}, 35 | \code{\link{extract_features}()}, 36 | \code{\link{extract_inforecpart}()}, 37 | \code{\link{extract_keywords}()}, 38 | \code{\link{extract_locus}()}, 39 | \code{\link{extract_organism}()}, 40 | \code{\link{extract_seqrecpart}()}, 41 | \code{\link{extract_sequence}()}, 42 | \code{\link{extract_version}()}, 43 | \code{\link{file_download}()}, 44 | \code{\link{filename_log}()}, 45 | \code{\link{flatfile_read}()}, 46 | \code{\link{gb_build}()}, 47 | \code{\link{gb_df_create}()}, 48 | \code{\link{gb_df_generate}()}, 49 | \code{\link{gb_sql_add}()}, 50 | \code{\link{gb_sql_query}()}, 51 | \code{\link{gbrelease_check}()}, 52 | \code{\link{gbrelease_get}()}, 53 | \code{\link{gbrelease_log}()}, 54 | \code{\link{has_data}()}, 55 | \code{\link{identify_downloadable_files}()}, 56 | \code{\link{last_add_get}()}, 57 | \code{\link{last_dwnld_get}()}, 58 | \code{\link{last_entry_get}()}, 59 | \code{\link{latest_genbank_release}()}, 60 | \code{\link{latest_genbank_release_notes}()}, 61 | \code{\link{message_missing}()}, 62 | \code{\link{mock_def}()}, 63 | \code{\link{mock_gb_df_generate}()}, 64 | \code{\link{mock_org}()}, 65 | \code{\link{mock_rec}()}, 66 | \code{\link{mock_seq}()}, 67 | \code{\link{predict_datasizes}()}, 68 | \code{\link{readme_log}()}, 69 | \code{\link{restez_connect}()}, 70 | \code{\link{restez_disconnect}()}, 71 | \code{\link{restez_path_check}()}, 72 | \code{\link{restez_rl}()}, 73 | \code{\link{search_gz}()}, 74 | \code{\link{seshinfo_log}()}, 75 | \code{\link{setup}()}, 76 | \code{\link{slctn_get}()}, 77 | \code{\link{slctn_log}()}, 78 | \code{\link{sql_path_get}()}, 79 | \code{\link{stat}()}, 80 | \code{\link{status_class}()}, 81 | \code{\link{testdatadir_get}()} 82 | } 83 | \concept{private} 84 | -------------------------------------------------------------------------------- /man/connection_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/connection-tools.R 3 | \name{connection_get} 4 | \alias{connection_get} 5 | \title{Retrieve restez connection} 6 | \usage{ 7 | connection_get() 8 | } 9 | \value{ 10 | connection 11 | } 12 | \description{ 13 | Safely acquire the restez connection. Raises error if no 14 | connection set. 15 | } 16 | \seealso{ 17 | Other private: 18 | \code{\link{add_rcrd_log}()}, 19 | \code{\link{cat_line}()}, 20 | \code{\link{char}()}, 21 | \code{\link{check_connection}()}, 22 | \code{\link{cleanup}()}, 23 | \code{\link{connected}()}, 24 | \code{\link{db_download_intern}()}, 25 | \code{\link{db_sqlngths_get}()}, 26 | \code{\link{db_sqlngths_log}()}, 27 | \code{\link{dir_size}()}, 28 | \code{\link{dwnld_path_get}()}, 29 | \code{\link{dwnld_rcrd_log}()}, 30 | \code{\link{entrez_fasta_get}()}, 31 | \code{\link{entrez_gb_get}()}, 32 | \code{\link{extract_accession}()}, 33 | \code{\link{extract_by_patterns}()}, 34 | \code{\link{extract_clean_sequence}()}, 35 | \code{\link{extract_definition}()}, 36 | \code{\link{extract_features}()}, 37 | \code{\link{extract_inforecpart}()}, 38 | \code{\link{extract_keywords}()}, 39 | \code{\link{extract_locus}()}, 40 | \code{\link{extract_organism}()}, 41 | \code{\link{extract_seqrecpart}()}, 42 | \code{\link{extract_sequence}()}, 43 | \code{\link{extract_version}()}, 44 | \code{\link{file_download}()}, 45 | \code{\link{filename_log}()}, 46 | \code{\link{flatfile_read}()}, 47 | \code{\link{gb_build}()}, 48 | \code{\link{gb_df_create}()}, 49 | \code{\link{gb_df_generate}()}, 50 | \code{\link{gb_sql_add}()}, 51 | \code{\link{gb_sql_query}()}, 52 | \code{\link{gbrelease_check}()}, 53 | \code{\link{gbrelease_get}()}, 54 | \code{\link{gbrelease_log}()}, 55 | \code{\link{has_data}()}, 56 | \code{\link{identify_downloadable_files}()}, 57 | \code{\link{last_add_get}()}, 58 | \code{\link{last_dwnld_get}()}, 59 | \code{\link{last_entry_get}()}, 60 | \code{\link{latest_genbank_release}()}, 61 | \code{\link{latest_genbank_release_notes}()}, 62 | \code{\link{message_missing}()}, 63 | \code{\link{mock_def}()}, 64 | \code{\link{mock_gb_df_generate}()}, 65 | \code{\link{mock_org}()}, 66 | \code{\link{mock_rec}()}, 67 | \code{\link{mock_seq}()}, 68 | \code{\link{predict_datasizes}()}, 69 | \code{\link{readme_log}()}, 70 | \code{\link{restez_connect}()}, 71 | \code{\link{restez_disconnect}()}, 72 | \code{\link{restez_path_check}()}, 73 | \code{\link{restez_rl}()}, 74 | \code{\link{search_gz}()}, 75 | \code{\link{seshinfo_log}()}, 76 | \code{\link{setup}()}, 77 | \code{\link{slctn_get}()}, 78 | \code{\link{slctn_log}()}, 79 | \code{\link{sql_path_get}()}, 80 | \code{\link{stat}()}, 81 | \code{\link{status_class}()}, 82 | \code{\link{testdatadir_get}()} 83 | } 84 | \concept{private} 85 | -------------------------------------------------------------------------------- /man/count_db_ids.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/db-get-tools.R 3 | \name{count_db_ids} 4 | \alias{count_db_ids} 5 | \title{Return the number of ids} 6 | \usage{ 7 | count_db_ids(db = "nucleotide") 8 | } 9 | \arguments{ 10 | \item{db}{character, database name} 11 | } 12 | \value{ 13 | integer 14 | } 15 | \description{ 16 | Return the number of ids in a user's restez database. 17 | } 18 | \details{ 19 | Requires an open connection. If no connection or db 0 is returned. 20 | } 21 | \examples{ 22 | library(restez) 23 | restez_path_set(filepath = tempdir()) 24 | demo_db_create(n = 5) 25 | (count_db_ids()) 26 | 27 | # delete demo after example 28 | db_delete(everything = TRUE) 29 | } 30 | \seealso{ 31 | Other database: 32 | \code{\link{db_create}()}, 33 | \code{\link{db_delete}()}, 34 | \code{\link{db_download}()}, 35 | \code{\link{demo_db_create}()}, 36 | \code{\link{is_in_db}()}, 37 | \code{\link{list_db_ids}()} 38 | } 39 | \concept{database} 40 | -------------------------------------------------------------------------------- /man/db_delete.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/db-setup-tools.R 3 | \name{db_delete} 4 | \alias{db_delete} 5 | \title{Delete database} 6 | \usage{ 7 | db_delete(everything = FALSE) 8 | } 9 | \arguments{ 10 | \item{everything}{T/F, delete the whole restez folder as well?} 11 | } 12 | \description{ 13 | Delete the local SQL database and/or restez folder. 14 | } 15 | \details{ 16 | Any connected database will be automatically disconnected. 17 | } 18 | \examples{ 19 | library(restez) 20 | fp <- tempdir() 21 | restez_path_set(filepath = fp) 22 | demo_db_create(n = 10) 23 | db_delete(everything = FALSE) 24 | # Will not run: gb_sequence_get(id = 'demo_1') 25 | # only the SQL database is deleted 26 | db_delete(everything = TRUE) 27 | # Now returns NULL 28 | (restez_path_get()) 29 | } 30 | \seealso{ 31 | Other database: 32 | \code{\link{count_db_ids}()}, 33 | \code{\link{db_create}()}, 34 | \code{\link{db_download}()}, 35 | \code{\link{demo_db_create}()}, 36 | \code{\link{is_in_db}()}, 37 | \code{\link{list_db_ids}()} 38 | } 39 | \concept{database} 40 | -------------------------------------------------------------------------------- /man/db_download.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/db-setup-tools.R 3 | \name{db_download} 4 | \alias{db_download} 5 | \title{Download database} 6 | \usage{ 7 | db_download( 8 | db = "nucleotide", 9 | overwrite = FALSE, 10 | preselection = NULL, 11 | max_tries = 1 12 | ) 13 | } 14 | \arguments{ 15 | \item{db}{Database type, only 'nucleotide' currently available.} 16 | 17 | \item{overwrite}{T/F, overwrite pre-existing downloaded files?} 18 | 19 | \item{preselection}{Character vector of length 1; GenBank domains to 20 | download. If not specified (default), a menu will be provided for 21 | selection. 22 | To specify, provide either a single number or a single character string 23 | of numbers separated by spaces, e.g. "19 20" for 'Phage' (19) and 24 | 'Unannotated' (20).} 25 | 26 | \item{max_tries}{Numeric vector of length 1; maximum number of times to 27 | attempt downloading database (default 1).} 28 | } 29 | \value{ 30 | T/F, if all files download correctly, TRUE else FALSE. 31 | } 32 | \description{ 33 | Download .seq.tar files from the latest GenBank release. 34 | } 35 | \details{ 36 | In default mode, the user interactively selects the parts (i.e., "domains") 37 | of GenBank to download (e.g. primates, plants, bacteria ...). Alternatively, 38 | the selected domains can be provided as a character string to \code{preselection}. 39 | 40 | The \code{max_tries} argument is useful for large databases that may otherwise 41 | fail due to periodic lapses in internet connectivity. This value can be set 42 | to \code{Inf} to continuously try until the database download succeeds (not 43 | recommended if you do not have an internet connection!). 44 | } 45 | \examples{ 46 | \dontrun{ 47 | library(restez) 48 | restez_path_set(filepath = 'path/for/downloads') 49 | db_download() 50 | } 51 | } 52 | \seealso{ 53 | \code{\link[=ncbi_acc_get]{ncbi_acc_get()}} 54 | 55 | Other database: 56 | \code{\link{count_db_ids}()}, 57 | \code{\link{db_create}()}, 58 | \code{\link{db_delete}()}, 59 | \code{\link{demo_db_create}()}, 60 | \code{\link{is_in_db}()}, 61 | \code{\link{list_db_ids}()} 62 | } 63 | \concept{database} 64 | -------------------------------------------------------------------------------- /man/db_sqlngths_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{db_sqlngths_get} 4 | \alias{db_sqlngths_get} 5 | \title{Return the minimum and maximum sequence lengths in db} 6 | \usage{ 7 | db_sqlngths_get() 8 | } 9 | \value{ 10 | vector of integers 11 | } 12 | \description{ 13 | Returns the maximum and minimum sequence lengths as set by the 14 | user upon db creation. 15 | } 16 | \details{ 17 | If no file found, returns empty character vector. 18 | } 19 | \seealso{ 20 | Other private: 21 | \code{\link{add_rcrd_log}()}, 22 | \code{\link{cat_line}()}, 23 | \code{\link{char}()}, 24 | \code{\link{check_connection}()}, 25 | \code{\link{cleanup}()}, 26 | \code{\link{connected}()}, 27 | \code{\link{connection_get}()}, 28 | \code{\link{db_download_intern}()}, 29 | \code{\link{db_sqlngths_log}()}, 30 | \code{\link{dir_size}()}, 31 | \code{\link{dwnld_path_get}()}, 32 | \code{\link{dwnld_rcrd_log}()}, 33 | \code{\link{entrez_fasta_get}()}, 34 | \code{\link{entrez_gb_get}()}, 35 | \code{\link{extract_accession}()}, 36 | \code{\link{extract_by_patterns}()}, 37 | \code{\link{extract_clean_sequence}()}, 38 | \code{\link{extract_definition}()}, 39 | \code{\link{extract_features}()}, 40 | \code{\link{extract_inforecpart}()}, 41 | \code{\link{extract_keywords}()}, 42 | \code{\link{extract_locus}()}, 43 | \code{\link{extract_organism}()}, 44 | \code{\link{extract_seqrecpart}()}, 45 | \code{\link{extract_sequence}()}, 46 | \code{\link{extract_version}()}, 47 | \code{\link{file_download}()}, 48 | \code{\link{filename_log}()}, 49 | \code{\link{flatfile_read}()}, 50 | \code{\link{gb_build}()}, 51 | \code{\link{gb_df_create}()}, 52 | \code{\link{gb_df_generate}()}, 53 | \code{\link{gb_sql_add}()}, 54 | \code{\link{gb_sql_query}()}, 55 | \code{\link{gbrelease_check}()}, 56 | \code{\link{gbrelease_get}()}, 57 | \code{\link{gbrelease_log}()}, 58 | \code{\link{has_data}()}, 59 | \code{\link{identify_downloadable_files}()}, 60 | \code{\link{last_add_get}()}, 61 | \code{\link{last_dwnld_get}()}, 62 | \code{\link{last_entry_get}()}, 63 | \code{\link{latest_genbank_release}()}, 64 | \code{\link{latest_genbank_release_notes}()}, 65 | \code{\link{message_missing}()}, 66 | \code{\link{mock_def}()}, 67 | \code{\link{mock_gb_df_generate}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_get}()}, 81 | \code{\link{slctn_log}()}, 82 | \code{\link{sql_path_get}()}, 83 | \code{\link{stat}()}, 84 | \code{\link{status_class}()}, 85 | \code{\link{testdatadir_get}()} 86 | } 87 | \concept{private} 88 | -------------------------------------------------------------------------------- /man/db_sqlngths_log.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{db_sqlngths_log} 4 | \alias{db_sqlngths_log} 5 | \title{Log the min and max sequence lengths} 6 | \usage{ 7 | db_sqlngths_log(min_lngth, max_lngth) 8 | } 9 | \arguments{ 10 | \item{min_lngth}{Minimum length} 11 | 12 | \item{max_lngth}{Maximum length} 13 | } 14 | \description{ 15 | Log the min and maximum sequence length used in the created db. 16 | } 17 | \seealso{ 18 | Other private: 19 | \code{\link{add_rcrd_log}()}, 20 | \code{\link{cat_line}()}, 21 | \code{\link{char}()}, 22 | \code{\link{check_connection}()}, 23 | \code{\link{cleanup}()}, 24 | \code{\link{connected}()}, 25 | \code{\link{connection_get}()}, 26 | \code{\link{db_download_intern}()}, 27 | \code{\link{db_sqlngths_get}()}, 28 | \code{\link{dir_size}()}, 29 | \code{\link{dwnld_path_get}()}, 30 | \code{\link{dwnld_rcrd_log}()}, 31 | \code{\link{entrez_fasta_get}()}, 32 | \code{\link{entrez_gb_get}()}, 33 | \code{\link{extract_accession}()}, 34 | \code{\link{extract_by_patterns}()}, 35 | \code{\link{extract_clean_sequence}()}, 36 | \code{\link{extract_definition}()}, 37 | \code{\link{extract_features}()}, 38 | \code{\link{extract_inforecpart}()}, 39 | \code{\link{extract_keywords}()}, 40 | \code{\link{extract_locus}()}, 41 | \code{\link{extract_organism}()}, 42 | \code{\link{extract_seqrecpart}()}, 43 | \code{\link{extract_sequence}()}, 44 | \code{\link{extract_version}()}, 45 | \code{\link{file_download}()}, 46 | \code{\link{filename_log}()}, 47 | \code{\link{flatfile_read}()}, 48 | \code{\link{gb_build}()}, 49 | \code{\link{gb_df_create}()}, 50 | \code{\link{gb_df_generate}()}, 51 | \code{\link{gb_sql_add}()}, 52 | \code{\link{gb_sql_query}()}, 53 | \code{\link{gbrelease_check}()}, 54 | \code{\link{gbrelease_get}()}, 55 | \code{\link{gbrelease_log}()}, 56 | \code{\link{has_data}()}, 57 | \code{\link{identify_downloadable_files}()}, 58 | \code{\link{last_add_get}()}, 59 | \code{\link{last_dwnld_get}()}, 60 | \code{\link{last_entry_get}()}, 61 | \code{\link{latest_genbank_release}()}, 62 | \code{\link{latest_genbank_release_notes}()}, 63 | \code{\link{message_missing}()}, 64 | \code{\link{mock_def}()}, 65 | \code{\link{mock_gb_df_generate}()}, 66 | \code{\link{mock_org}()}, 67 | \code{\link{mock_rec}()}, 68 | \code{\link{mock_seq}()}, 69 | \code{\link{predict_datasizes}()}, 70 | \code{\link{readme_log}()}, 71 | \code{\link{restez_connect}()}, 72 | \code{\link{restez_disconnect}()}, 73 | \code{\link{restez_path_check}()}, 74 | \code{\link{restez_rl}()}, 75 | \code{\link{search_gz}()}, 76 | \code{\link{seshinfo_log}()}, 77 | \code{\link{setup}()}, 78 | \code{\link{slctn_get}()}, 79 | \code{\link{slctn_log}()}, 80 | \code{\link{sql_path_get}()}, 81 | \code{\link{stat}()}, 82 | \code{\link{status_class}()}, 83 | \code{\link{testdatadir_get}()} 84 | } 85 | \concept{private} 86 | -------------------------------------------------------------------------------- /man/demo_db_create.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/db-setup-tools.R 3 | \name{demo_db_create} 4 | \alias{demo_db_create} 5 | \title{Create demo database} 6 | \usage{ 7 | demo_db_create(db_type = "nucleotide", n = 100) 8 | } 9 | \arguments{ 10 | \item{db_type}{character, database type} 11 | 12 | \item{n}{integer, number of mock sequences} 13 | } 14 | \description{ 15 | Creates a local mock SQL database 16 | from package test data for demonstration purposes. 17 | No internet connection required. 18 | } 19 | \examples{ 20 | library(restez) 21 | # set the restez path to a temporary dir 22 | restez_path_set(filepath = tempdir()) 23 | # create demo database 24 | demo_db_create(n = 5) 25 | # in the demo, IDs are 'demo_1', 'demo_2' ... 26 | (gb_sequence_get(id = 'demo_1')) 27 | 28 | # Delete a demo database after an example 29 | db_delete(everything = TRUE) 30 | } 31 | \seealso{ 32 | Other database: 33 | \code{\link{count_db_ids}()}, 34 | \code{\link{db_create}()}, 35 | \code{\link{db_delete}()}, 36 | \code{\link{db_download}()}, 37 | \code{\link{is_in_db}()}, 38 | \code{\link{list_db_ids}()} 39 | } 40 | \concept{database} 41 | -------------------------------------------------------------------------------- /man/dir_size.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{dir_size} 4 | \alias{dir_size} 5 | \title{Calculate the size of a directory} 6 | \usage{ 7 | dir_size(fp) 8 | } 9 | \arguments{ 10 | \item{fp}{File path, character} 11 | } 12 | \value{ 13 | numeric 14 | } 15 | \description{ 16 | Returns the size of directory in GB 17 | } 18 | \seealso{ 19 | Other private: 20 | \code{\link{add_rcrd_log}()}, 21 | \code{\link{cat_line}()}, 22 | \code{\link{char}()}, 23 | \code{\link{check_connection}()}, 24 | \code{\link{cleanup}()}, 25 | \code{\link{connected}()}, 26 | \code{\link{connection_get}()}, 27 | \code{\link{db_download_intern}()}, 28 | \code{\link{db_sqlngths_get}()}, 29 | \code{\link{db_sqlngths_log}()}, 30 | \code{\link{dwnld_path_get}()}, 31 | \code{\link{dwnld_rcrd_log}()}, 32 | \code{\link{entrez_fasta_get}()}, 33 | \code{\link{entrez_gb_get}()}, 34 | \code{\link{extract_accession}()}, 35 | \code{\link{extract_by_patterns}()}, 36 | \code{\link{extract_clean_sequence}()}, 37 | \code{\link{extract_definition}()}, 38 | \code{\link{extract_features}()}, 39 | \code{\link{extract_inforecpart}()}, 40 | \code{\link{extract_keywords}()}, 41 | \code{\link{extract_locus}()}, 42 | \code{\link{extract_organism}()}, 43 | \code{\link{extract_seqrecpart}()}, 44 | \code{\link{extract_sequence}()}, 45 | \code{\link{extract_version}()}, 46 | \code{\link{file_download}()}, 47 | \code{\link{filename_log}()}, 48 | \code{\link{flatfile_read}()}, 49 | \code{\link{gb_build}()}, 50 | \code{\link{gb_df_create}()}, 51 | \code{\link{gb_df_generate}()}, 52 | \code{\link{gb_sql_add}()}, 53 | \code{\link{gb_sql_query}()}, 54 | \code{\link{gbrelease_check}()}, 55 | \code{\link{gbrelease_get}()}, 56 | \code{\link{gbrelease_log}()}, 57 | \code{\link{has_data}()}, 58 | \code{\link{identify_downloadable_files}()}, 59 | \code{\link{last_add_get}()}, 60 | \code{\link{last_dwnld_get}()}, 61 | \code{\link{last_entry_get}()}, 62 | \code{\link{latest_genbank_release}()}, 63 | \code{\link{latest_genbank_release_notes}()}, 64 | \code{\link{message_missing}()}, 65 | \code{\link{mock_def}()}, 66 | \code{\link{mock_gb_df_generate}()}, 67 | \code{\link{mock_org}()}, 68 | \code{\link{mock_rec}()}, 69 | \code{\link{mock_seq}()}, 70 | \code{\link{predict_datasizes}()}, 71 | \code{\link{readme_log}()}, 72 | \code{\link{restez_connect}()}, 73 | \code{\link{restez_disconnect}()}, 74 | \code{\link{restez_path_check}()}, 75 | \code{\link{restez_rl}()}, 76 | \code{\link{search_gz}()}, 77 | \code{\link{seshinfo_log}()}, 78 | \code{\link{setup}()}, 79 | \code{\link{slctn_get}()}, 80 | \code{\link{slctn_log}()}, 81 | \code{\link{sql_path_get}()}, 82 | \code{\link{stat}()}, 83 | \code{\link{status_class}()}, 84 | \code{\link{testdatadir_get}()} 85 | } 86 | \concept{private} 87 | -------------------------------------------------------------------------------- /man/dwnld_path_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/filepath-tools.R 3 | \name{dwnld_path_get} 4 | \alias{dwnld_path_get} 5 | \title{Get dwnld path} 6 | \usage{ 7 | dwnld_path_get() 8 | } 9 | \value{ 10 | character 11 | } 12 | \description{ 13 | Return path to folder where raw .seq files 14 | are stored. 15 | } 16 | \seealso{ 17 | Other private: 18 | \code{\link{add_rcrd_log}()}, 19 | \code{\link{cat_line}()}, 20 | \code{\link{char}()}, 21 | \code{\link{check_connection}()}, 22 | \code{\link{cleanup}()}, 23 | \code{\link{connected}()}, 24 | \code{\link{connection_get}()}, 25 | \code{\link{db_download_intern}()}, 26 | \code{\link{db_sqlngths_get}()}, 27 | \code{\link{db_sqlngths_log}()}, 28 | \code{\link{dir_size}()}, 29 | \code{\link{dwnld_rcrd_log}()}, 30 | \code{\link{entrez_fasta_get}()}, 31 | \code{\link{entrez_gb_get}()}, 32 | \code{\link{extract_accession}()}, 33 | \code{\link{extract_by_patterns}()}, 34 | \code{\link{extract_clean_sequence}()}, 35 | \code{\link{extract_definition}()}, 36 | \code{\link{extract_features}()}, 37 | \code{\link{extract_inforecpart}()}, 38 | \code{\link{extract_keywords}()}, 39 | \code{\link{extract_locus}()}, 40 | \code{\link{extract_organism}()}, 41 | \code{\link{extract_seqrecpart}()}, 42 | \code{\link{extract_sequence}()}, 43 | \code{\link{extract_version}()}, 44 | \code{\link{file_download}()}, 45 | \code{\link{filename_log}()}, 46 | \code{\link{flatfile_read}()}, 47 | \code{\link{gb_build}()}, 48 | \code{\link{gb_df_create}()}, 49 | \code{\link{gb_df_generate}()}, 50 | \code{\link{gb_sql_add}()}, 51 | \code{\link{gb_sql_query}()}, 52 | \code{\link{gbrelease_check}()}, 53 | \code{\link{gbrelease_get}()}, 54 | \code{\link{gbrelease_log}()}, 55 | \code{\link{has_data}()}, 56 | \code{\link{identify_downloadable_files}()}, 57 | \code{\link{last_add_get}()}, 58 | \code{\link{last_dwnld_get}()}, 59 | \code{\link{last_entry_get}()}, 60 | \code{\link{latest_genbank_release}()}, 61 | \code{\link{latest_genbank_release_notes}()}, 62 | \code{\link{message_missing}()}, 63 | \code{\link{mock_def}()}, 64 | \code{\link{mock_gb_df_generate}()}, 65 | \code{\link{mock_org}()}, 66 | \code{\link{mock_rec}()}, 67 | \code{\link{mock_seq}()}, 68 | \code{\link{predict_datasizes}()}, 69 | \code{\link{readme_log}()}, 70 | \code{\link{restez_connect}()}, 71 | \code{\link{restez_disconnect}()}, 72 | \code{\link{restez_path_check}()}, 73 | \code{\link{restez_rl}()}, 74 | \code{\link{search_gz}()}, 75 | \code{\link{seshinfo_log}()}, 76 | \code{\link{setup}()}, 77 | \code{\link{slctn_get}()}, 78 | \code{\link{slctn_log}()}, 79 | \code{\link{sql_path_get}()}, 80 | \code{\link{stat}()}, 81 | \code{\link{status_class}()}, 82 | \code{\link{testdatadir_get}()} 83 | } 84 | \concept{private} 85 | -------------------------------------------------------------------------------- /man/entrez_fetch.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/rentrez-wrappers.R 3 | \name{entrez_fetch} 4 | \alias{entrez_fetch} 5 | \title{Entrez fetch} 6 | \usage{ 7 | entrez_fetch(db, id = NULL, rettype, retmode = "", ...) 8 | } 9 | \arguments{ 10 | \item{db}{character, name of the database} 11 | 12 | \item{id}{vector, unique ID(s) for record(s)} 13 | 14 | \item{rettype}{character, data format} 15 | 16 | \item{retmode}{character, data mode} 17 | 18 | \item{...}{Arguments to be passed on to rentrez} 19 | } 20 | \value{ 21 | character string containing the file created 22 | } 23 | \description{ 24 | Wrapper for rentrez::entrez_fetch. 25 | } 26 | \details{ 27 | Attempts to first search local database with user-specified 28 | parameters, if the record is missing in the database, the function then 29 | calls rentrez::entrez_fetch to search GenBank remotely. 30 | 31 | \code{rettype='fasta'} and \code{rettype='gb'} are respectively equivalent to 32 | \code{\link[=gb_fasta_get]{gb_fasta_get()}} and \code{\link[=gb_record_get]{gb_record_get()}}. 33 | } 34 | \note{ 35 | It is advisable to call restez and rentrez functions with '::' notation 36 | rather than library() calls to avoid namespace issues. e.g. 37 | restez::entrez_fetch(). 38 | } 39 | \section{Supported return types and modes}{ 40 | 41 | XML retmode is not supported. Rettypes 'seqid', 'ft', 'acc' and 'uilist' 42 | are also not supported. 43 | } 44 | 45 | \examples{ 46 | library(restez) 47 | restez_path_set(tempdir()) 48 | demo_db_create(n = 5) 49 | # return fasta record 50 | fasta_res <- entrez_fetch(db = 'nucleotide', 51 | id = c('demo_1', 'demo_2'), 52 | rettype = 'fasta') 53 | cat(fasta_res) 54 | # return whole GB record in text format 55 | gb_res <- entrez_fetch(db = 'nucleotide', 56 | id = c('demo_1', 'demo_2'), 57 | rettype = 'gb') 58 | cat(gb_res) 59 | # NOT RUN 60 | # whereas these request would go through rentrez 61 | # fasta_res <- entrez_fetch(db = 'nucleotide', 62 | # id = c('S71333', 'S71334'), 63 | # rettype = 'fasta') 64 | # gb_res <- entrez_fetch(db = 'nucleotide', 65 | # id = c('S71333', 'S71334'), 66 | # rettype = 'gb') 67 | 68 | # delete demo after example 69 | db_delete(everything = TRUE) 70 | } 71 | \seealso{ 72 | \code{\link[rentrez:entrez_fetch]{rentrez::entrez_fetch()}} 73 | } 74 | \concept{entrez} 75 | -------------------------------------------------------------------------------- /man/extract_accession.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{extract_accession} 4 | \alias{extract_accession} 5 | \title{Extract accession} 6 | \usage{ 7 | extract_accession(record) 8 | } 9 | \arguments{ 10 | \item{record}{GenBank record in text format, character} 11 | } 12 | \value{ 13 | character 14 | } 15 | \description{ 16 | Return accession ID from GenBank record 17 | } 18 | \details{ 19 | If element is not found, '' returned. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_by_patterns}()}, 39 | \code{\link{extract_clean_sequence}()}, 40 | \code{\link{extract_definition}()}, 41 | \code{\link{extract_features}()}, 42 | \code{\link{extract_inforecpart}()}, 43 | \code{\link{extract_keywords}()}, 44 | \code{\link{extract_locus}()}, 45 | \code{\link{extract_organism}()}, 46 | \code{\link{extract_seqrecpart}()}, 47 | \code{\link{extract_sequence}()}, 48 | \code{\link{extract_version}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/extract_definition.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{extract_definition} 4 | \alias{extract_definition} 5 | \title{Extract definition} 6 | \usage{ 7 | extract_definition(record) 8 | } 9 | \arguments{ 10 | \item{record}{GenBank record in text format, character} 11 | } 12 | \value{ 13 | character 14 | } 15 | \description{ 16 | Return definition from GenBank record. 17 | } 18 | \details{ 19 | If element is not found, '' returned. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_accession}()}, 39 | \code{\link{extract_by_patterns}()}, 40 | \code{\link{extract_clean_sequence}()}, 41 | \code{\link{extract_features}()}, 42 | \code{\link{extract_inforecpart}()}, 43 | \code{\link{extract_keywords}()}, 44 | \code{\link{extract_locus}()}, 45 | \code{\link{extract_organism}()}, 46 | \code{\link{extract_seqrecpart}()}, 47 | \code{\link{extract_sequence}()}, 48 | \code{\link{extract_version}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/extract_features.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{extract_features} 4 | \alias{extract_features} 5 | \title{Extract features} 6 | \usage{ 7 | extract_features(record) 8 | } 9 | \arguments{ 10 | \item{record}{GenBank record in text format, character} 11 | } 12 | \value{ 13 | list of lists 14 | } 15 | \description{ 16 | Return feature table as list from GenBank record 17 | } 18 | \details{ 19 | If element is not found, empty list returned. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_accession}()}, 39 | \code{\link{extract_by_patterns}()}, 40 | \code{\link{extract_clean_sequence}()}, 41 | \code{\link{extract_definition}()}, 42 | \code{\link{extract_inforecpart}()}, 43 | \code{\link{extract_keywords}()}, 44 | \code{\link{extract_locus}()}, 45 | \code{\link{extract_organism}()}, 46 | \code{\link{extract_seqrecpart}()}, 47 | \code{\link{extract_sequence}()}, 48 | \code{\link{extract_version}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/extract_keywords.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{extract_keywords} 4 | \alias{extract_keywords} 5 | \title{Extract keywords} 6 | \usage{ 7 | extract_keywords(record) 8 | } 9 | \arguments{ 10 | \item{record}{GenBank record in text format, character} 11 | } 12 | \value{ 13 | character vector 14 | } 15 | \description{ 16 | Return keywords as list from GenBank record 17 | } 18 | \details{ 19 | If element is not found, '' returned. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_accession}()}, 39 | \code{\link{extract_by_patterns}()}, 40 | \code{\link{extract_clean_sequence}()}, 41 | \code{\link{extract_definition}()}, 42 | \code{\link{extract_features}()}, 43 | \code{\link{extract_inforecpart}()}, 44 | \code{\link{extract_locus}()}, 45 | \code{\link{extract_organism}()}, 46 | \code{\link{extract_seqrecpart}()}, 47 | \code{\link{extract_sequence}()}, 48 | \code{\link{extract_version}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/extract_locus.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{extract_locus} 4 | \alias{extract_locus} 5 | \title{Extract locus} 6 | \usage{ 7 | extract_locus(record) 8 | } 9 | \arguments{ 10 | \item{record}{GenBank record in text format, character} 11 | } 12 | \value{ 13 | named character vector 14 | } 15 | \description{ 16 | Return locus information from GenBank record 17 | } 18 | \details{ 19 | If element is not found, '' returned. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_accession}()}, 39 | \code{\link{extract_by_patterns}()}, 40 | \code{\link{extract_clean_sequence}()}, 41 | \code{\link{extract_definition}()}, 42 | \code{\link{extract_features}()}, 43 | \code{\link{extract_inforecpart}()}, 44 | \code{\link{extract_keywords}()}, 45 | \code{\link{extract_organism}()}, 46 | \code{\link{extract_seqrecpart}()}, 47 | \code{\link{extract_sequence}()}, 48 | \code{\link{extract_version}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/extract_organism.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{extract_organism} 4 | \alias{extract_organism} 5 | \title{Extract organism} 6 | \usage{ 7 | extract_organism(record) 8 | } 9 | \arguments{ 10 | \item{record}{GenBank record in text format, character} 11 | } 12 | \value{ 13 | character 14 | } 15 | \description{ 16 | Return organism name from GenBank record 17 | } 18 | \details{ 19 | If element is not found, '' returned. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_accession}()}, 39 | \code{\link{extract_by_patterns}()}, 40 | \code{\link{extract_clean_sequence}()}, 41 | \code{\link{extract_definition}()}, 42 | \code{\link{extract_features}()}, 43 | \code{\link{extract_inforecpart}()}, 44 | \code{\link{extract_keywords}()}, 45 | \code{\link{extract_locus}()}, 46 | \code{\link{extract_seqrecpart}()}, 47 | \code{\link{extract_sequence}()}, 48 | \code{\link{extract_version}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/extract_sequence.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{extract_sequence} 4 | \alias{extract_sequence} 5 | \title{Extract sequence} 6 | \usage{ 7 | extract_sequence(record) 8 | } 9 | \arguments{ 10 | \item{record}{GenBank record in text format, character} 11 | } 12 | \value{ 13 | character 14 | } 15 | \description{ 16 | Return sequence from GenBank record 17 | } 18 | \details{ 19 | If element is not found, '' returned. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_accession}()}, 39 | \code{\link{extract_by_patterns}()}, 40 | \code{\link{extract_clean_sequence}()}, 41 | \code{\link{extract_definition}()}, 42 | \code{\link{extract_features}()}, 43 | \code{\link{extract_inforecpart}()}, 44 | \code{\link{extract_keywords}()}, 45 | \code{\link{extract_locus}()}, 46 | \code{\link{extract_organism}()}, 47 | \code{\link{extract_seqrecpart}()}, 48 | \code{\link{extract_version}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/extract_version.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{extract_version} 4 | \alias{extract_version} 5 | \title{Extract version} 6 | \usage{ 7 | extract_version(record) 8 | } 9 | \arguments{ 10 | \item{record}{GenBank record in text format, character} 11 | } 12 | \value{ 13 | character 14 | } 15 | \description{ 16 | Return accession + version ID from GenBank record 17 | } 18 | \details{ 19 | If element is not found, '' returned. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_accession}()}, 39 | \code{\link{extract_by_patterns}()}, 40 | \code{\link{extract_clean_sequence}()}, 41 | \code{\link{extract_definition}()}, 42 | \code{\link{extract_features}()}, 43 | \code{\link{extract_inforecpart}()}, 44 | \code{\link{extract_keywords}()}, 45 | \code{\link{extract_locus}()}, 46 | \code{\link{extract_organism}()}, 47 | \code{\link{extract_seqrecpart}()}, 48 | \code{\link{extract_sequence}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/filename_log.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{filename_log} 4 | \alias{filename_log} 5 | \title{Write filenames to log files} 6 | \usage{ 7 | filename_log(fl, fp) 8 | } 9 | \arguments{ 10 | \item{fl}{file name, character} 11 | 12 | \item{fp}{filepath to log file, character} 13 | } 14 | \description{ 15 | Record a filename in a log file along with GB release and time. 16 | } 17 | \seealso{ 18 | Other private: 19 | \code{\link{add_rcrd_log}()}, 20 | \code{\link{cat_line}()}, 21 | \code{\link{char}()}, 22 | \code{\link{check_connection}()}, 23 | \code{\link{cleanup}()}, 24 | \code{\link{connected}()}, 25 | \code{\link{connection_get}()}, 26 | \code{\link{db_download_intern}()}, 27 | \code{\link{db_sqlngths_get}()}, 28 | \code{\link{db_sqlngths_log}()}, 29 | \code{\link{dir_size}()}, 30 | \code{\link{dwnld_path_get}()}, 31 | \code{\link{dwnld_rcrd_log}()}, 32 | \code{\link{entrez_fasta_get}()}, 33 | \code{\link{entrez_gb_get}()}, 34 | \code{\link{extract_accession}()}, 35 | \code{\link{extract_by_patterns}()}, 36 | \code{\link{extract_clean_sequence}()}, 37 | \code{\link{extract_definition}()}, 38 | \code{\link{extract_features}()}, 39 | \code{\link{extract_inforecpart}()}, 40 | \code{\link{extract_keywords}()}, 41 | \code{\link{extract_locus}()}, 42 | \code{\link{extract_organism}()}, 43 | \code{\link{extract_seqrecpart}()}, 44 | \code{\link{extract_sequence}()}, 45 | \code{\link{extract_version}()}, 46 | \code{\link{file_download}()}, 47 | \code{\link{flatfile_read}()}, 48 | \code{\link{gb_build}()}, 49 | \code{\link{gb_df_create}()}, 50 | \code{\link{gb_df_generate}()}, 51 | \code{\link{gb_sql_add}()}, 52 | \code{\link{gb_sql_query}()}, 53 | \code{\link{gbrelease_check}()}, 54 | \code{\link{gbrelease_get}()}, 55 | \code{\link{gbrelease_log}()}, 56 | \code{\link{has_data}()}, 57 | \code{\link{identify_downloadable_files}()}, 58 | \code{\link{last_add_get}()}, 59 | \code{\link{last_dwnld_get}()}, 60 | \code{\link{last_entry_get}()}, 61 | \code{\link{latest_genbank_release}()}, 62 | \code{\link{latest_genbank_release_notes}()}, 63 | \code{\link{message_missing}()}, 64 | \code{\link{mock_def}()}, 65 | \code{\link{mock_gb_df_generate}()}, 66 | \code{\link{mock_org}()}, 67 | \code{\link{mock_rec}()}, 68 | \code{\link{mock_seq}()}, 69 | \code{\link{predict_datasizes}()}, 70 | \code{\link{readme_log}()}, 71 | \code{\link{restez_connect}()}, 72 | \code{\link{restez_disconnect}()}, 73 | \code{\link{restez_path_check}()}, 74 | \code{\link{restez_rl}()}, 75 | \code{\link{search_gz}()}, 76 | \code{\link{seshinfo_log}()}, 77 | \code{\link{setup}()}, 78 | \code{\link{slctn_get}()}, 79 | \code{\link{slctn_log}()}, 80 | \code{\link{sql_path_get}()}, 81 | \code{\link{stat}()}, 82 | \code{\link{status_class}()}, 83 | \code{\link{testdatadir_get}()} 84 | } 85 | \concept{private} 86 | -------------------------------------------------------------------------------- /man/flatfile_read.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-setup-tools.R 3 | \name{flatfile_read} 4 | \alias{flatfile_read} 5 | \title{Read flatfile sequence records} 6 | \usage{ 7 | flatfile_read(flpth) 8 | } 9 | \arguments{ 10 | \item{flpth}{Path to .seq file} 11 | } 12 | \value{ 13 | list of GenBank records in text format 14 | } 15 | \description{ 16 | Read records from a .seq file. 17 | } 18 | \seealso{ 19 | Other private: 20 | \code{\link{add_rcrd_log}()}, 21 | \code{\link{cat_line}()}, 22 | \code{\link{char}()}, 23 | \code{\link{check_connection}()}, 24 | \code{\link{cleanup}()}, 25 | \code{\link{connected}()}, 26 | \code{\link{connection_get}()}, 27 | \code{\link{db_download_intern}()}, 28 | \code{\link{db_sqlngths_get}()}, 29 | \code{\link{db_sqlngths_log}()}, 30 | \code{\link{dir_size}()}, 31 | \code{\link{dwnld_path_get}()}, 32 | \code{\link{dwnld_rcrd_log}()}, 33 | \code{\link{entrez_fasta_get}()}, 34 | \code{\link{entrez_gb_get}()}, 35 | \code{\link{extract_accession}()}, 36 | \code{\link{extract_by_patterns}()}, 37 | \code{\link{extract_clean_sequence}()}, 38 | \code{\link{extract_definition}()}, 39 | \code{\link{extract_features}()}, 40 | \code{\link{extract_inforecpart}()}, 41 | \code{\link{extract_keywords}()}, 42 | \code{\link{extract_locus}()}, 43 | \code{\link{extract_organism}()}, 44 | \code{\link{extract_seqrecpart}()}, 45 | \code{\link{extract_sequence}()}, 46 | \code{\link{extract_version}()}, 47 | \code{\link{file_download}()}, 48 | \code{\link{filename_log}()}, 49 | \code{\link{gb_build}()}, 50 | \code{\link{gb_df_create}()}, 51 | \code{\link{gb_df_generate}()}, 52 | \code{\link{gb_sql_add}()}, 53 | \code{\link{gb_sql_query}()}, 54 | \code{\link{gbrelease_check}()}, 55 | \code{\link{gbrelease_get}()}, 56 | \code{\link{gbrelease_log}()}, 57 | \code{\link{has_data}()}, 58 | \code{\link{identify_downloadable_files}()}, 59 | \code{\link{last_add_get}()}, 60 | \code{\link{last_dwnld_get}()}, 61 | \code{\link{last_entry_get}()}, 62 | \code{\link{latest_genbank_release}()}, 63 | \code{\link{latest_genbank_release_notes}()}, 64 | \code{\link{message_missing}()}, 65 | \code{\link{mock_def}()}, 66 | \code{\link{mock_gb_df_generate}()}, 67 | \code{\link{mock_org}()}, 68 | \code{\link{mock_rec}()}, 69 | \code{\link{mock_seq}()}, 70 | \code{\link{predict_datasizes}()}, 71 | \code{\link{readme_log}()}, 72 | \code{\link{restez_connect}()}, 73 | \code{\link{restez_disconnect}()}, 74 | \code{\link{restez_path_check}()}, 75 | \code{\link{restez_rl}()}, 76 | \code{\link{search_gz}()}, 77 | \code{\link{seshinfo_log}()}, 78 | \code{\link{setup}()}, 79 | \code{\link{slctn_get}()}, 80 | \code{\link{slctn_log}()}, 81 | \code{\link{sql_path_get}()}, 82 | \code{\link{stat}()}, 83 | \code{\link{status_class}()}, 84 | \code{\link{testdatadir_get}()} 85 | } 86 | \concept{private} 87 | -------------------------------------------------------------------------------- /man/gb_definition_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-get-tools.R 3 | \name{gb_definition_get} 4 | \alias{gb_definition_get} 5 | \title{Get definition from GenBank} 6 | \usage{ 7 | gb_definition_get(id) 8 | } 9 | \arguments{ 10 | \item{id}{character, sequence accession ID(s)} 11 | } 12 | \value{ 13 | named vector of definitions, if no results found NULL 14 | } 15 | \description{ 16 | Return the definition line 17 | for an accession ID. 18 | } 19 | \examples{ 20 | library(restez) 21 | restez_path_set(filepath = tempdir()) 22 | demo_db_create(n = 5) 23 | (def <- gb_definition_get(id = 'demo_1')) 24 | (defs <- gb_definition_get(id = c('demo_1', 'demo_2'))) 25 | 26 | 27 | # delete demo after example 28 | db_delete(everything = TRUE) 29 | } 30 | \seealso{ 31 | \code{\link[=ncbi_acc_get]{ncbi_acc_get()}} 32 | 33 | Other get: 34 | \code{\link{gb_fasta_get}()}, 35 | \code{\link{gb_organism_get}()}, 36 | \code{\link{gb_record_get}()}, 37 | \code{\link{gb_sequence_get}()}, 38 | \code{\link{gb_version_get}()} 39 | } 40 | \concept{get} 41 | -------------------------------------------------------------------------------- /man/gb_extract.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/extract-tools.R 3 | \name{gb_extract} 4 | \alias{gb_extract} 5 | \title{Extract elements of a GenBank record} 6 | \usage{ 7 | gb_extract( 8 | record, 9 | what = c("accession", "version", "organism", "sequence", "definition", "locus", 10 | "features", "keywords") 11 | ) 12 | } 13 | \arguments{ 14 | \item{record}{GenBank record in text format, character} 15 | 16 | \item{what}{Which element to extract} 17 | } 18 | \value{ 19 | character or list of lists (what='features') or named character 20 | vector (what='locus') 21 | } 22 | \description{ 23 | Return elements of GenBank record e.g. sequence, definition ... 24 | } 25 | \details{ 26 | This function uses a REGEX to extract particular elements of a 27 | GenBank record. All of the what options return a single character with the 28 | exception of 'locus' or 'keywords' that return character vectors and 29 | 'features' that returns a list of lists for all features. 30 | 31 | The accuracy of these functions cannot be guaranteed due to the enormity of 32 | the GenBank database. But the function is regularly tested on a range of 33 | GenBank records. 34 | 35 | Note: all non-latin1 characters are converted to '-'. 36 | } 37 | \examples{ 38 | library(restez) 39 | data('record') 40 | (gb_extract(record = record, what = 'locus')) 41 | } 42 | \concept{parse} 43 | -------------------------------------------------------------------------------- /man/gb_fasta_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-get-tools.R 3 | \name{gb_fasta_get} 4 | \alias{gb_fasta_get} 5 | \title{Get fasta from GenBank} 6 | \usage{ 7 | gb_fasta_get(id, width = 70) 8 | } 9 | \arguments{ 10 | \item{id}{character, sequence accession ID(s)} 11 | 12 | \item{width}{integer, maximum number of characters in a line} 13 | } 14 | \value{ 15 | named vector of fasta sequences, if no results found NULL 16 | } 17 | \description{ 18 | Get sequence and definition data in FASTA format. Equivalent to 19 | \code{rettype='fasta'} in \code{\link[rentrez:entrez_fetch]{rentrez::entrez_fetch()}}. 20 | } 21 | \examples{ 22 | library(restez) 23 | restez_path_set(filepath = tempdir()) 24 | demo_db_create(n = 5) 25 | (fasta <- gb_fasta_get(id = 'demo_1')) 26 | (fastas <- gb_fasta_get(id = c('demo_1', 'demo_2'))) 27 | 28 | 29 | # delete demo after example 30 | db_delete(everything = TRUE) 31 | } 32 | \seealso{ 33 | \code{\link[=ncbi_acc_get]{ncbi_acc_get()}} 34 | 35 | Other get: 36 | \code{\link{gb_definition_get}()}, 37 | \code{\link{gb_organism_get}()}, 38 | \code{\link{gb_record_get}()}, 39 | \code{\link{gb_sequence_get}()}, 40 | \code{\link{gb_version_get}()} 41 | } 42 | \concept{get} 43 | -------------------------------------------------------------------------------- /man/gb_organism_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-get-tools.R 3 | \name{gb_organism_get} 4 | \alias{gb_organism_get} 5 | \title{Get organism from GenBank} 6 | \usage{ 7 | gb_organism_get(id) 8 | } 9 | \arguments{ 10 | \item{id}{character, sequence accession ID(s)} 11 | } 12 | \value{ 13 | named vector of definitions, if no results found NULL 14 | } 15 | \description{ 16 | Return the organism name 17 | for an accession ID. 18 | } 19 | \examples{ 20 | library(restez) 21 | restez_path_set(filepath = tempdir()) 22 | demo_db_create(n = 5) 23 | (org <- gb_organism_get(id = 'demo_1')) 24 | (orgs <- gb_organism_get(id = c('demo_1', 'demo_2'))) 25 | 26 | 27 | # delete demo after example 28 | db_delete(everything = TRUE) 29 | } 30 | \seealso{ 31 | \code{\link[=ncbi_acc_get]{ncbi_acc_get()}} 32 | 33 | Other get: 34 | \code{\link{gb_definition_get}()}, 35 | \code{\link{gb_fasta_get}()}, 36 | \code{\link{gb_record_get}()}, 37 | \code{\link{gb_sequence_get}()}, 38 | \code{\link{gb_version_get}()} 39 | } 40 | \concept{get} 41 | -------------------------------------------------------------------------------- /man/gb_record_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-get-tools.R 3 | \name{gb_record_get} 4 | \alias{gb_record_get} 5 | \title{Get record from GenBank} 6 | \usage{ 7 | gb_record_get(id) 8 | } 9 | \arguments{ 10 | \item{id}{character, sequence accession ID(s)} 11 | } 12 | \value{ 13 | named vector of records, if no results found NULL 14 | } 15 | \description{ 16 | Return the entire GenBank record for an accession ID. 17 | Equivalent to \code{rettype='gb'} in \code{\link[rentrez:entrez_fetch]{rentrez::entrez_fetch()}}. 18 | } 19 | \examples{ 20 | library(restez) 21 | restez_path_set(filepath = tempdir()) 22 | demo_db_create(n = 5) 23 | (rec <- gb_record_get(id = 'demo_1')) 24 | (recs <- gb_record_get(id = c('demo_1', 'demo_2'))) 25 | 26 | 27 | # delete demo after example 28 | db_delete(everything = TRUE) 29 | } 30 | \seealso{ 31 | \code{\link[=ncbi_acc_get]{ncbi_acc_get()}} 32 | 33 | Other get: 34 | \code{\link{gb_definition_get}()}, 35 | \code{\link{gb_fasta_get}()}, 36 | \code{\link{gb_organism_get}()}, 37 | \code{\link{gb_sequence_get}()}, 38 | \code{\link{gb_version_get}()} 39 | } 40 | \concept{get} 41 | -------------------------------------------------------------------------------- /man/gb_sequence_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-get-tools.R 3 | \name{gb_sequence_get} 4 | \alias{gb_sequence_get} 5 | \title{Get sequence from GenBank} 6 | \usage{ 7 | gb_sequence_get(id, dnabin = FALSE) 8 | } 9 | \arguments{ 10 | \item{id}{character, sequence accession ID(s)} 11 | 12 | \item{dnabin}{Logical vector of length 1; should the sequences 13 | be returned using the bit-level coding scheme of the ape package? 14 | Default FALSE.} 15 | } 16 | \value{ 17 | named vector of sequences, if no results found NULL 18 | } 19 | \description{ 20 | Return the sequence(s) for a record(s) 21 | from the accession ID(s). 22 | } 23 | \details{ 24 | For more information about the \code{dnabin} format, see \code{\link[ape:DNAbin]{ape::DNAbin()}}. 25 | } 26 | \examples{ 27 | library(restez) 28 | restez_path_set(filepath = tempdir()) 29 | demo_db_create(n = 5) 30 | (seq <- gb_sequence_get(id = 'demo_1')) 31 | (seqs <- gb_sequence_get(id = c('demo_1', 'demo_2'))) 32 | (fasta_dnabin <- gb_sequence_get(id = 'demo_1', dnabin = TRUE)) 33 | 34 | # delete demo after example 35 | db_delete(everything = TRUE) 36 | 37 | } 38 | \seealso{ 39 | \code{\link[=ncbi_acc_get]{ncbi_acc_get()}} 40 | 41 | Other get: 42 | \code{\link{gb_definition_get}()}, 43 | \code{\link{gb_fasta_get}()}, 44 | \code{\link{gb_organism_get}()}, 45 | \code{\link{gb_record_get}()}, 46 | \code{\link{gb_version_get}()} 47 | } 48 | \concept{get} 49 | -------------------------------------------------------------------------------- /man/gb_sql_add.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-setup-tools.R 3 | \name{gb_sql_add} 4 | \alias{gb_sql_add} 5 | \title{Add to GenBank SQL database} 6 | \usage{ 7 | gb_sql_add(df) 8 | } 9 | \arguments{ 10 | \item{df}{Records data.frame} 11 | } 12 | \description{ 13 | Add records data.frame to SQL-like database. 14 | } 15 | \seealso{ 16 | Other private: 17 | \code{\link{add_rcrd_log}()}, 18 | \code{\link{cat_line}()}, 19 | \code{\link{char}()}, 20 | \code{\link{check_connection}()}, 21 | \code{\link{cleanup}()}, 22 | \code{\link{connected}()}, 23 | \code{\link{connection_get}()}, 24 | \code{\link{db_download_intern}()}, 25 | \code{\link{db_sqlngths_get}()}, 26 | \code{\link{db_sqlngths_log}()}, 27 | \code{\link{dir_size}()}, 28 | \code{\link{dwnld_path_get}()}, 29 | \code{\link{dwnld_rcrd_log}()}, 30 | \code{\link{entrez_fasta_get}()}, 31 | \code{\link{entrez_gb_get}()}, 32 | \code{\link{extract_accession}()}, 33 | \code{\link{extract_by_patterns}()}, 34 | \code{\link{extract_clean_sequence}()}, 35 | \code{\link{extract_definition}()}, 36 | \code{\link{extract_features}()}, 37 | \code{\link{extract_inforecpart}()}, 38 | \code{\link{extract_keywords}()}, 39 | \code{\link{extract_locus}()}, 40 | \code{\link{extract_organism}()}, 41 | \code{\link{extract_seqrecpart}()}, 42 | \code{\link{extract_sequence}()}, 43 | \code{\link{extract_version}()}, 44 | \code{\link{file_download}()}, 45 | \code{\link{filename_log}()}, 46 | \code{\link{flatfile_read}()}, 47 | \code{\link{gb_build}()}, 48 | \code{\link{gb_df_create}()}, 49 | \code{\link{gb_df_generate}()}, 50 | \code{\link{gb_sql_query}()}, 51 | \code{\link{gbrelease_check}()}, 52 | \code{\link{gbrelease_get}()}, 53 | \code{\link{gbrelease_log}()}, 54 | \code{\link{has_data}()}, 55 | \code{\link{identify_downloadable_files}()}, 56 | \code{\link{last_add_get}()}, 57 | \code{\link{last_dwnld_get}()}, 58 | \code{\link{last_entry_get}()}, 59 | \code{\link{latest_genbank_release}()}, 60 | \code{\link{latest_genbank_release_notes}()}, 61 | \code{\link{message_missing}()}, 62 | \code{\link{mock_def}()}, 63 | \code{\link{mock_gb_df_generate}()}, 64 | \code{\link{mock_org}()}, 65 | \code{\link{mock_rec}()}, 66 | \code{\link{mock_seq}()}, 67 | \code{\link{predict_datasizes}()}, 68 | \code{\link{readme_log}()}, 69 | \code{\link{restez_connect}()}, 70 | \code{\link{restez_disconnect}()}, 71 | \code{\link{restez_path_check}()}, 72 | \code{\link{restez_rl}()}, 73 | \code{\link{search_gz}()}, 74 | \code{\link{seshinfo_log}()}, 75 | \code{\link{setup}()}, 76 | \code{\link{slctn_get}()}, 77 | \code{\link{slctn_log}()}, 78 | \code{\link{sql_path_get}()}, 79 | \code{\link{stat}()}, 80 | \code{\link{status_class}()}, 81 | \code{\link{testdatadir_get}()} 82 | } 83 | \concept{private} 84 | -------------------------------------------------------------------------------- /man/gb_sql_query.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-get-tools.R 3 | \name{gb_sql_query} 4 | \alias{gb_sql_query} 5 | \title{Query the GenBank SQL} 6 | \usage{ 7 | gb_sql_query(nm, id) 8 | } 9 | \arguments{ 10 | \item{nm}{character, column name} 11 | 12 | \item{id}{character, sequence accession ID(s)} 13 | } 14 | \value{ 15 | data.frame 16 | } 17 | \description{ 18 | Generic query function for retrieving 19 | data from the SQL database for the get functions. 20 | } 21 | \seealso{ 22 | Other private: 23 | \code{\link{add_rcrd_log}()}, 24 | \code{\link{cat_line}()}, 25 | \code{\link{char}()}, 26 | \code{\link{check_connection}()}, 27 | \code{\link{cleanup}()}, 28 | \code{\link{connected}()}, 29 | \code{\link{connection_get}()}, 30 | \code{\link{db_download_intern}()}, 31 | \code{\link{db_sqlngths_get}()}, 32 | \code{\link{db_sqlngths_log}()}, 33 | \code{\link{dir_size}()}, 34 | \code{\link{dwnld_path_get}()}, 35 | \code{\link{dwnld_rcrd_log}()}, 36 | \code{\link{entrez_fasta_get}()}, 37 | \code{\link{entrez_gb_get}()}, 38 | \code{\link{extract_accession}()}, 39 | \code{\link{extract_by_patterns}()}, 40 | \code{\link{extract_clean_sequence}()}, 41 | \code{\link{extract_definition}()}, 42 | \code{\link{extract_features}()}, 43 | \code{\link{extract_inforecpart}()}, 44 | \code{\link{extract_keywords}()}, 45 | \code{\link{extract_locus}()}, 46 | \code{\link{extract_organism}()}, 47 | \code{\link{extract_seqrecpart}()}, 48 | \code{\link{extract_sequence}()}, 49 | \code{\link{extract_version}()}, 50 | \code{\link{file_download}()}, 51 | \code{\link{filename_log}()}, 52 | \code{\link{flatfile_read}()}, 53 | \code{\link{gb_build}()}, 54 | \code{\link{gb_df_create}()}, 55 | \code{\link{gb_df_generate}()}, 56 | \code{\link{gb_sql_add}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{mock_seq}()}, 73 | \code{\link{predict_datasizes}()}, 74 | \code{\link{readme_log}()}, 75 | \code{\link{restez_connect}()}, 76 | \code{\link{restez_disconnect}()}, 77 | \code{\link{restez_path_check}()}, 78 | \code{\link{restez_rl}()}, 79 | \code{\link{search_gz}()}, 80 | \code{\link{seshinfo_log}()}, 81 | \code{\link{setup}()}, 82 | \code{\link{slctn_get}()}, 83 | \code{\link{slctn_log}()}, 84 | \code{\link{sql_path_get}()}, 85 | \code{\link{stat}()}, 86 | \code{\link{status_class}()}, 87 | \code{\link{testdatadir_get}()} 88 | } 89 | \concept{private} 90 | -------------------------------------------------------------------------------- /man/gb_version_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/gb-get-tools.R 3 | \name{gb_version_get} 4 | \alias{gb_version_get} 5 | \title{Get version from GenBank} 6 | \usage{ 7 | gb_version_get(id) 8 | } 9 | \arguments{ 10 | \item{id}{character, sequence accession ID(s)} 11 | } 12 | \value{ 13 | named vector of versions, if no results found NULL 14 | } 15 | \description{ 16 | Return the accession version 17 | for an accession ID. 18 | } 19 | \examples{ 20 | library(restez) 21 | restez_path_set(filepath = tempdir()) 22 | demo_db_create(n = 5) 23 | (ver <- gb_version_get(id = 'demo_1')) 24 | (vers <- gb_version_get(id = c('demo_1', 'demo_2'))) 25 | 26 | 27 | # delete demo after example 28 | db_delete(everything = TRUE) 29 | 30 | 31 | } 32 | \seealso{ 33 | \code{\link[=ncbi_acc_get]{ncbi_acc_get()}} 34 | 35 | Other get: 36 | \code{\link{gb_definition_get}()}, 37 | \code{\link{gb_fasta_get}()}, 38 | \code{\link{gb_organism_get}()}, 39 | \code{\link{gb_record_get}()}, 40 | \code{\link{gb_sequence_get}()} 41 | } 42 | \concept{get} 43 | -------------------------------------------------------------------------------- /man/gbrelease_check.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{gbrelease_check} 4 | \alias{gbrelease_check} 5 | \title{Check if the last GenBank release number is the latest} 6 | \usage{ 7 | gbrelease_check() 8 | } 9 | \value{ 10 | logical 11 | } 12 | \description{ 13 | Returns TRUE if the GenBank release number is the most recent 14 | GenBank release available. 15 | } 16 | \seealso{ 17 | Other private: 18 | \code{\link{add_rcrd_log}()}, 19 | \code{\link{cat_line}()}, 20 | \code{\link{char}()}, 21 | \code{\link{check_connection}()}, 22 | \code{\link{cleanup}()}, 23 | \code{\link{connected}()}, 24 | \code{\link{connection_get}()}, 25 | \code{\link{db_download_intern}()}, 26 | \code{\link{db_sqlngths_get}()}, 27 | \code{\link{db_sqlngths_log}()}, 28 | \code{\link{dir_size}()}, 29 | \code{\link{dwnld_path_get}()}, 30 | \code{\link{dwnld_rcrd_log}()}, 31 | \code{\link{entrez_fasta_get}()}, 32 | \code{\link{entrez_gb_get}()}, 33 | \code{\link{extract_accession}()}, 34 | \code{\link{extract_by_patterns}()}, 35 | \code{\link{extract_clean_sequence}()}, 36 | \code{\link{extract_definition}()}, 37 | \code{\link{extract_features}()}, 38 | \code{\link{extract_inforecpart}()}, 39 | \code{\link{extract_keywords}()}, 40 | \code{\link{extract_locus}()}, 41 | \code{\link{extract_organism}()}, 42 | \code{\link{extract_seqrecpart}()}, 43 | \code{\link{extract_sequence}()}, 44 | \code{\link{extract_version}()}, 45 | \code{\link{file_download}()}, 46 | \code{\link{filename_log}()}, 47 | \code{\link{flatfile_read}()}, 48 | \code{\link{gb_build}()}, 49 | \code{\link{gb_df_create}()}, 50 | \code{\link{gb_df_generate}()}, 51 | \code{\link{gb_sql_add}()}, 52 | \code{\link{gb_sql_query}()}, 53 | \code{\link{gbrelease_get}()}, 54 | \code{\link{gbrelease_log}()}, 55 | \code{\link{has_data}()}, 56 | \code{\link{identify_downloadable_files}()}, 57 | \code{\link{last_add_get}()}, 58 | \code{\link{last_dwnld_get}()}, 59 | \code{\link{last_entry_get}()}, 60 | \code{\link{latest_genbank_release}()}, 61 | \code{\link{latest_genbank_release_notes}()}, 62 | \code{\link{message_missing}()}, 63 | \code{\link{mock_def}()}, 64 | \code{\link{mock_gb_df_generate}()}, 65 | \code{\link{mock_org}()}, 66 | \code{\link{mock_rec}()}, 67 | \code{\link{mock_seq}()}, 68 | \code{\link{predict_datasizes}()}, 69 | \code{\link{readme_log}()}, 70 | \code{\link{restez_connect}()}, 71 | \code{\link{restez_disconnect}()}, 72 | \code{\link{restez_path_check}()}, 73 | \code{\link{restez_rl}()}, 74 | \code{\link{search_gz}()}, 75 | \code{\link{seshinfo_log}()}, 76 | \code{\link{setup}()}, 77 | \code{\link{slctn_get}()}, 78 | \code{\link{slctn_log}()}, 79 | \code{\link{sql_path_get}()}, 80 | \code{\link{stat}()}, 81 | \code{\link{status_class}()}, 82 | \code{\link{testdatadir_get}()} 83 | } 84 | \concept{private} 85 | -------------------------------------------------------------------------------- /man/gbrelease_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{gbrelease_get} 4 | \alias{gbrelease_get} 5 | \title{Get the GenBank release number in the restez path} 6 | \usage{ 7 | gbrelease_get() 8 | } 9 | \value{ 10 | character 11 | } 12 | \description{ 13 | Returns the GenBank release number. Returns empty character 14 | if none found. 15 | } 16 | \details{ 17 | If no file found, returns empty character vector. 18 | } 19 | \seealso{ 20 | Other private: 21 | \code{\link{add_rcrd_log}()}, 22 | \code{\link{cat_line}()}, 23 | \code{\link{char}()}, 24 | \code{\link{check_connection}()}, 25 | \code{\link{cleanup}()}, 26 | \code{\link{connected}()}, 27 | \code{\link{connection_get}()}, 28 | \code{\link{db_download_intern}()}, 29 | \code{\link{db_sqlngths_get}()}, 30 | \code{\link{db_sqlngths_log}()}, 31 | \code{\link{dir_size}()}, 32 | \code{\link{dwnld_path_get}()}, 33 | \code{\link{dwnld_rcrd_log}()}, 34 | \code{\link{entrez_fasta_get}()}, 35 | \code{\link{entrez_gb_get}()}, 36 | \code{\link{extract_accession}()}, 37 | \code{\link{extract_by_patterns}()}, 38 | \code{\link{extract_clean_sequence}()}, 39 | \code{\link{extract_definition}()}, 40 | \code{\link{extract_features}()}, 41 | \code{\link{extract_inforecpart}()}, 42 | \code{\link{extract_keywords}()}, 43 | \code{\link{extract_locus}()}, 44 | \code{\link{extract_organism}()}, 45 | \code{\link{extract_seqrecpart}()}, 46 | \code{\link{extract_sequence}()}, 47 | \code{\link{extract_version}()}, 48 | \code{\link{file_download}()}, 49 | \code{\link{filename_log}()}, 50 | \code{\link{flatfile_read}()}, 51 | \code{\link{gb_build}()}, 52 | \code{\link{gb_df_create}()}, 53 | \code{\link{gb_df_generate}()}, 54 | \code{\link{gb_sql_add}()}, 55 | \code{\link{gb_sql_query}()}, 56 | \code{\link{gbrelease_check}()}, 57 | \code{\link{gbrelease_log}()}, 58 | \code{\link{has_data}()}, 59 | \code{\link{identify_downloadable_files}()}, 60 | \code{\link{last_add_get}()}, 61 | \code{\link{last_dwnld_get}()}, 62 | \code{\link{last_entry_get}()}, 63 | \code{\link{latest_genbank_release}()}, 64 | \code{\link{latest_genbank_release_notes}()}, 65 | \code{\link{message_missing}()}, 66 | \code{\link{mock_def}()}, 67 | \code{\link{mock_gb_df_generate}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_get}()}, 81 | \code{\link{slctn_log}()}, 82 | \code{\link{sql_path_get}()}, 83 | \code{\link{stat}()}, 84 | \code{\link{status_class}()}, 85 | \code{\link{testdatadir_get}()} 86 | } 87 | \concept{private} 88 | -------------------------------------------------------------------------------- /man/has_data.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/connection-tools.R 3 | \name{has_data} 4 | \alias{has_data} 5 | \title{Does the connected database have data?} 6 | \usage{ 7 | has_data() 8 | } 9 | \value{ 10 | Logical 11 | } 12 | \description{ 13 | Returns TRUE if a restez SQL database has data. 14 | } 15 | \seealso{ 16 | Other private: 17 | \code{\link{add_rcrd_log}()}, 18 | \code{\link{cat_line}()}, 19 | \code{\link{char}()}, 20 | \code{\link{check_connection}()}, 21 | \code{\link{cleanup}()}, 22 | \code{\link{connected}()}, 23 | \code{\link{connection_get}()}, 24 | \code{\link{db_download_intern}()}, 25 | \code{\link{db_sqlngths_get}()}, 26 | \code{\link{db_sqlngths_log}()}, 27 | \code{\link{dir_size}()}, 28 | \code{\link{dwnld_path_get}()}, 29 | \code{\link{dwnld_rcrd_log}()}, 30 | \code{\link{entrez_fasta_get}()}, 31 | \code{\link{entrez_gb_get}()}, 32 | \code{\link{extract_accession}()}, 33 | \code{\link{extract_by_patterns}()}, 34 | \code{\link{extract_clean_sequence}()}, 35 | \code{\link{extract_definition}()}, 36 | \code{\link{extract_features}()}, 37 | \code{\link{extract_inforecpart}()}, 38 | \code{\link{extract_keywords}()}, 39 | \code{\link{extract_locus}()}, 40 | \code{\link{extract_organism}()}, 41 | \code{\link{extract_seqrecpart}()}, 42 | \code{\link{extract_sequence}()}, 43 | \code{\link{extract_version}()}, 44 | \code{\link{file_download}()}, 45 | \code{\link{filename_log}()}, 46 | \code{\link{flatfile_read}()}, 47 | \code{\link{gb_build}()}, 48 | \code{\link{gb_df_create}()}, 49 | \code{\link{gb_df_generate}()}, 50 | \code{\link{gb_sql_add}()}, 51 | \code{\link{gb_sql_query}()}, 52 | \code{\link{gbrelease_check}()}, 53 | \code{\link{gbrelease_get}()}, 54 | \code{\link{gbrelease_log}()}, 55 | \code{\link{identify_downloadable_files}()}, 56 | \code{\link{last_add_get}()}, 57 | \code{\link{last_dwnld_get}()}, 58 | \code{\link{last_entry_get}()}, 59 | \code{\link{latest_genbank_release}()}, 60 | \code{\link{latest_genbank_release_notes}()}, 61 | \code{\link{message_missing}()}, 62 | \code{\link{mock_def}()}, 63 | \code{\link{mock_gb_df_generate}()}, 64 | \code{\link{mock_org}()}, 65 | \code{\link{mock_rec}()}, 66 | \code{\link{mock_seq}()}, 67 | \code{\link{predict_datasizes}()}, 68 | \code{\link{readme_log}()}, 69 | \code{\link{restez_connect}()}, 70 | \code{\link{restez_disconnect}()}, 71 | \code{\link{restez_path_check}()}, 72 | \code{\link{restez_rl}()}, 73 | \code{\link{search_gz}()}, 74 | \code{\link{seshinfo_log}()}, 75 | \code{\link{setup}()}, 76 | \code{\link{slctn_get}()}, 77 | \code{\link{slctn_log}()}, 78 | \code{\link{sql_path_get}()}, 79 | \code{\link{stat}()}, 80 | \code{\link{status_class}()}, 81 | \code{\link{testdatadir_get}()} 82 | } 83 | \concept{private} 84 | -------------------------------------------------------------------------------- /man/identify_downloadable_files.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/download-tools.R 3 | \name{identify_downloadable_files} 4 | \alias{identify_downloadable_files} 5 | \title{Identify downloadable files} 6 | \usage{ 7 | identify_downloadable_files() 8 | } 9 | \value{ 10 | data.frame 11 | } 12 | \description{ 13 | Searches through the release notes 14 | for a GenBank release to find all listed .seq files. 15 | Returns a data.frame for all .seq files and their 16 | description. 17 | } 18 | \seealso{ 19 | Other private: 20 | \code{\link{add_rcrd_log}()}, 21 | \code{\link{cat_line}()}, 22 | \code{\link{char}()}, 23 | \code{\link{check_connection}()}, 24 | \code{\link{cleanup}()}, 25 | \code{\link{connected}()}, 26 | \code{\link{connection_get}()}, 27 | \code{\link{db_download_intern}()}, 28 | \code{\link{db_sqlngths_get}()}, 29 | \code{\link{db_sqlngths_log}()}, 30 | \code{\link{dir_size}()}, 31 | \code{\link{dwnld_path_get}()}, 32 | \code{\link{dwnld_rcrd_log}()}, 33 | \code{\link{entrez_fasta_get}()}, 34 | \code{\link{entrez_gb_get}()}, 35 | \code{\link{extract_accession}()}, 36 | \code{\link{extract_by_patterns}()}, 37 | \code{\link{extract_clean_sequence}()}, 38 | \code{\link{extract_definition}()}, 39 | \code{\link{extract_features}()}, 40 | \code{\link{extract_inforecpart}()}, 41 | \code{\link{extract_keywords}()}, 42 | \code{\link{extract_locus}()}, 43 | \code{\link{extract_organism}()}, 44 | \code{\link{extract_seqrecpart}()}, 45 | \code{\link{extract_sequence}()}, 46 | \code{\link{extract_version}()}, 47 | \code{\link{file_download}()}, 48 | \code{\link{filename_log}()}, 49 | \code{\link{flatfile_read}()}, 50 | \code{\link{gb_build}()}, 51 | \code{\link{gb_df_create}()}, 52 | \code{\link{gb_df_generate}()}, 53 | \code{\link{gb_sql_add}()}, 54 | \code{\link{gb_sql_query}()}, 55 | \code{\link{gbrelease_check}()}, 56 | \code{\link{gbrelease_get}()}, 57 | \code{\link{gbrelease_log}()}, 58 | \code{\link{has_data}()}, 59 | \code{\link{last_add_get}()}, 60 | \code{\link{last_dwnld_get}()}, 61 | \code{\link{last_entry_get}()}, 62 | \code{\link{latest_genbank_release}()}, 63 | \code{\link{latest_genbank_release_notes}()}, 64 | \code{\link{message_missing}()}, 65 | \code{\link{mock_def}()}, 66 | \code{\link{mock_gb_df_generate}()}, 67 | \code{\link{mock_org}()}, 68 | \code{\link{mock_rec}()}, 69 | \code{\link{mock_seq}()}, 70 | \code{\link{predict_datasizes}()}, 71 | \code{\link{readme_log}()}, 72 | \code{\link{restez_connect}()}, 73 | \code{\link{restez_disconnect}()}, 74 | \code{\link{restez_path_check}()}, 75 | \code{\link{restez_rl}()}, 76 | \code{\link{search_gz}()}, 77 | \code{\link{seshinfo_log}()}, 78 | \code{\link{setup}()}, 79 | \code{\link{slctn_get}()}, 80 | \code{\link{slctn_log}()}, 81 | \code{\link{sql_path_get}()}, 82 | \code{\link{stat}()}, 83 | \code{\link{status_class}()}, 84 | \code{\link{testdatadir_get}()} 85 | } 86 | \concept{private} 87 | -------------------------------------------------------------------------------- /man/is_in_db.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/db-get-tools.R 3 | \name{is_in_db} 4 | \alias{is_in_db} 5 | \title{Is in db} 6 | \usage{ 7 | is_in_db(id, db = "nucleotide") 8 | } 9 | \arguments{ 10 | \item{id}{character, sequence accession ID(s)} 11 | 12 | \item{db}{character, database name} 13 | } 14 | \value{ 15 | named vector of booleans 16 | } 17 | \description{ 18 | Determine whether an id(s) 19 | is/are present in a database. 20 | } 21 | \examples{ 22 | library(restez) 23 | # set the restez path to a temporary dir 24 | restez_path_set(filepath = tempdir()) 25 | # create demo database 26 | demo_db_create(n = 5) 27 | # in the demo, IDs are 'demo_1', 'demo_2' ... 28 | ids <- c('thisisnotanid', 'demo_1', 'demo_2') 29 | (is_in_db(id = ids)) 30 | 31 | 32 | # delete demo after example 33 | db_delete(everything = TRUE) 34 | } 35 | \seealso{ 36 | Other database: 37 | \code{\link{count_db_ids}()}, 38 | \code{\link{db_create}()}, 39 | \code{\link{db_delete}()}, 40 | \code{\link{db_download}()}, 41 | \code{\link{demo_db_create}()}, 42 | \code{\link{list_db_ids}()} 43 | } 44 | \concept{database} 45 | -------------------------------------------------------------------------------- /man/last_add_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{last_add_get} 4 | \alias{last_add_get} 5 | \title{Return date and time of the last added sequence} 6 | \usage{ 7 | last_add_get() 8 | } 9 | \value{ 10 | character 11 | } 12 | \description{ 13 | Return the date and time of the last added sequence as 14 | determined using the 'add_log.tsv'. 15 | } 16 | \details{ 17 | If no file found, returns empty character vector. 18 | } 19 | \seealso{ 20 | Other private: 21 | \code{\link{add_rcrd_log}()}, 22 | \code{\link{cat_line}()}, 23 | \code{\link{char}()}, 24 | \code{\link{check_connection}()}, 25 | \code{\link{cleanup}()}, 26 | \code{\link{connected}()}, 27 | \code{\link{connection_get}()}, 28 | \code{\link{db_download_intern}()}, 29 | \code{\link{db_sqlngths_get}()}, 30 | \code{\link{db_sqlngths_log}()}, 31 | \code{\link{dir_size}()}, 32 | \code{\link{dwnld_path_get}()}, 33 | \code{\link{dwnld_rcrd_log}()}, 34 | \code{\link{entrez_fasta_get}()}, 35 | \code{\link{entrez_gb_get}()}, 36 | \code{\link{extract_accession}()}, 37 | \code{\link{extract_by_patterns}()}, 38 | \code{\link{extract_clean_sequence}()}, 39 | \code{\link{extract_definition}()}, 40 | \code{\link{extract_features}()}, 41 | \code{\link{extract_inforecpart}()}, 42 | \code{\link{extract_keywords}()}, 43 | \code{\link{extract_locus}()}, 44 | \code{\link{extract_organism}()}, 45 | \code{\link{extract_seqrecpart}()}, 46 | \code{\link{extract_sequence}()}, 47 | \code{\link{extract_version}()}, 48 | \code{\link{file_download}()}, 49 | \code{\link{filename_log}()}, 50 | \code{\link{flatfile_read}()}, 51 | \code{\link{gb_build}()}, 52 | \code{\link{gb_df_create}()}, 53 | \code{\link{gb_df_generate}()}, 54 | \code{\link{gb_sql_add}()}, 55 | \code{\link{gb_sql_query}()}, 56 | \code{\link{gbrelease_check}()}, 57 | \code{\link{gbrelease_get}()}, 58 | \code{\link{gbrelease_log}()}, 59 | \code{\link{has_data}()}, 60 | \code{\link{identify_downloadable_files}()}, 61 | \code{\link{last_dwnld_get}()}, 62 | \code{\link{last_entry_get}()}, 63 | \code{\link{latest_genbank_release}()}, 64 | \code{\link{latest_genbank_release_notes}()}, 65 | \code{\link{message_missing}()}, 66 | \code{\link{mock_def}()}, 67 | \code{\link{mock_gb_df_generate}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_get}()}, 81 | \code{\link{slctn_log}()}, 82 | \code{\link{sql_path_get}()}, 83 | \code{\link{stat}()}, 84 | \code{\link{status_class}()}, 85 | \code{\link{testdatadir_get}()} 86 | } 87 | \concept{private} 88 | -------------------------------------------------------------------------------- /man/last_dwnld_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{last_dwnld_get} 4 | \alias{last_dwnld_get} 5 | \title{Return date and time of the last download} 6 | \usage{ 7 | last_dwnld_get() 8 | } 9 | \value{ 10 | character 11 | } 12 | \description{ 13 | Return the date and time of the last download as determined 14 | using the 'download_log.tsv'. 15 | } 16 | \details{ 17 | If no file found, returns empty character vector. 18 | } 19 | \seealso{ 20 | Other private: 21 | \code{\link{add_rcrd_log}()}, 22 | \code{\link{cat_line}()}, 23 | \code{\link{char}()}, 24 | \code{\link{check_connection}()}, 25 | \code{\link{cleanup}()}, 26 | \code{\link{connected}()}, 27 | \code{\link{connection_get}()}, 28 | \code{\link{db_download_intern}()}, 29 | \code{\link{db_sqlngths_get}()}, 30 | \code{\link{db_sqlngths_log}()}, 31 | \code{\link{dir_size}()}, 32 | \code{\link{dwnld_path_get}()}, 33 | \code{\link{dwnld_rcrd_log}()}, 34 | \code{\link{entrez_fasta_get}()}, 35 | \code{\link{entrez_gb_get}()}, 36 | \code{\link{extract_accession}()}, 37 | \code{\link{extract_by_patterns}()}, 38 | \code{\link{extract_clean_sequence}()}, 39 | \code{\link{extract_definition}()}, 40 | \code{\link{extract_features}()}, 41 | \code{\link{extract_inforecpart}()}, 42 | \code{\link{extract_keywords}()}, 43 | \code{\link{extract_locus}()}, 44 | \code{\link{extract_organism}()}, 45 | \code{\link{extract_seqrecpart}()}, 46 | \code{\link{extract_sequence}()}, 47 | \code{\link{extract_version}()}, 48 | \code{\link{file_download}()}, 49 | \code{\link{filename_log}()}, 50 | \code{\link{flatfile_read}()}, 51 | \code{\link{gb_build}()}, 52 | \code{\link{gb_df_create}()}, 53 | \code{\link{gb_df_generate}()}, 54 | \code{\link{gb_sql_add}()}, 55 | \code{\link{gb_sql_query}()}, 56 | \code{\link{gbrelease_check}()}, 57 | \code{\link{gbrelease_get}()}, 58 | \code{\link{gbrelease_log}()}, 59 | \code{\link{has_data}()}, 60 | \code{\link{identify_downloadable_files}()}, 61 | \code{\link{last_add_get}()}, 62 | \code{\link{last_entry_get}()}, 63 | \code{\link{latest_genbank_release}()}, 64 | \code{\link{latest_genbank_release_notes}()}, 65 | \code{\link{message_missing}()}, 66 | \code{\link{mock_def}()}, 67 | \code{\link{mock_gb_df_generate}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_get}()}, 81 | \code{\link{slctn_log}()}, 82 | \code{\link{sql_path_get}()}, 83 | \code{\link{stat}()}, 84 | \code{\link{status_class}()}, 85 | \code{\link{testdatadir_get}()} 86 | } 87 | \concept{private} 88 | -------------------------------------------------------------------------------- /man/last_entry_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{last_entry_get} 4 | \alias{last_entry_get} 5 | \title{Return the last entry} 6 | \usage{ 7 | last_entry_get(fp) 8 | } 9 | \arguments{ 10 | \item{fp}{Filepath, character} 11 | } 12 | \value{ 13 | vector 14 | } 15 | \description{ 16 | Return the last entry from a tab-delimited log file. 17 | } 18 | \seealso{ 19 | Other private: 20 | \code{\link{add_rcrd_log}()}, 21 | \code{\link{cat_line}()}, 22 | \code{\link{char}()}, 23 | \code{\link{check_connection}()}, 24 | \code{\link{cleanup}()}, 25 | \code{\link{connected}()}, 26 | \code{\link{connection_get}()}, 27 | \code{\link{db_download_intern}()}, 28 | \code{\link{db_sqlngths_get}()}, 29 | \code{\link{db_sqlngths_log}()}, 30 | \code{\link{dir_size}()}, 31 | \code{\link{dwnld_path_get}()}, 32 | \code{\link{dwnld_rcrd_log}()}, 33 | \code{\link{entrez_fasta_get}()}, 34 | \code{\link{entrez_gb_get}()}, 35 | \code{\link{extract_accession}()}, 36 | \code{\link{extract_by_patterns}()}, 37 | \code{\link{extract_clean_sequence}()}, 38 | \code{\link{extract_definition}()}, 39 | \code{\link{extract_features}()}, 40 | \code{\link{extract_inforecpart}()}, 41 | \code{\link{extract_keywords}()}, 42 | \code{\link{extract_locus}()}, 43 | \code{\link{extract_organism}()}, 44 | \code{\link{extract_seqrecpart}()}, 45 | \code{\link{extract_sequence}()}, 46 | \code{\link{extract_version}()}, 47 | \code{\link{file_download}()}, 48 | \code{\link{filename_log}()}, 49 | \code{\link{flatfile_read}()}, 50 | \code{\link{gb_build}()}, 51 | \code{\link{gb_df_create}()}, 52 | \code{\link{gb_df_generate}()}, 53 | \code{\link{gb_sql_add}()}, 54 | \code{\link{gb_sql_query}()}, 55 | \code{\link{gbrelease_check}()}, 56 | \code{\link{gbrelease_get}()}, 57 | \code{\link{gbrelease_log}()}, 58 | \code{\link{has_data}()}, 59 | \code{\link{identify_downloadable_files}()}, 60 | \code{\link{last_add_get}()}, 61 | \code{\link{last_dwnld_get}()}, 62 | \code{\link{latest_genbank_release}()}, 63 | \code{\link{latest_genbank_release_notes}()}, 64 | \code{\link{message_missing}()}, 65 | \code{\link{mock_def}()}, 66 | \code{\link{mock_gb_df_generate}()}, 67 | \code{\link{mock_org}()}, 68 | \code{\link{mock_rec}()}, 69 | \code{\link{mock_seq}()}, 70 | \code{\link{predict_datasizes}()}, 71 | \code{\link{readme_log}()}, 72 | \code{\link{restez_connect}()}, 73 | \code{\link{restez_disconnect}()}, 74 | \code{\link{restez_path_check}()}, 75 | \code{\link{restez_rl}()}, 76 | \code{\link{search_gz}()}, 77 | \code{\link{seshinfo_log}()}, 78 | \code{\link{setup}()}, 79 | \code{\link{slctn_get}()}, 80 | \code{\link{slctn_log}()}, 81 | \code{\link{sql_path_get}()}, 82 | \code{\link{stat}()}, 83 | \code{\link{status_class}()}, 84 | \code{\link{testdatadir_get}()} 85 | } 86 | \concept{private} 87 | -------------------------------------------------------------------------------- /man/latest_genbank_release.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/download-tools.R 3 | \name{latest_genbank_release} 4 | \alias{latest_genbank_release} 5 | \title{Retrieve latest GenBank release number} 6 | \usage{ 7 | latest_genbank_release() 8 | } 9 | \value{ 10 | character 11 | } 12 | \description{ 13 | Downloads the latest GenBank release number and returns it. 14 | } 15 | \seealso{ 16 | Other private: 17 | \code{\link{add_rcrd_log}()}, 18 | \code{\link{cat_line}()}, 19 | \code{\link{char}()}, 20 | \code{\link{check_connection}()}, 21 | \code{\link{cleanup}()}, 22 | \code{\link{connected}()}, 23 | \code{\link{connection_get}()}, 24 | \code{\link{db_download_intern}()}, 25 | \code{\link{db_sqlngths_get}()}, 26 | \code{\link{db_sqlngths_log}()}, 27 | \code{\link{dir_size}()}, 28 | \code{\link{dwnld_path_get}()}, 29 | \code{\link{dwnld_rcrd_log}()}, 30 | \code{\link{entrez_fasta_get}()}, 31 | \code{\link{entrez_gb_get}()}, 32 | \code{\link{extract_accession}()}, 33 | \code{\link{extract_by_patterns}()}, 34 | \code{\link{extract_clean_sequence}()}, 35 | \code{\link{extract_definition}()}, 36 | \code{\link{extract_features}()}, 37 | \code{\link{extract_inforecpart}()}, 38 | \code{\link{extract_keywords}()}, 39 | \code{\link{extract_locus}()}, 40 | \code{\link{extract_organism}()}, 41 | \code{\link{extract_seqrecpart}()}, 42 | \code{\link{extract_sequence}()}, 43 | \code{\link{extract_version}()}, 44 | \code{\link{file_download}()}, 45 | \code{\link{filename_log}()}, 46 | \code{\link{flatfile_read}()}, 47 | \code{\link{gb_build}()}, 48 | \code{\link{gb_df_create}()}, 49 | \code{\link{gb_df_generate}()}, 50 | \code{\link{gb_sql_add}()}, 51 | \code{\link{gb_sql_query}()}, 52 | \code{\link{gbrelease_check}()}, 53 | \code{\link{gbrelease_get}()}, 54 | \code{\link{gbrelease_log}()}, 55 | \code{\link{has_data}()}, 56 | \code{\link{identify_downloadable_files}()}, 57 | \code{\link{last_add_get}()}, 58 | \code{\link{last_dwnld_get}()}, 59 | \code{\link{last_entry_get}()}, 60 | \code{\link{latest_genbank_release_notes}()}, 61 | \code{\link{message_missing}()}, 62 | \code{\link{mock_def}()}, 63 | \code{\link{mock_gb_df_generate}()}, 64 | \code{\link{mock_org}()}, 65 | \code{\link{mock_rec}()}, 66 | \code{\link{mock_seq}()}, 67 | \code{\link{predict_datasizes}()}, 68 | \code{\link{readme_log}()}, 69 | \code{\link{restez_connect}()}, 70 | \code{\link{restez_disconnect}()}, 71 | \code{\link{restez_path_check}()}, 72 | \code{\link{restez_rl}()}, 73 | \code{\link{search_gz}()}, 74 | \code{\link{seshinfo_log}()}, 75 | \code{\link{setup}()}, 76 | \code{\link{slctn_get}()}, 77 | \code{\link{slctn_log}()}, 78 | \code{\link{sql_path_get}()}, 79 | \code{\link{stat}()}, 80 | \code{\link{status_class}()}, 81 | \code{\link{testdatadir_get}()} 82 | } 83 | \concept{private} 84 | -------------------------------------------------------------------------------- /man/latest_genbank_release_notes.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/download-tools.R 3 | \name{latest_genbank_release_notes} 4 | \alias{latest_genbank_release_notes} 5 | \title{Download the latest GenBank Release Notes} 6 | \usage{ 7 | latest_genbank_release_notes() 8 | } 9 | \description{ 10 | Downloads the latest GenBank release notes to a user's restez 11 | download path. 12 | } 13 | \seealso{ 14 | Other private: 15 | \code{\link{add_rcrd_log}()}, 16 | \code{\link{cat_line}()}, 17 | \code{\link{char}()}, 18 | \code{\link{check_connection}()}, 19 | \code{\link{cleanup}()}, 20 | \code{\link{connected}()}, 21 | \code{\link{connection_get}()}, 22 | \code{\link{db_download_intern}()}, 23 | \code{\link{db_sqlngths_get}()}, 24 | \code{\link{db_sqlngths_log}()}, 25 | \code{\link{dir_size}()}, 26 | \code{\link{dwnld_path_get}()}, 27 | \code{\link{dwnld_rcrd_log}()}, 28 | \code{\link{entrez_fasta_get}()}, 29 | \code{\link{entrez_gb_get}()}, 30 | \code{\link{extract_accession}()}, 31 | \code{\link{extract_by_patterns}()}, 32 | \code{\link{extract_clean_sequence}()}, 33 | \code{\link{extract_definition}()}, 34 | \code{\link{extract_features}()}, 35 | \code{\link{extract_inforecpart}()}, 36 | \code{\link{extract_keywords}()}, 37 | \code{\link{extract_locus}()}, 38 | \code{\link{extract_organism}()}, 39 | \code{\link{extract_seqrecpart}()}, 40 | \code{\link{extract_sequence}()}, 41 | \code{\link{extract_version}()}, 42 | \code{\link{file_download}()}, 43 | \code{\link{filename_log}()}, 44 | \code{\link{flatfile_read}()}, 45 | \code{\link{gb_build}()}, 46 | \code{\link{gb_df_create}()}, 47 | \code{\link{gb_df_generate}()}, 48 | \code{\link{gb_sql_add}()}, 49 | \code{\link{gb_sql_query}()}, 50 | \code{\link{gbrelease_check}()}, 51 | \code{\link{gbrelease_get}()}, 52 | \code{\link{gbrelease_log}()}, 53 | \code{\link{has_data}()}, 54 | \code{\link{identify_downloadable_files}()}, 55 | \code{\link{last_add_get}()}, 56 | \code{\link{last_dwnld_get}()}, 57 | \code{\link{last_entry_get}()}, 58 | \code{\link{latest_genbank_release}()}, 59 | \code{\link{message_missing}()}, 60 | \code{\link{mock_def}()}, 61 | \code{\link{mock_gb_df_generate}()}, 62 | \code{\link{mock_org}()}, 63 | \code{\link{mock_rec}()}, 64 | \code{\link{mock_seq}()}, 65 | \code{\link{predict_datasizes}()}, 66 | \code{\link{readme_log}()}, 67 | \code{\link{restez_connect}()}, 68 | \code{\link{restez_disconnect}()}, 69 | \code{\link{restez_path_check}()}, 70 | \code{\link{restez_rl}()}, 71 | \code{\link{search_gz}()}, 72 | \code{\link{seshinfo_log}()}, 73 | \code{\link{setup}()}, 74 | \code{\link{slctn_get}()}, 75 | \code{\link{slctn_log}()}, 76 | \code{\link{sql_path_get}()}, 77 | \code{\link{stat}()}, 78 | \code{\link{status_class}()}, 79 | \code{\link{testdatadir_get}()} 80 | } 81 | \concept{private} 82 | -------------------------------------------------------------------------------- /man/list_db_ids.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/db-get-tools.R 3 | \name{list_db_ids} 4 | \alias{list_db_ids} 5 | \title{List database IDs} 6 | \usage{ 7 | list_db_ids(db = "nucleotide", n = 100) 8 | } 9 | \arguments{ 10 | \item{db}{character, database name} 11 | 12 | \item{n}{Maximum number of IDs to return, if NULL returns all} 13 | } 14 | \value{ 15 | vector of characters 16 | } 17 | \description{ 18 | Return a vector of all IDs in 19 | a database. 20 | } 21 | \details{ 22 | Warning: can return very large vectors 23 | for large databases. 24 | } 25 | \examples{ 26 | library(restez) 27 | restez_path_set(filepath = tempdir()) 28 | demo_db_create(n = 5) 29 | # Warning: not recommended for real databases 30 | # with potentially millions of IDs 31 | all_ids <- list_db_ids() 32 | 33 | 34 | # What shall we do with these IDs? 35 | # ... how about make a mock fasta file 36 | seqs <- gb_sequence_get(id = all_ids) 37 | defs <- gb_definition_get(id = all_ids) 38 | # paste together 39 | fasta_seqs <- paste0('>', defs, '\n', seqs) 40 | fasta_file <- paste0(fasta_seqs, collapse = '\n') 41 | cat(fasta_file) 42 | 43 | 44 | # delete after example 45 | db_delete(everything = TRUE) 46 | } 47 | \seealso{ 48 | Other database: 49 | \code{\link{count_db_ids}()}, 50 | \code{\link{db_create}()}, 51 | \code{\link{db_delete}()}, 52 | \code{\link{db_download}()}, 53 | \code{\link{demo_db_create}()}, 54 | \code{\link{is_in_db}()} 55 | } 56 | \concept{database} 57 | -------------------------------------------------------------------------------- /man/message_missing.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/entrez-tools.R 3 | \name{message_missing} 4 | \alias{message_missing} 5 | \title{Produce message of missing IDs} 6 | \usage{ 7 | message_missing(n) 8 | } 9 | \arguments{ 10 | \item{n}{Number of missing IDs} 11 | } 12 | \description{ 13 | Sends message to console stating number of missing IDs. 14 | } 15 | \seealso{ 16 | Other private: 17 | \code{\link{add_rcrd_log}()}, 18 | \code{\link{cat_line}()}, 19 | \code{\link{char}()}, 20 | \code{\link{check_connection}()}, 21 | \code{\link{cleanup}()}, 22 | \code{\link{connected}()}, 23 | \code{\link{connection_get}()}, 24 | \code{\link{db_download_intern}()}, 25 | \code{\link{db_sqlngths_get}()}, 26 | \code{\link{db_sqlngths_log}()}, 27 | \code{\link{dir_size}()}, 28 | \code{\link{dwnld_path_get}()}, 29 | \code{\link{dwnld_rcrd_log}()}, 30 | \code{\link{entrez_fasta_get}()}, 31 | \code{\link{entrez_gb_get}()}, 32 | \code{\link{extract_accession}()}, 33 | \code{\link{extract_by_patterns}()}, 34 | \code{\link{extract_clean_sequence}()}, 35 | \code{\link{extract_definition}()}, 36 | \code{\link{extract_features}()}, 37 | \code{\link{extract_inforecpart}()}, 38 | \code{\link{extract_keywords}()}, 39 | \code{\link{extract_locus}()}, 40 | \code{\link{extract_organism}()}, 41 | \code{\link{extract_seqrecpart}()}, 42 | \code{\link{extract_sequence}()}, 43 | \code{\link{extract_version}()}, 44 | \code{\link{file_download}()}, 45 | \code{\link{filename_log}()}, 46 | \code{\link{flatfile_read}()}, 47 | \code{\link{gb_build}()}, 48 | \code{\link{gb_df_create}()}, 49 | \code{\link{gb_df_generate}()}, 50 | \code{\link{gb_sql_add}()}, 51 | \code{\link{gb_sql_query}()}, 52 | \code{\link{gbrelease_check}()}, 53 | \code{\link{gbrelease_get}()}, 54 | \code{\link{gbrelease_log}()}, 55 | \code{\link{has_data}()}, 56 | \code{\link{identify_downloadable_files}()}, 57 | \code{\link{last_add_get}()}, 58 | \code{\link{last_dwnld_get}()}, 59 | \code{\link{last_entry_get}()}, 60 | \code{\link{latest_genbank_release}()}, 61 | \code{\link{latest_genbank_release_notes}()}, 62 | \code{\link{mock_def}()}, 63 | \code{\link{mock_gb_df_generate}()}, 64 | \code{\link{mock_org}()}, 65 | \code{\link{mock_rec}()}, 66 | \code{\link{mock_seq}()}, 67 | \code{\link{predict_datasizes}()}, 68 | \code{\link{readme_log}()}, 69 | \code{\link{restez_connect}()}, 70 | \code{\link{restez_disconnect}()}, 71 | \code{\link{restez_path_check}()}, 72 | \code{\link{restez_rl}()}, 73 | \code{\link{search_gz}()}, 74 | \code{\link{seshinfo_log}()}, 75 | \code{\link{setup}()}, 76 | \code{\link{slctn_get}()}, 77 | \code{\link{slctn_log}()}, 78 | \code{\link{sql_path_get}()}, 79 | \code{\link{stat}()}, 80 | \code{\link{status_class}()}, 81 | \code{\link{testdatadir_get}()} 82 | } 83 | \concept{private} 84 | -------------------------------------------------------------------------------- /man/mock_def.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/mock-tools.R 3 | \name{mock_def} 4 | \alias{mock_def} 5 | \title{Mock def} 6 | \usage{ 7 | mock_def(i) 8 | } 9 | \arguments{ 10 | \item{i}{integer, iterator} 11 | } 12 | \value{ 13 | character 14 | } 15 | \description{ 16 | Make a mock sequence definition. 17 | Designed to be part of a loop. 18 | } 19 | \seealso{ 20 | Other private: 21 | \code{\link{add_rcrd_log}()}, 22 | \code{\link{cat_line}()}, 23 | \code{\link{char}()}, 24 | \code{\link{check_connection}()}, 25 | \code{\link{cleanup}()}, 26 | \code{\link{connected}()}, 27 | \code{\link{connection_get}()}, 28 | \code{\link{db_download_intern}()}, 29 | \code{\link{db_sqlngths_get}()}, 30 | \code{\link{db_sqlngths_log}()}, 31 | \code{\link{dir_size}()}, 32 | \code{\link{dwnld_path_get}()}, 33 | \code{\link{dwnld_rcrd_log}()}, 34 | \code{\link{entrez_fasta_get}()}, 35 | \code{\link{entrez_gb_get}()}, 36 | \code{\link{extract_accession}()}, 37 | \code{\link{extract_by_patterns}()}, 38 | \code{\link{extract_clean_sequence}()}, 39 | \code{\link{extract_definition}()}, 40 | \code{\link{extract_features}()}, 41 | \code{\link{extract_inforecpart}()}, 42 | \code{\link{extract_keywords}()}, 43 | \code{\link{extract_locus}()}, 44 | \code{\link{extract_organism}()}, 45 | \code{\link{extract_seqrecpart}()}, 46 | \code{\link{extract_sequence}()}, 47 | \code{\link{extract_version}()}, 48 | \code{\link{file_download}()}, 49 | \code{\link{filename_log}()}, 50 | \code{\link{flatfile_read}()}, 51 | \code{\link{gb_build}()}, 52 | \code{\link{gb_df_create}()}, 53 | \code{\link{gb_df_generate}()}, 54 | \code{\link{gb_sql_add}()}, 55 | \code{\link{gb_sql_query}()}, 56 | \code{\link{gbrelease_check}()}, 57 | \code{\link{gbrelease_get}()}, 58 | \code{\link{gbrelease_log}()}, 59 | \code{\link{has_data}()}, 60 | \code{\link{identify_downloadable_files}()}, 61 | \code{\link{last_add_get}()}, 62 | \code{\link{last_dwnld_get}()}, 63 | \code{\link{last_entry_get}()}, 64 | \code{\link{latest_genbank_release}()}, 65 | \code{\link{latest_genbank_release_notes}()}, 66 | \code{\link{message_missing}()}, 67 | \code{\link{mock_gb_df_generate}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_get}()}, 81 | \code{\link{slctn_log}()}, 82 | \code{\link{sql_path_get}()}, 83 | \code{\link{stat}()}, 84 | \code{\link{status_class}()}, 85 | \code{\link{testdatadir_get}()} 86 | } 87 | \concept{private} 88 | -------------------------------------------------------------------------------- /man/mock_gb_df_generate.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/mock-tools.R 3 | \name{mock_gb_df_generate} 4 | \alias{mock_gb_df_generate} 5 | \title{Generate mock GenBank records data.frame} 6 | \usage{ 7 | mock_gb_df_generate(n) 8 | } 9 | \arguments{ 10 | \item{n}{integer, number of entries} 11 | } 12 | \value{ 13 | data.frame 14 | } 15 | \description{ 16 | Make a mock nucleotide data.frame 17 | for entry into a demonstration SQL database. 18 | } 19 | \seealso{ 20 | Other private: 21 | \code{\link{add_rcrd_log}()}, 22 | \code{\link{cat_line}()}, 23 | \code{\link{char}()}, 24 | \code{\link{check_connection}()}, 25 | \code{\link{cleanup}()}, 26 | \code{\link{connected}()}, 27 | \code{\link{connection_get}()}, 28 | \code{\link{db_download_intern}()}, 29 | \code{\link{db_sqlngths_get}()}, 30 | \code{\link{db_sqlngths_log}()}, 31 | \code{\link{dir_size}()}, 32 | \code{\link{dwnld_path_get}()}, 33 | \code{\link{dwnld_rcrd_log}()}, 34 | \code{\link{entrez_fasta_get}()}, 35 | \code{\link{entrez_gb_get}()}, 36 | \code{\link{extract_accession}()}, 37 | \code{\link{extract_by_patterns}()}, 38 | \code{\link{extract_clean_sequence}()}, 39 | \code{\link{extract_definition}()}, 40 | \code{\link{extract_features}()}, 41 | \code{\link{extract_inforecpart}()}, 42 | \code{\link{extract_keywords}()}, 43 | \code{\link{extract_locus}()}, 44 | \code{\link{extract_organism}()}, 45 | \code{\link{extract_seqrecpart}()}, 46 | \code{\link{extract_sequence}()}, 47 | \code{\link{extract_version}()}, 48 | \code{\link{file_download}()}, 49 | \code{\link{filename_log}()}, 50 | \code{\link{flatfile_read}()}, 51 | \code{\link{gb_build}()}, 52 | \code{\link{gb_df_create}()}, 53 | \code{\link{gb_df_generate}()}, 54 | \code{\link{gb_sql_add}()}, 55 | \code{\link{gb_sql_query}()}, 56 | \code{\link{gbrelease_check}()}, 57 | \code{\link{gbrelease_get}()}, 58 | \code{\link{gbrelease_log}()}, 59 | \code{\link{has_data}()}, 60 | \code{\link{identify_downloadable_files}()}, 61 | \code{\link{last_add_get}()}, 62 | \code{\link{last_dwnld_get}()}, 63 | \code{\link{last_entry_get}()}, 64 | \code{\link{latest_genbank_release}()}, 65 | \code{\link{latest_genbank_release_notes}()}, 66 | \code{\link{message_missing}()}, 67 | \code{\link{mock_def}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_get}()}, 81 | \code{\link{slctn_log}()}, 82 | \code{\link{sql_path_get}()}, 83 | \code{\link{stat}()}, 84 | \code{\link{status_class}()}, 85 | \code{\link{testdatadir_get}()} 86 | } 87 | \concept{private} 88 | -------------------------------------------------------------------------------- /man/mock_org.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/mock-tools.R 3 | \name{mock_org} 4 | \alias{mock_org} 5 | \title{Mock org} 6 | \usage{ 7 | mock_org(i) 8 | } 9 | \arguments{ 10 | \item{i}{integer, iterator} 11 | } 12 | \value{ 13 | character 14 | } 15 | \description{ 16 | Make a mock sequence organism. 17 | Designed to be part of a loop. 18 | } 19 | \seealso{ 20 | Other private: 21 | \code{\link{add_rcrd_log}()}, 22 | \code{\link{cat_line}()}, 23 | \code{\link{char}()}, 24 | \code{\link{check_connection}()}, 25 | \code{\link{cleanup}()}, 26 | \code{\link{connected}()}, 27 | \code{\link{connection_get}()}, 28 | \code{\link{db_download_intern}()}, 29 | \code{\link{db_sqlngths_get}()}, 30 | \code{\link{db_sqlngths_log}()}, 31 | \code{\link{dir_size}()}, 32 | \code{\link{dwnld_path_get}()}, 33 | \code{\link{dwnld_rcrd_log}()}, 34 | \code{\link{entrez_fasta_get}()}, 35 | \code{\link{entrez_gb_get}()}, 36 | \code{\link{extract_accession}()}, 37 | \code{\link{extract_by_patterns}()}, 38 | \code{\link{extract_clean_sequence}()}, 39 | \code{\link{extract_definition}()}, 40 | \code{\link{extract_features}()}, 41 | \code{\link{extract_inforecpart}()}, 42 | \code{\link{extract_keywords}()}, 43 | \code{\link{extract_locus}()}, 44 | \code{\link{extract_organism}()}, 45 | \code{\link{extract_seqrecpart}()}, 46 | \code{\link{extract_sequence}()}, 47 | \code{\link{extract_version}()}, 48 | \code{\link{file_download}()}, 49 | \code{\link{filename_log}()}, 50 | \code{\link{flatfile_read}()}, 51 | \code{\link{gb_build}()}, 52 | \code{\link{gb_df_create}()}, 53 | \code{\link{gb_df_generate}()}, 54 | \code{\link{gb_sql_add}()}, 55 | \code{\link{gb_sql_query}()}, 56 | \code{\link{gbrelease_check}()}, 57 | \code{\link{gbrelease_get}()}, 58 | \code{\link{gbrelease_log}()}, 59 | \code{\link{has_data}()}, 60 | \code{\link{identify_downloadable_files}()}, 61 | \code{\link{last_add_get}()}, 62 | \code{\link{last_dwnld_get}()}, 63 | \code{\link{last_entry_get}()}, 64 | \code{\link{latest_genbank_release}()}, 65 | \code{\link{latest_genbank_release_notes}()}, 66 | \code{\link{message_missing}()}, 67 | \code{\link{mock_def}()}, 68 | \code{\link{mock_gb_df_generate}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_get}()}, 81 | \code{\link{slctn_log}()}, 82 | \code{\link{sql_path_get}()}, 83 | \code{\link{stat}()}, 84 | \code{\link{status_class}()}, 85 | \code{\link{testdatadir_get}()} 86 | } 87 | \concept{private} 88 | -------------------------------------------------------------------------------- /man/mock_seq.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/mock-tools.R 3 | \name{mock_seq} 4 | \alias{mock_seq} 5 | \title{Mock seq} 6 | \usage{ 7 | mock_seq(i, sqlngth = 10) 8 | } 9 | \arguments{ 10 | \item{i}{integer, iterator} 11 | 12 | \item{sqlngth}{integer, sequence length} 13 | } 14 | \value{ 15 | character 16 | } 17 | \description{ 18 | Make a mock sequence. Designed to be part of a loop. 19 | } 20 | \seealso{ 21 | Other private: 22 | \code{\link{add_rcrd_log}()}, 23 | \code{\link{cat_line}()}, 24 | \code{\link{char}()}, 25 | \code{\link{check_connection}()}, 26 | \code{\link{cleanup}()}, 27 | \code{\link{connected}()}, 28 | \code{\link{connection_get}()}, 29 | \code{\link{db_download_intern}()}, 30 | \code{\link{db_sqlngths_get}()}, 31 | \code{\link{db_sqlngths_log}()}, 32 | \code{\link{dir_size}()}, 33 | \code{\link{dwnld_path_get}()}, 34 | \code{\link{dwnld_rcrd_log}()}, 35 | \code{\link{entrez_fasta_get}()}, 36 | \code{\link{entrez_gb_get}()}, 37 | \code{\link{extract_accession}()}, 38 | \code{\link{extract_by_patterns}()}, 39 | \code{\link{extract_clean_sequence}()}, 40 | \code{\link{extract_definition}()}, 41 | \code{\link{extract_features}()}, 42 | \code{\link{extract_inforecpart}()}, 43 | \code{\link{extract_keywords}()}, 44 | \code{\link{extract_locus}()}, 45 | \code{\link{extract_organism}()}, 46 | \code{\link{extract_seqrecpart}()}, 47 | \code{\link{extract_sequence}()}, 48 | \code{\link{extract_version}()}, 49 | \code{\link{file_download}()}, 50 | \code{\link{filename_log}()}, 51 | \code{\link{flatfile_read}()}, 52 | \code{\link{gb_build}()}, 53 | \code{\link{gb_df_create}()}, 54 | \code{\link{gb_df_generate}()}, 55 | \code{\link{gb_sql_add}()}, 56 | \code{\link{gb_sql_query}()}, 57 | \code{\link{gbrelease_check}()}, 58 | \code{\link{gbrelease_get}()}, 59 | \code{\link{gbrelease_log}()}, 60 | \code{\link{has_data}()}, 61 | \code{\link{identify_downloadable_files}()}, 62 | \code{\link{last_add_get}()}, 63 | \code{\link{last_dwnld_get}()}, 64 | \code{\link{last_entry_get}()}, 65 | \code{\link{latest_genbank_release}()}, 66 | \code{\link{latest_genbank_release_notes}()}, 67 | \code{\link{message_missing}()}, 68 | \code{\link{mock_def}()}, 69 | \code{\link{mock_gb_df_generate}()}, 70 | \code{\link{mock_org}()}, 71 | \code{\link{mock_rec}()}, 72 | \code{\link{predict_datasizes}()}, 73 | \code{\link{readme_log}()}, 74 | \code{\link{restez_connect}()}, 75 | \code{\link{restez_disconnect}()}, 76 | \code{\link{restez_path_check}()}, 77 | \code{\link{restez_rl}()}, 78 | \code{\link{search_gz}()}, 79 | \code{\link{seshinfo_log}()}, 80 | \code{\link{setup}()}, 81 | \code{\link{slctn_get}()}, 82 | \code{\link{slctn_log}()}, 83 | \code{\link{sql_path_get}()}, 84 | \code{\link{stat}()}, 85 | \code{\link{status_class}()}, 86 | \code{\link{testdatadir_get}()} 87 | } 88 | \concept{private} 89 | -------------------------------------------------------------------------------- /man/ncbi_acc_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/db-setup-tools.R 3 | \name{ncbi_acc_get} 4 | \alias{ncbi_acc_get} 5 | \title{Get accession numbers by querying NCBI GenBank} 6 | \usage{ 7 | ncbi_acc_get(query, strict = TRUE, drop_ver = TRUE) 8 | } 9 | \arguments{ 10 | \item{query}{Character vector of length 1; query string to search GenBank.} 11 | 12 | \item{strict}{Logical vector of length 1; should an error be issued if 13 | the number of unique accessions retrieved does not match the number of hits 14 | from GenBank? Default TRUE.} 15 | 16 | \item{drop_ver}{Logical vector of length 1; should the version part of the 17 | accession number (e.g., '.1' in 'AB001538.1') be dropped? Default TRUE.} 18 | } 19 | \value{ 20 | Character vector; accession numbers resulting from query. 21 | } 22 | \description{ 23 | The query string can be formatted using 24 | \href{https://www.ncbi.nlm.nih.gov/nuccore/advanced}{GenBank advanced query terms} 25 | to obtain accession numbers corresponding to a specific set of criteria. 26 | } 27 | \details{ 28 | Note this queries NCBI GenBank, not the local database generated with restez. 29 | 30 | It can be used either to restrict the accessions used to construct the local 31 | database (\code{acc_filter} argument of \code{\link[=db_create]{db_create()}}) or to specify accessions 32 | to read from the local database (\code{id} argument of \code{\link[=gb_fasta_get]{gb_fasta_get()}} and other 33 | gb_*_get() functions). 34 | } 35 | \examples{ 36 | \dontrun{ 37 | # requires an internet connection 38 | cmin_accs <- ncbi_acc_get("Crepidomanes minutum") 39 | length(cmin_accs) 40 | head(cmin_accs) 41 | } 42 | } 43 | \seealso{ 44 | \code{\link[=db_create]{db_create()}}, \code{\link[=gb_fasta_get]{gb_fasta_get()}} 45 | } 46 | -------------------------------------------------------------------------------- /man/print.status.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/status-tools.R 3 | \name{print.status} 4 | \alias{print.status} 5 | \title{Print method for status class} 6 | \usage{ 7 | \method{print}{status}(x, ...) 8 | } 9 | \arguments{ 10 | \item{x}{Status object} 11 | 12 | \item{...}{Other arguments (not used by this function)} 13 | } 14 | \description{ 15 | Prints to screen the three sections of the status class. Not 16 | meant to be used interactively. 17 | } 18 | -------------------------------------------------------------------------------- /man/readme_log.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{readme_log} 4 | \alias{readme_log} 5 | \title{Create README in restez_path} 6 | \usage{ 7 | readme_log() 8 | } 9 | \description{ 10 | Write notes for the curious sorts who peruse the restez_path. 11 | } 12 | \seealso{ 13 | Other private: 14 | \code{\link{add_rcrd_log}()}, 15 | \code{\link{cat_line}()}, 16 | \code{\link{char}()}, 17 | \code{\link{check_connection}()}, 18 | \code{\link{cleanup}()}, 19 | \code{\link{connected}()}, 20 | \code{\link{connection_get}()}, 21 | \code{\link{db_download_intern}()}, 22 | \code{\link{db_sqlngths_get}()}, 23 | \code{\link{db_sqlngths_log}()}, 24 | \code{\link{dir_size}()}, 25 | \code{\link{dwnld_path_get}()}, 26 | \code{\link{dwnld_rcrd_log}()}, 27 | \code{\link{entrez_fasta_get}()}, 28 | \code{\link{entrez_gb_get}()}, 29 | \code{\link{extract_accession}()}, 30 | \code{\link{extract_by_patterns}()}, 31 | \code{\link{extract_clean_sequence}()}, 32 | \code{\link{extract_definition}()}, 33 | \code{\link{extract_features}()}, 34 | \code{\link{extract_inforecpart}()}, 35 | \code{\link{extract_keywords}()}, 36 | \code{\link{extract_locus}()}, 37 | \code{\link{extract_organism}()}, 38 | \code{\link{extract_seqrecpart}()}, 39 | \code{\link{extract_sequence}()}, 40 | \code{\link{extract_version}()}, 41 | \code{\link{file_download}()}, 42 | \code{\link{filename_log}()}, 43 | \code{\link{flatfile_read}()}, 44 | \code{\link{gb_build}()}, 45 | \code{\link{gb_df_create}()}, 46 | \code{\link{gb_df_generate}()}, 47 | \code{\link{gb_sql_add}()}, 48 | \code{\link{gb_sql_query}()}, 49 | \code{\link{gbrelease_check}()}, 50 | \code{\link{gbrelease_get}()}, 51 | \code{\link{gbrelease_log}()}, 52 | \code{\link{has_data}()}, 53 | \code{\link{identify_downloadable_files}()}, 54 | \code{\link{last_add_get}()}, 55 | \code{\link{last_dwnld_get}()}, 56 | \code{\link{last_entry_get}()}, 57 | \code{\link{latest_genbank_release}()}, 58 | \code{\link{latest_genbank_release_notes}()}, 59 | \code{\link{message_missing}()}, 60 | \code{\link{mock_def}()}, 61 | \code{\link{mock_gb_df_generate}()}, 62 | \code{\link{mock_org}()}, 63 | \code{\link{mock_rec}()}, 64 | \code{\link{mock_seq}()}, 65 | \code{\link{predict_datasizes}()}, 66 | \code{\link{restez_connect}()}, 67 | \code{\link{restez_disconnect}()}, 68 | \code{\link{restez_path_check}()}, 69 | \code{\link{restez_rl}()}, 70 | \code{\link{search_gz}()}, 71 | \code{\link{seshinfo_log}()}, 72 | \code{\link{setup}()}, 73 | \code{\link{slctn_get}()}, 74 | \code{\link{slctn_log}()}, 75 | \code{\link{sql_path_get}()}, 76 | \code{\link{stat}()}, 77 | \code{\link{status_class}()}, 78 | \code{\link{testdatadir_get}()} 79 | } 80 | \concept{private} 81 | -------------------------------------------------------------------------------- /man/record.Rd: -------------------------------------------------------------------------------- 1 | \name{record} 2 | \alias{record} 3 | \docType{data} 4 | \title{ 5 | Example GenBank record 6 | } 7 | \description{ 8 | Example GenBank record in text format for demonstration purposes. 9 | } 10 | \usage{data("record")} 11 | \format{ 12 | A large character object containing record information and DNA sequence. 13 | } 14 | \source{ 15 | \url{https://www.ncbi.nlm.nih.gov/nuccore/AY952423.1} 16 | } 17 | \references{ 18 | GenBank 19 | } 20 | \examples{ 21 | data(record) 22 | cat(record) 23 | } 24 | \keyword{datasets} 25 | -------------------------------------------------------------------------------- /man/restez.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/restez.R 3 | \docType{package} 4 | \name{restez} 5 | \alias{restez-package} 6 | \alias{restez} 7 | \title{restez: Create and Query a Local Copy of GenBank in R} 8 | \description{ 9 | The restez package comes with five families of functions: 10 | setup, database, get, entrez and internal/private. 11 | } 12 | \section{Setup functions}{ 13 | 14 | These functions allow a user to set the filepath for where the GenBank files 15 | should be stored, create connections and verify these settings. 16 | } 17 | 18 | \section{Database functions}{ 19 | 20 | These functions download specific parts of GenBank and create the local 21 | SQL-like database. 22 | } 23 | 24 | \section{GenBank functions}{ 25 | 26 | These functions allow a user to query the local SQL-like database. A 27 | user can use an NCBI accession ID to retrieve sequences or whole GenBank 28 | records. 29 | } 30 | 31 | \section{Entrez functions}{ 32 | 33 | The entrez functions are wrappers to the \verb{entrez_*} functions in the 34 | rentrez package. e.g the restez's entrez_fetch will first try to search the 35 | local database, if it fails it will then call rentrez's 36 | \code{\link[rentrez:entrez_fetch]{rentrez::entrez_fetch()}} with the same arguments. 37 | } 38 | 39 | \section{Private/internal functions}{ 40 | 41 | These functions work behind the scenes to make everything work. If you're 42 | curious you can read their documentation using the form 43 | \code{?restez:::functionname}. 44 | } 45 | 46 | \seealso{ 47 | Useful links: 48 | \itemize{ 49 | \item \url{https://github.com/ropensci/restez} 50 | \item \url{https://docs.ropensci.org/restez/} 51 | \item Report bugs at \url{https://github.com/ropensci/restez/issues} 52 | } 53 | 54 | } 55 | \author{ 56 | \strong{Maintainer}: Joel H. Nitta \email{joelnitta@gmail.com} (\href{https://orcid.org/0000-0003-4719-7472}{ORCID}) 57 | 58 | Authors: 59 | \itemize{ 60 | \item Dom Bennett \email{dominic.john.bennett@gmail.com} (\href{https://orcid.org/0000-0003-2722-1359}{ORCID}) 61 | } 62 | 63 | } 64 | \keyword{internal} 65 | -------------------------------------------------------------------------------- /man/restez_disconnect.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/connection-tools.R 3 | \name{restez_disconnect} 4 | \alias{restez_disconnect} 5 | \title{Disconnect from restez database} 6 | \usage{ 7 | restez_disconnect() 8 | } 9 | \description{ 10 | Safely disconnect from the restez connection 11 | } 12 | \seealso{ 13 | Other private: 14 | \code{\link{add_rcrd_log}()}, 15 | \code{\link{cat_line}()}, 16 | \code{\link{char}()}, 17 | \code{\link{check_connection}()}, 18 | \code{\link{cleanup}()}, 19 | \code{\link{connected}()}, 20 | \code{\link{connection_get}()}, 21 | \code{\link{db_download_intern}()}, 22 | \code{\link{db_sqlngths_get}()}, 23 | \code{\link{db_sqlngths_log}()}, 24 | \code{\link{dir_size}()}, 25 | \code{\link{dwnld_path_get}()}, 26 | \code{\link{dwnld_rcrd_log}()}, 27 | \code{\link{entrez_fasta_get}()}, 28 | \code{\link{entrez_gb_get}()}, 29 | \code{\link{extract_accession}()}, 30 | \code{\link{extract_by_patterns}()}, 31 | \code{\link{extract_clean_sequence}()}, 32 | \code{\link{extract_definition}()}, 33 | \code{\link{extract_features}()}, 34 | \code{\link{extract_inforecpart}()}, 35 | \code{\link{extract_keywords}()}, 36 | \code{\link{extract_locus}()}, 37 | \code{\link{extract_organism}()}, 38 | \code{\link{extract_seqrecpart}()}, 39 | \code{\link{extract_sequence}()}, 40 | \code{\link{extract_version}()}, 41 | \code{\link{file_download}()}, 42 | \code{\link{filename_log}()}, 43 | \code{\link{flatfile_read}()}, 44 | \code{\link{gb_build}()}, 45 | \code{\link{gb_df_create}()}, 46 | \code{\link{gb_df_generate}()}, 47 | \code{\link{gb_sql_add}()}, 48 | \code{\link{gb_sql_query}()}, 49 | \code{\link{gbrelease_check}()}, 50 | \code{\link{gbrelease_get}()}, 51 | \code{\link{gbrelease_log}()}, 52 | \code{\link{has_data}()}, 53 | \code{\link{identify_downloadable_files}()}, 54 | \code{\link{last_add_get}()}, 55 | \code{\link{last_dwnld_get}()}, 56 | \code{\link{last_entry_get}()}, 57 | \code{\link{latest_genbank_release}()}, 58 | \code{\link{latest_genbank_release_notes}()}, 59 | \code{\link{message_missing}()}, 60 | \code{\link{mock_def}()}, 61 | \code{\link{mock_gb_df_generate}()}, 62 | \code{\link{mock_org}()}, 63 | \code{\link{mock_rec}()}, 64 | \code{\link{mock_seq}()}, 65 | \code{\link{predict_datasizes}()}, 66 | \code{\link{readme_log}()}, 67 | \code{\link{restez_connect}()}, 68 | \code{\link{restez_path_check}()}, 69 | \code{\link{restez_rl}()}, 70 | \code{\link{search_gz}()}, 71 | \code{\link{seshinfo_log}()}, 72 | \code{\link{setup}()}, 73 | \code{\link{slctn_get}()}, 74 | \code{\link{slctn_log}()}, 75 | \code{\link{sql_path_get}()}, 76 | \code{\link{stat}()}, 77 | \code{\link{status_class}()}, 78 | \code{\link{testdatadir_get}()} 79 | } 80 | \concept{private} 81 | -------------------------------------------------------------------------------- /man/restez_path_check.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/filepath-tools.R 3 | \name{restez_path_check} 4 | \alias{restez_path_check} 5 | \title{Check restez filepath} 6 | \usage{ 7 | restez_path_check() 8 | } 9 | \description{ 10 | Raises error if restez path does 11 | not exist. 12 | } 13 | \seealso{ 14 | Other private: 15 | \code{\link{add_rcrd_log}()}, 16 | \code{\link{cat_line}()}, 17 | \code{\link{char}()}, 18 | \code{\link{check_connection}()}, 19 | \code{\link{cleanup}()}, 20 | \code{\link{connected}()}, 21 | \code{\link{connection_get}()}, 22 | \code{\link{db_download_intern}()}, 23 | \code{\link{db_sqlngths_get}()}, 24 | \code{\link{db_sqlngths_log}()}, 25 | \code{\link{dir_size}()}, 26 | \code{\link{dwnld_path_get}()}, 27 | \code{\link{dwnld_rcrd_log}()}, 28 | \code{\link{entrez_fasta_get}()}, 29 | \code{\link{entrez_gb_get}()}, 30 | \code{\link{extract_accession}()}, 31 | \code{\link{extract_by_patterns}()}, 32 | \code{\link{extract_clean_sequence}()}, 33 | \code{\link{extract_definition}()}, 34 | \code{\link{extract_features}()}, 35 | \code{\link{extract_inforecpart}()}, 36 | \code{\link{extract_keywords}()}, 37 | \code{\link{extract_locus}()}, 38 | \code{\link{extract_organism}()}, 39 | \code{\link{extract_seqrecpart}()}, 40 | \code{\link{extract_sequence}()}, 41 | \code{\link{extract_version}()}, 42 | \code{\link{file_download}()}, 43 | \code{\link{filename_log}()}, 44 | \code{\link{flatfile_read}()}, 45 | \code{\link{gb_build}()}, 46 | \code{\link{gb_df_create}()}, 47 | \code{\link{gb_df_generate}()}, 48 | \code{\link{gb_sql_add}()}, 49 | \code{\link{gb_sql_query}()}, 50 | \code{\link{gbrelease_check}()}, 51 | \code{\link{gbrelease_get}()}, 52 | \code{\link{gbrelease_log}()}, 53 | \code{\link{has_data}()}, 54 | \code{\link{identify_downloadable_files}()}, 55 | \code{\link{last_add_get}()}, 56 | \code{\link{last_dwnld_get}()}, 57 | \code{\link{last_entry_get}()}, 58 | \code{\link{latest_genbank_release}()}, 59 | \code{\link{latest_genbank_release_notes}()}, 60 | \code{\link{message_missing}()}, 61 | \code{\link{mock_def}()}, 62 | \code{\link{mock_gb_df_generate}()}, 63 | \code{\link{mock_org}()}, 64 | \code{\link{mock_rec}()}, 65 | \code{\link{mock_seq}()}, 66 | \code{\link{predict_datasizes}()}, 67 | \code{\link{readme_log}()}, 68 | \code{\link{restez_connect}()}, 69 | \code{\link{restez_disconnect}()}, 70 | \code{\link{restez_rl}()}, 71 | \code{\link{search_gz}()}, 72 | \code{\link{seshinfo_log}()}, 73 | \code{\link{setup}()}, 74 | \code{\link{slctn_get}()}, 75 | \code{\link{slctn_log}()}, 76 | \code{\link{sql_path_get}()}, 77 | \code{\link{stat}()}, 78 | \code{\link{status_class}()}, 79 | \code{\link{testdatadir_get}()} 80 | } 81 | \concept{private} 82 | -------------------------------------------------------------------------------- /man/restez_path_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/filepath-tools.R 3 | \name{restez_path_get} 4 | \alias{restez_path_get} 5 | \title{Get restez path} 6 | \usage{ 7 | restez_path_get() 8 | } 9 | \value{ 10 | character 11 | } 12 | \description{ 13 | Return filepath to where the restez 14 | database is stored. 15 | } 16 | \examples{ 17 | library(restez) 18 | # set a restez path with a tempdir 19 | restez_path_set(filepath = tempdir()) 20 | # check what the set path is 21 | (restez_path_get()) 22 | } 23 | \seealso{ 24 | Other setup: 25 | \code{\link{restez_path_set}()}, 26 | \code{\link{restez_path_unset}()}, 27 | \code{\link{restez_ready}()}, 28 | \code{\link{restez_status}()} 29 | } 30 | \concept{setup} 31 | -------------------------------------------------------------------------------- /man/restez_path_set.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/filepath-tools.R 3 | \name{restez_path_set} 4 | \alias{restez_path_set} 5 | \title{Set restez path} 6 | \usage{ 7 | restez_path_set(filepath) 8 | } 9 | \arguments{ 10 | \item{filepath}{character, valid filepath to the folder where the 11 | database should be stored.} 12 | } 13 | \description{ 14 | Specify the filepath for the local GenBank database. 15 | } 16 | \details{ 17 | Adds 'restez_path' to options(). In this path 18 | the folder 'restez' will be created and all downloaded and 19 | database files will be stored there. 20 | } 21 | \examples{ 22 | \dontrun{ 23 | library(restez) 24 | restez_path_set(filepath = 'path/to/where/you/want/files/to/download') 25 | } 26 | } 27 | \seealso{ 28 | Other setup: 29 | \code{\link{restez_path_get}()}, 30 | \code{\link{restez_path_unset}()}, 31 | \code{\link{restez_ready}()}, 32 | \code{\link{restez_status}()} 33 | } 34 | \concept{setup} 35 | -------------------------------------------------------------------------------- /man/restez_path_unset.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/filepath-tools.R 3 | \name{restez_path_unset} 4 | \alias{restez_path_unset} 5 | \title{Unset restez path} 6 | \usage{ 7 | restez_path_unset() 8 | } 9 | \description{ 10 | Set the restez path to NULL 11 | } 12 | \seealso{ 13 | Other setup: 14 | \code{\link{restez_path_get}()}, 15 | \code{\link{restez_path_set}()}, 16 | \code{\link{restez_ready}()}, 17 | \code{\link{restez_status}()} 18 | } 19 | \concept{setup} 20 | -------------------------------------------------------------------------------- /man/restez_ready.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/connection-tools.R 3 | \name{restez_ready} 4 | \alias{restez_ready} 5 | \title{Is restez ready?} 6 | \usage{ 7 | restez_ready() 8 | } 9 | \value{ 10 | Logical 11 | } 12 | \description{ 13 | Returns TRUE if a restez SQL database is available. 14 | Use restez_status() for more information. 15 | } 16 | \examples{ 17 | library(restez) 18 | fp <- tempdir() 19 | restez_path_set(filepath = fp) 20 | demo_db_create(n = 5) 21 | (restez_ready()) 22 | db_delete(everything = TRUE) 23 | (restez_ready()) 24 | } 25 | \seealso{ 26 | Other setup: 27 | \code{\link{restez_path_get}()}, 28 | \code{\link{restez_path_set}()}, 29 | \code{\link{restez_path_unset}()}, 30 | \code{\link{restez_status}()} 31 | } 32 | \concept{setup} 33 | -------------------------------------------------------------------------------- /man/restez_rl.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/wrappers.R 3 | \name{restez_rl} 4 | \alias{restez_rl} 5 | \title{Restez readline} 6 | \usage{ 7 | restez_rl(prompt) 8 | } 9 | \arguments{ 10 | \item{prompt}{character, display text} 11 | } 12 | \value{ 13 | character 14 | } 15 | \description{ 16 | Wrapper for base readline. 17 | } 18 | \seealso{ 19 | Other private: 20 | \code{\link{add_rcrd_log}()}, 21 | \code{\link{cat_line}()}, 22 | \code{\link{char}()}, 23 | \code{\link{check_connection}()}, 24 | \code{\link{cleanup}()}, 25 | \code{\link{connected}()}, 26 | \code{\link{connection_get}()}, 27 | \code{\link{db_download_intern}()}, 28 | \code{\link{db_sqlngths_get}()}, 29 | \code{\link{db_sqlngths_log}()}, 30 | \code{\link{dir_size}()}, 31 | \code{\link{dwnld_path_get}()}, 32 | \code{\link{dwnld_rcrd_log}()}, 33 | \code{\link{entrez_fasta_get}()}, 34 | \code{\link{entrez_gb_get}()}, 35 | \code{\link{extract_accession}()}, 36 | \code{\link{extract_by_patterns}()}, 37 | \code{\link{extract_clean_sequence}()}, 38 | \code{\link{extract_definition}()}, 39 | \code{\link{extract_features}()}, 40 | \code{\link{extract_inforecpart}()}, 41 | \code{\link{extract_keywords}()}, 42 | \code{\link{extract_locus}()}, 43 | \code{\link{extract_organism}()}, 44 | \code{\link{extract_seqrecpart}()}, 45 | \code{\link{extract_sequence}()}, 46 | \code{\link{extract_version}()}, 47 | \code{\link{file_download}()}, 48 | \code{\link{filename_log}()}, 49 | \code{\link{flatfile_read}()}, 50 | \code{\link{gb_build}()}, 51 | \code{\link{gb_df_create}()}, 52 | \code{\link{gb_df_generate}()}, 53 | \code{\link{gb_sql_add}()}, 54 | \code{\link{gb_sql_query}()}, 55 | \code{\link{gbrelease_check}()}, 56 | \code{\link{gbrelease_get}()}, 57 | \code{\link{gbrelease_log}()}, 58 | \code{\link{has_data}()}, 59 | \code{\link{identify_downloadable_files}()}, 60 | \code{\link{last_add_get}()}, 61 | \code{\link{last_dwnld_get}()}, 62 | \code{\link{last_entry_get}()}, 63 | \code{\link{latest_genbank_release}()}, 64 | \code{\link{latest_genbank_release_notes}()}, 65 | \code{\link{message_missing}()}, 66 | \code{\link{mock_def}()}, 67 | \code{\link{mock_gb_df_generate}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{search_gz}()}, 77 | \code{\link{seshinfo_log}()}, 78 | \code{\link{setup}()}, 79 | \code{\link{slctn_get}()}, 80 | \code{\link{slctn_log}()}, 81 | \code{\link{sql_path_get}()}, 82 | \code{\link{stat}()}, 83 | \code{\link{status_class}()}, 84 | \code{\link{testdatadir_get}()} 85 | } 86 | \concept{private} 87 | -------------------------------------------------------------------------------- /man/restez_status.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/status-tools.R 3 | \name{restez_status} 4 | \alias{restez_status} 5 | \title{Check restez status} 6 | \usage{ 7 | restez_status(gb_check = FALSE) 8 | } 9 | \arguments{ 10 | \item{gb_check}{Check whether last download was from latest GenBank release? 11 | Default FALSE.} 12 | } 13 | \value{ 14 | Status class 15 | } 16 | \description{ 17 | Report to console current setup status of restez. 18 | } 19 | \details{ 20 | Set gb_check=TRUE to see if your downloads are up-to-date. 21 | } 22 | \examples{ 23 | library(restez) 24 | fp <- tempdir() 25 | restez_path_set(filepath = fp) 26 | demo_db_create(n = 5) 27 | restez_status() 28 | db_delete(everything = TRUE) 29 | # Errors: 30 | # restez_status() 31 | } 32 | \seealso{ 33 | Other setup: 34 | \code{\link{restez_path_get}()}, 35 | \code{\link{restez_path_set}()}, 36 | \code{\link{restez_path_unset}()}, 37 | \code{\link{restez_ready}()} 38 | } 39 | \concept{setup} 40 | -------------------------------------------------------------------------------- /man/seshinfo_log.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{seshinfo_log} 4 | \alias{seshinfo_log} 5 | \title{Log the system session information in restez path} 6 | \usage{ 7 | seshinfo_log() 8 | } 9 | \description{ 10 | Records the session and system information to file. 11 | } 12 | \seealso{ 13 | Other private: 14 | \code{\link{add_rcrd_log}()}, 15 | \code{\link{cat_line}()}, 16 | \code{\link{char}()}, 17 | \code{\link{check_connection}()}, 18 | \code{\link{cleanup}()}, 19 | \code{\link{connected}()}, 20 | \code{\link{connection_get}()}, 21 | \code{\link{db_download_intern}()}, 22 | \code{\link{db_sqlngths_get}()}, 23 | \code{\link{db_sqlngths_log}()}, 24 | \code{\link{dir_size}()}, 25 | \code{\link{dwnld_path_get}()}, 26 | \code{\link{dwnld_rcrd_log}()}, 27 | \code{\link{entrez_fasta_get}()}, 28 | \code{\link{entrez_gb_get}()}, 29 | \code{\link{extract_accession}()}, 30 | \code{\link{extract_by_patterns}()}, 31 | \code{\link{extract_clean_sequence}()}, 32 | \code{\link{extract_definition}()}, 33 | \code{\link{extract_features}()}, 34 | \code{\link{extract_inforecpart}()}, 35 | \code{\link{extract_keywords}()}, 36 | \code{\link{extract_locus}()}, 37 | \code{\link{extract_organism}()}, 38 | \code{\link{extract_seqrecpart}()}, 39 | \code{\link{extract_sequence}()}, 40 | \code{\link{extract_version}()}, 41 | \code{\link{file_download}()}, 42 | \code{\link{filename_log}()}, 43 | \code{\link{flatfile_read}()}, 44 | \code{\link{gb_build}()}, 45 | \code{\link{gb_df_create}()}, 46 | \code{\link{gb_df_generate}()}, 47 | \code{\link{gb_sql_add}()}, 48 | \code{\link{gb_sql_query}()}, 49 | \code{\link{gbrelease_check}()}, 50 | \code{\link{gbrelease_get}()}, 51 | \code{\link{gbrelease_log}()}, 52 | \code{\link{has_data}()}, 53 | \code{\link{identify_downloadable_files}()}, 54 | \code{\link{last_add_get}()}, 55 | \code{\link{last_dwnld_get}()}, 56 | \code{\link{last_entry_get}()}, 57 | \code{\link{latest_genbank_release}()}, 58 | \code{\link{latest_genbank_release_notes}()}, 59 | \code{\link{message_missing}()}, 60 | \code{\link{mock_def}()}, 61 | \code{\link{mock_gb_df_generate}()}, 62 | \code{\link{mock_org}()}, 63 | \code{\link{mock_rec}()}, 64 | \code{\link{mock_seq}()}, 65 | \code{\link{predict_datasizes}()}, 66 | \code{\link{readme_log}()}, 67 | \code{\link{restez_connect}()}, 68 | \code{\link{restez_disconnect}()}, 69 | \code{\link{restez_path_check}()}, 70 | \code{\link{restez_rl}()}, 71 | \code{\link{search_gz}()}, 72 | \code{\link{setup}()}, 73 | \code{\link{slctn_get}()}, 74 | \code{\link{slctn_log}()}, 75 | \code{\link{sql_path_get}()}, 76 | \code{\link{stat}()}, 77 | \code{\link{status_class}()}, 78 | \code{\link{testdatadir_get}()} 79 | } 80 | \concept{private} 81 | -------------------------------------------------------------------------------- /man/setup.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/test-tools.R 3 | \name{setup} 4 | \alias{setup} 5 | \title{Set up test common test data} 6 | \usage{ 7 | setup() 8 | } 9 | \description{ 10 | Creates temporary test folders. 11 | } 12 | \seealso{ 13 | Other private: 14 | \code{\link{add_rcrd_log}()}, 15 | \code{\link{cat_line}()}, 16 | \code{\link{char}()}, 17 | \code{\link{check_connection}()}, 18 | \code{\link{cleanup}()}, 19 | \code{\link{connected}()}, 20 | \code{\link{connection_get}()}, 21 | \code{\link{db_download_intern}()}, 22 | \code{\link{db_sqlngths_get}()}, 23 | \code{\link{db_sqlngths_log}()}, 24 | \code{\link{dir_size}()}, 25 | \code{\link{dwnld_path_get}()}, 26 | \code{\link{dwnld_rcrd_log}()}, 27 | \code{\link{entrez_fasta_get}()}, 28 | \code{\link{entrez_gb_get}()}, 29 | \code{\link{extract_accession}()}, 30 | \code{\link{extract_by_patterns}()}, 31 | \code{\link{extract_clean_sequence}()}, 32 | \code{\link{extract_definition}()}, 33 | \code{\link{extract_features}()}, 34 | \code{\link{extract_inforecpart}()}, 35 | \code{\link{extract_keywords}()}, 36 | \code{\link{extract_locus}()}, 37 | \code{\link{extract_organism}()}, 38 | \code{\link{extract_seqrecpart}()}, 39 | \code{\link{extract_sequence}()}, 40 | \code{\link{extract_version}()}, 41 | \code{\link{file_download}()}, 42 | \code{\link{filename_log}()}, 43 | \code{\link{flatfile_read}()}, 44 | \code{\link{gb_build}()}, 45 | \code{\link{gb_df_create}()}, 46 | \code{\link{gb_df_generate}()}, 47 | \code{\link{gb_sql_add}()}, 48 | \code{\link{gb_sql_query}()}, 49 | \code{\link{gbrelease_check}()}, 50 | \code{\link{gbrelease_get}()}, 51 | \code{\link{gbrelease_log}()}, 52 | \code{\link{has_data}()}, 53 | \code{\link{identify_downloadable_files}()}, 54 | \code{\link{last_add_get}()}, 55 | \code{\link{last_dwnld_get}()}, 56 | \code{\link{last_entry_get}()}, 57 | \code{\link{latest_genbank_release}()}, 58 | \code{\link{latest_genbank_release_notes}()}, 59 | \code{\link{message_missing}()}, 60 | \code{\link{mock_def}()}, 61 | \code{\link{mock_gb_df_generate}()}, 62 | \code{\link{mock_org}()}, 63 | \code{\link{mock_rec}()}, 64 | \code{\link{mock_seq}()}, 65 | \code{\link{predict_datasizes}()}, 66 | \code{\link{readme_log}()}, 67 | \code{\link{restez_connect}()}, 68 | \code{\link{restez_disconnect}()}, 69 | \code{\link{restez_path_check}()}, 70 | \code{\link{restez_rl}()}, 71 | \code{\link{search_gz}()}, 72 | \code{\link{seshinfo_log}()}, 73 | \code{\link{slctn_get}()}, 74 | \code{\link{slctn_log}()}, 75 | \code{\link{sql_path_get}()}, 76 | \code{\link{stat}()}, 77 | \code{\link{status_class}()}, 78 | \code{\link{testdatadir_get}()} 79 | } 80 | \concept{private} 81 | -------------------------------------------------------------------------------- /man/slctn_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/log-tools.R 3 | \name{slctn_get} 4 | \alias{slctn_get} 5 | \title{Retrieve GenBank selections made by user} 6 | \usage{ 7 | slctn_get() 8 | } 9 | \value{ 10 | character vector 11 | } 12 | \description{ 13 | Returns the selections made by the user. 14 | } 15 | \details{ 16 | If no file found, returns empty character vector. 17 | } 18 | \seealso{ 19 | Other private: 20 | \code{\link{add_rcrd_log}()}, 21 | \code{\link{cat_line}()}, 22 | \code{\link{char}()}, 23 | \code{\link{check_connection}()}, 24 | \code{\link{cleanup}()}, 25 | \code{\link{connected}()}, 26 | \code{\link{connection_get}()}, 27 | \code{\link{db_download_intern}()}, 28 | \code{\link{db_sqlngths_get}()}, 29 | \code{\link{db_sqlngths_log}()}, 30 | \code{\link{dir_size}()}, 31 | \code{\link{dwnld_path_get}()}, 32 | \code{\link{dwnld_rcrd_log}()}, 33 | \code{\link{entrez_fasta_get}()}, 34 | \code{\link{entrez_gb_get}()}, 35 | \code{\link{extract_accession}()}, 36 | \code{\link{extract_by_patterns}()}, 37 | \code{\link{extract_clean_sequence}()}, 38 | \code{\link{extract_definition}()}, 39 | \code{\link{extract_features}()}, 40 | \code{\link{extract_inforecpart}()}, 41 | \code{\link{extract_keywords}()}, 42 | \code{\link{extract_locus}()}, 43 | \code{\link{extract_organism}()}, 44 | \code{\link{extract_seqrecpart}()}, 45 | \code{\link{extract_sequence}()}, 46 | \code{\link{extract_version}()}, 47 | \code{\link{file_download}()}, 48 | \code{\link{filename_log}()}, 49 | \code{\link{flatfile_read}()}, 50 | \code{\link{gb_build}()}, 51 | \code{\link{gb_df_create}()}, 52 | \code{\link{gb_df_generate}()}, 53 | \code{\link{gb_sql_add}()}, 54 | \code{\link{gb_sql_query}()}, 55 | \code{\link{gbrelease_check}()}, 56 | \code{\link{gbrelease_get}()}, 57 | \code{\link{gbrelease_log}()}, 58 | \code{\link{has_data}()}, 59 | \code{\link{identify_downloadable_files}()}, 60 | \code{\link{last_add_get}()}, 61 | \code{\link{last_dwnld_get}()}, 62 | \code{\link{last_entry_get}()}, 63 | \code{\link{latest_genbank_release}()}, 64 | \code{\link{latest_genbank_release_notes}()}, 65 | \code{\link{message_missing}()}, 66 | \code{\link{mock_def}()}, 67 | \code{\link{mock_gb_df_generate}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_log}()}, 81 | \code{\link{sql_path_get}()}, 82 | \code{\link{stat}()}, 83 | \code{\link{status_class}()}, 84 | \code{\link{testdatadir_get}()} 85 | } 86 | \concept{private} 87 | -------------------------------------------------------------------------------- /man/sql_path_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/filepath-tools.R 3 | \name{sql_path_get} 4 | \alias{sql_path_get} 5 | \title{Get SQL path} 6 | \usage{ 7 | sql_path_get() 8 | } 9 | \value{ 10 | character 11 | } 12 | \description{ 13 | Return path to where SQL database is stored. 14 | } 15 | \seealso{ 16 | Other private: 17 | \code{\link{add_rcrd_log}()}, 18 | \code{\link{cat_line}()}, 19 | \code{\link{char}()}, 20 | \code{\link{check_connection}()}, 21 | \code{\link{cleanup}()}, 22 | \code{\link{connected}()}, 23 | \code{\link{connection_get}()}, 24 | \code{\link{db_download_intern}()}, 25 | \code{\link{db_sqlngths_get}()}, 26 | \code{\link{db_sqlngths_log}()}, 27 | \code{\link{dir_size}()}, 28 | \code{\link{dwnld_path_get}()}, 29 | \code{\link{dwnld_rcrd_log}()}, 30 | \code{\link{entrez_fasta_get}()}, 31 | \code{\link{entrez_gb_get}()}, 32 | \code{\link{extract_accession}()}, 33 | \code{\link{extract_by_patterns}()}, 34 | \code{\link{extract_clean_sequence}()}, 35 | \code{\link{extract_definition}()}, 36 | \code{\link{extract_features}()}, 37 | \code{\link{extract_inforecpart}()}, 38 | \code{\link{extract_keywords}()}, 39 | \code{\link{extract_locus}()}, 40 | \code{\link{extract_organism}()}, 41 | \code{\link{extract_seqrecpart}()}, 42 | \code{\link{extract_sequence}()}, 43 | \code{\link{extract_version}()}, 44 | \code{\link{file_download}()}, 45 | \code{\link{filename_log}()}, 46 | \code{\link{flatfile_read}()}, 47 | \code{\link{gb_build}()}, 48 | \code{\link{gb_df_create}()}, 49 | \code{\link{gb_df_generate}()}, 50 | \code{\link{gb_sql_add}()}, 51 | \code{\link{gb_sql_query}()}, 52 | \code{\link{gbrelease_check}()}, 53 | \code{\link{gbrelease_get}()}, 54 | \code{\link{gbrelease_log}()}, 55 | \code{\link{has_data}()}, 56 | \code{\link{identify_downloadable_files}()}, 57 | \code{\link{last_add_get}()}, 58 | \code{\link{last_dwnld_get}()}, 59 | \code{\link{last_entry_get}()}, 60 | \code{\link{latest_genbank_release}()}, 61 | \code{\link{latest_genbank_release_notes}()}, 62 | \code{\link{message_missing}()}, 63 | \code{\link{mock_def}()}, 64 | \code{\link{mock_gb_df_generate}()}, 65 | \code{\link{mock_org}()}, 66 | \code{\link{mock_rec}()}, 67 | \code{\link{mock_seq}()}, 68 | \code{\link{predict_datasizes}()}, 69 | \code{\link{readme_log}()}, 70 | \code{\link{restez_connect}()}, 71 | \code{\link{restez_disconnect}()}, 72 | \code{\link{restez_path_check}()}, 73 | \code{\link{restez_rl}()}, 74 | \code{\link{search_gz}()}, 75 | \code{\link{seshinfo_log}()}, 76 | \code{\link{setup}()}, 77 | \code{\link{slctn_get}()}, 78 | \code{\link{slctn_log}()}, 79 | \code{\link{stat}()}, 80 | \code{\link{status_class}()}, 81 | \code{\link{testdatadir_get}()} 82 | } 83 | \concept{private} 84 | -------------------------------------------------------------------------------- /man/stat.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/print-tools.R 3 | \name{stat} 4 | \alias{stat} 5 | \title{Print blue} 6 | \usage{ 7 | stat(...) 8 | } 9 | \arguments{ 10 | \item{...}{Any number of text arguments to print, character} 11 | } 12 | \value{ 13 | coloured character encoding, character 14 | } 15 | \description{ 16 | Print to console blue text to indicate a number/statistic. 17 | } 18 | \seealso{ 19 | Other private: 20 | \code{\link{add_rcrd_log}()}, 21 | \code{\link{cat_line}()}, 22 | \code{\link{char}()}, 23 | \code{\link{check_connection}()}, 24 | \code{\link{cleanup}()}, 25 | \code{\link{connected}()}, 26 | \code{\link{connection_get}()}, 27 | \code{\link{db_download_intern}()}, 28 | \code{\link{db_sqlngths_get}()}, 29 | \code{\link{db_sqlngths_log}()}, 30 | \code{\link{dir_size}()}, 31 | \code{\link{dwnld_path_get}()}, 32 | \code{\link{dwnld_rcrd_log}()}, 33 | \code{\link{entrez_fasta_get}()}, 34 | \code{\link{entrez_gb_get}()}, 35 | \code{\link{extract_accession}()}, 36 | \code{\link{extract_by_patterns}()}, 37 | \code{\link{extract_clean_sequence}()}, 38 | \code{\link{extract_definition}()}, 39 | \code{\link{extract_features}()}, 40 | \code{\link{extract_inforecpart}()}, 41 | \code{\link{extract_keywords}()}, 42 | \code{\link{extract_locus}()}, 43 | \code{\link{extract_organism}()}, 44 | \code{\link{extract_seqrecpart}()}, 45 | \code{\link{extract_sequence}()}, 46 | \code{\link{extract_version}()}, 47 | \code{\link{file_download}()}, 48 | \code{\link{filename_log}()}, 49 | \code{\link{flatfile_read}()}, 50 | \code{\link{gb_build}()}, 51 | \code{\link{gb_df_create}()}, 52 | \code{\link{gb_df_generate}()}, 53 | \code{\link{gb_sql_add}()}, 54 | \code{\link{gb_sql_query}()}, 55 | \code{\link{gbrelease_check}()}, 56 | \code{\link{gbrelease_get}()}, 57 | \code{\link{gbrelease_log}()}, 58 | \code{\link{has_data}()}, 59 | \code{\link{identify_downloadable_files}()}, 60 | \code{\link{last_add_get}()}, 61 | \code{\link{last_dwnld_get}()}, 62 | \code{\link{last_entry_get}()}, 63 | \code{\link{latest_genbank_release}()}, 64 | \code{\link{latest_genbank_release_notes}()}, 65 | \code{\link{message_missing}()}, 66 | \code{\link{mock_def}()}, 67 | \code{\link{mock_gb_df_generate}()}, 68 | \code{\link{mock_org}()}, 69 | \code{\link{mock_rec}()}, 70 | \code{\link{mock_seq}()}, 71 | \code{\link{predict_datasizes}()}, 72 | \code{\link{readme_log}()}, 73 | \code{\link{restez_connect}()}, 74 | \code{\link{restez_disconnect}()}, 75 | \code{\link{restez_path_check}()}, 76 | \code{\link{restez_rl}()}, 77 | \code{\link{search_gz}()}, 78 | \code{\link{seshinfo_log}()}, 79 | \code{\link{setup}()}, 80 | \code{\link{slctn_get}()}, 81 | \code{\link{slctn_log}()}, 82 | \code{\link{sql_path_get}()}, 83 | \code{\link{status_class}()}, 84 | \code{\link{testdatadir_get}()} 85 | } 86 | \concept{private} 87 | -------------------------------------------------------------------------------- /man/status_class.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/status-tools.R 3 | \name{status_class} 4 | \alias{status_class} 5 | \title{Generate a list class for storing status information} 6 | \usage{ 7 | status_class() 8 | } 9 | \value{ 10 | Status class 11 | } 12 | \description{ 13 | Creates a three-part list for holding information on the 14 | status of the restez file path. 15 | } 16 | \seealso{ 17 | Other private: 18 | \code{\link{add_rcrd_log}()}, 19 | \code{\link{cat_line}()}, 20 | \code{\link{char}()}, 21 | \code{\link{check_connection}()}, 22 | \code{\link{cleanup}()}, 23 | \code{\link{connected}()}, 24 | \code{\link{connection_get}()}, 25 | \code{\link{db_download_intern}()}, 26 | \code{\link{db_sqlngths_get}()}, 27 | \code{\link{db_sqlngths_log}()}, 28 | \code{\link{dir_size}()}, 29 | \code{\link{dwnld_path_get}()}, 30 | \code{\link{dwnld_rcrd_log}()}, 31 | \code{\link{entrez_fasta_get}()}, 32 | \code{\link{entrez_gb_get}()}, 33 | \code{\link{extract_accession}()}, 34 | \code{\link{extract_by_patterns}()}, 35 | \code{\link{extract_clean_sequence}()}, 36 | \code{\link{extract_definition}()}, 37 | \code{\link{extract_features}()}, 38 | \code{\link{extract_inforecpart}()}, 39 | \code{\link{extract_keywords}()}, 40 | \code{\link{extract_locus}()}, 41 | \code{\link{extract_organism}()}, 42 | \code{\link{extract_seqrecpart}()}, 43 | \code{\link{extract_sequence}()}, 44 | \code{\link{extract_version}()}, 45 | \code{\link{file_download}()}, 46 | \code{\link{filename_log}()}, 47 | \code{\link{flatfile_read}()}, 48 | \code{\link{gb_build}()}, 49 | \code{\link{gb_df_create}()}, 50 | \code{\link{gb_df_generate}()}, 51 | \code{\link{gb_sql_add}()}, 52 | \code{\link{gb_sql_query}()}, 53 | \code{\link{gbrelease_check}()}, 54 | \code{\link{gbrelease_get}()}, 55 | \code{\link{gbrelease_log}()}, 56 | \code{\link{has_data}()}, 57 | \code{\link{identify_downloadable_files}()}, 58 | \code{\link{last_add_get}()}, 59 | \code{\link{last_dwnld_get}()}, 60 | \code{\link{last_entry_get}()}, 61 | \code{\link{latest_genbank_release}()}, 62 | \code{\link{latest_genbank_release_notes}()}, 63 | \code{\link{message_missing}()}, 64 | \code{\link{mock_def}()}, 65 | \code{\link{mock_gb_df_generate}()}, 66 | \code{\link{mock_org}()}, 67 | \code{\link{mock_rec}()}, 68 | \code{\link{mock_seq}()}, 69 | \code{\link{predict_datasizes}()}, 70 | \code{\link{readme_log}()}, 71 | \code{\link{restez_connect}()}, 72 | \code{\link{restez_disconnect}()}, 73 | \code{\link{restez_path_check}()}, 74 | \code{\link{restez_rl}()}, 75 | \code{\link{search_gz}()}, 76 | \code{\link{seshinfo_log}()}, 77 | \code{\link{setup}()}, 78 | \code{\link{slctn_get}()}, 79 | \code{\link{slctn_log}()}, 80 | \code{\link{sql_path_get}()}, 81 | \code{\link{stat}()}, 82 | \code{\link{testdatadir_get}()} 83 | } 84 | \concept{private} 85 | -------------------------------------------------------------------------------- /man/testdatadir_get.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/test-tools.R 3 | \name{testdatadir_get} 4 | \alias{testdatadir_get} 5 | \title{Get test data directory} 6 | \usage{ 7 | testdatadir_get() 8 | } 9 | \description{ 10 | Get the folder containing test data. 11 | } 12 | \seealso{ 13 | Other private: 14 | \code{\link{add_rcrd_log}()}, 15 | \code{\link{cat_line}()}, 16 | \code{\link{char}()}, 17 | \code{\link{check_connection}()}, 18 | \code{\link{cleanup}()}, 19 | \code{\link{connected}()}, 20 | \code{\link{connection_get}()}, 21 | \code{\link{db_download_intern}()}, 22 | \code{\link{db_sqlngths_get}()}, 23 | \code{\link{db_sqlngths_log}()}, 24 | \code{\link{dir_size}()}, 25 | \code{\link{dwnld_path_get}()}, 26 | \code{\link{dwnld_rcrd_log}()}, 27 | \code{\link{entrez_fasta_get}()}, 28 | \code{\link{entrez_gb_get}()}, 29 | \code{\link{extract_accession}()}, 30 | \code{\link{extract_by_patterns}()}, 31 | \code{\link{extract_clean_sequence}()}, 32 | \code{\link{extract_definition}()}, 33 | \code{\link{extract_features}()}, 34 | \code{\link{extract_inforecpart}()}, 35 | \code{\link{extract_keywords}()}, 36 | \code{\link{extract_locus}()}, 37 | \code{\link{extract_organism}()}, 38 | \code{\link{extract_seqrecpart}()}, 39 | \code{\link{extract_sequence}()}, 40 | \code{\link{extract_version}()}, 41 | \code{\link{file_download}()}, 42 | \code{\link{filename_log}()}, 43 | \code{\link{flatfile_read}()}, 44 | \code{\link{gb_build}()}, 45 | \code{\link{gb_df_create}()}, 46 | \code{\link{gb_df_generate}()}, 47 | \code{\link{gb_sql_add}()}, 48 | \code{\link{gb_sql_query}()}, 49 | \code{\link{gbrelease_check}()}, 50 | \code{\link{gbrelease_get}()}, 51 | \code{\link{gbrelease_log}()}, 52 | \code{\link{has_data}()}, 53 | \code{\link{identify_downloadable_files}()}, 54 | \code{\link{last_add_get}()}, 55 | \code{\link{last_dwnld_get}()}, 56 | \code{\link{last_entry_get}()}, 57 | \code{\link{latest_genbank_release}()}, 58 | \code{\link{latest_genbank_release_notes}()}, 59 | \code{\link{message_missing}()}, 60 | \code{\link{mock_def}()}, 61 | \code{\link{mock_gb_df_generate}()}, 62 | \code{\link{mock_org}()}, 63 | \code{\link{mock_rec}()}, 64 | \code{\link{mock_seq}()}, 65 | \code{\link{predict_datasizes}()}, 66 | \code{\link{readme_log}()}, 67 | \code{\link{restez_connect}()}, 68 | \code{\link{restez_disconnect}()}, 69 | \code{\link{restez_path_check}()}, 70 | \code{\link{restez_rl}()}, 71 | \code{\link{search_gz}()}, 72 | \code{\link{seshinfo_log}()}, 73 | \code{\link{setup}()}, 74 | \code{\link{slctn_get}()}, 75 | \code{\link{slctn_log}()}, 76 | \code{\link{sql_path_get}()}, 77 | \code{\link{stat}()}, 78 | \code{\link{status_class}()} 79 | } 80 | \concept{private} 81 | -------------------------------------------------------------------------------- /other/demonstration_1.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(restez) 3 | 4 | # SETUP 5 | fp <- file.path(getwd(), 'demo') 6 | dir.create(fp) 7 | restez_path_set(fp) 8 | gb_download() 9 | db_create() 10 | 11 | # LOOK UP ACCESSIONS 12 | # 33553 - Sciuromorpha - squirrel-like things 13 | search_term <- 'txid33553[Organism:exp] AND COI [GENE]' 14 | search_object <- rentrez::entrez_search(db = 'nucleotide', term = search_term, 15 | use_history = TRUE, retmax = 0) 16 | accessions <- rentrez::entrez_fetch(db = 'nucleotide', 17 | web_history = search_object$web_history, 18 | rettype = 'acc') 19 | accessions <- strsplit(x = accessions, split = '\\n')[[1]] 20 | accessions <- sub(pattern = '\\.[0-9]+', replacement = '', x = accessions) 21 | 22 | # FETCH 23 | coi_sequences <- gb_fasta_get(id = accessions) 24 | # are all accessions in results? 25 | all(accessions %in% names(coi_sequences)) 26 | # .... no 27 | accessions[!accessions %in% names(coi_sequences)] 28 | # NC* refers to RefSeq sequences and are not currently available through restez 29 | # The sequence however exists in GB under a different id which we can find like so 30 | smmry <- rentrez::entrez_summary(db = 'nucleotide', id = 'NC_027278') 31 | # This ID does exist in our results. 32 | smmry$assemblyacc %in% accessions 33 | 34 | # TAKE DOWN 35 | db_delete() 36 | -------------------------------------------------------------------------------- /other/dev.R: -------------------------------------------------------------------------------- 1 | # Problem record 2 | devtools::load_all() 3 | record <- rentrez::entrez_fetch(db = 'nucleotide', id = 'AB467315', 4 | rettype = 'gb', retmode = 'text') 5 | extract_features(record) 6 | 7 | 8 | # TODO: 9 | problem_files <- c('gbinv17.seq', 'gbpln118.seq', 'gbpln12.seq') 10 | restez_path_set('.') 11 | for (problem_file in problem_files) { 12 | file_download(problem_file) 13 | } 14 | 15 | db_create() 16 | 17 | devtools::load_all('.') 18 | 19 | db_delete(everything = TRUE) 20 | restez_path_set('.') 21 | db_download() 22 | 23 | restez_connect() 24 | db_create(min_length = 10, max_length = 1000) 25 | restez_status() 26 | db_delete() 27 | restez_connect() 28 | db_create() 29 | restez_status() 30 | restez_disconnect() 31 | 32 | 33 | -------------------------------------------------------------------------------- /other/generate_records_data.R: -------------------------------------------------------------------------------- 1 | filepath <- file.path('other', "gbpri10.seq") 2 | records <- restez:::flatfile_read(filepath) 3 | records <- sample(records, 25) 4 | save(records, file = file.path('data', 'records.rda'), compress = 'xz') 5 | -------------------------------------------------------------------------------- /other/make_sure_things_work.R: -------------------------------------------------------------------------------- 1 | library(restez) 2 | # On my linux computer and I specify my desktop 3 | fp <- '~/Desktop' 4 | restez_path_set(filepath = fp) 5 | 6 | db_download(preselection = '15') 7 | db_create(db_type = 'nucleotide', min_length = 250, max_length = 2000) 8 | -------------------------------------------------------------------------------- /other/phylotar_demo.R: -------------------------------------------------------------------------------- 1 | # SPEED TEST DOWNLOAD STAGE WITH AND WITHOUT RESTEZ 2 | # LIBS ---- 3 | devtools::load_all('~/Coding/restez') 4 | devtools::load_all('~/Coding/phylotaR') 5 | 6 | # VARS ---- 7 | wd <- 'beavers' 8 | restez_path <- '~/Desktop/beavers_restez' 9 | txid <- 1963757 10 | ncbi_dr <- '/usr/bin' 11 | 12 | # RUN PHYLOTAR ---- 13 | if (dir.exists(wd)) { 14 | unlink(x = wd, recursive = TRUE) 15 | } 16 | dir.create(wd) 17 | phylotaR::setup(wd = wd, txid = txid, ncbi_dr = ncbi_dr, v = TRUE, btchsz = 100) 18 | taxise_run(wd = wd) 19 | 20 | # SPEED TEST 1 21 | without_restez <- system.time(expr = { 22 | download_run(wd = wd) 23 | }) 24 | # user system elapsed 25 | # 121.104 32.604 467.001 26 | # user system elapsed 27 | # 63.128 0.883 438.626 28 | 29 | # SET-UP RESTEZ ---- 30 | restez_path_set(filepath = restez_path) 31 | db_download(preselection = '15') # select 15 for rodents 32 | restez_connect() 33 | db_create() 34 | restez_disconnect() 35 | restez_status() 36 | 37 | # RESET PHYLOTAR ---- 38 | reset(wd = wd, stage = 'download', hard = TRUE) 39 | 40 | # SPEED TEST 2 41 | restez_path_set(filepath = restez_path) 42 | restez_connect() 43 | with_restez <- system.time(expr = { 44 | download_run(wd = wd) 45 | }) 46 | restez_disconnect() 47 | # user system elapsed 48 | # 71.983 14.783 234.640 49 | 50 | # TAKE HOME STAT 51 | # twice as faster 52 | pfaster <- signif((467.001/234.640), 3) 53 | cat(pfaster, 'x faster with restez\n', sep = '') 54 | -------------------------------------------------------------------------------- /other/rodent_build_times.rds: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/other/rodent_build_times.rds -------------------------------------------------------------------------------- /other/rodent_db.R: -------------------------------------------------------------------------------- 1 | rodents_path <- file.path(getwd(), 'rodents') 2 | if (!dir.exists(rodents_path)) { 3 | dir.create(rodents_path) 4 | } 5 | 6 | # Dynamically get Rodent index (may change with future GenBank updates) 7 | restez_path_set(tempdir()) 8 | latest_genbank_release_notes() 9 | downloadable_table <- identify_downloadable_files() 10 | types <- sort(table(downloadable_table[['descripts']]), decreasing = TRUE) 11 | rodent_index <- grep("Rodent|rodent", names(types)) 12 | 13 | if (length(rodent_index) != 1) stop("multiple rodent indexes detected") 14 | if (!is.numeric(rodent_index)) stop("rodent index should be numeric") 15 | 16 | # set the restez path to a memorable location 17 | restez_path_set(rodents_path) 18 | # download domain 10 and build db, record times 19 | dl_time <- system.time( 20 | db_download(preselection = rodent_index, max_tries = 100)) 21 | 22 | db_time <- system.time(db_create(min_length = 100, max_length = 1000)) 23 | 24 | # save times to RDS for documentation in vignette 25 | rodent_times <- list( 26 | dl_time = dl_time, 27 | db_time = db_time 28 | ) 29 | 30 | saveRDS(rodent_times, "other/rodent_build_times.rds") -------------------------------------------------------------------------------- /other/spell_check.R: -------------------------------------------------------------------------------- 1 | ignore <- c('restez', 2 | 'db', 3 | "restez's", 4 | 'acc', 5 | 'booleans', 6 | 'entrez', 7 | 'Entrez', 8 | 'fasta', 9 | 'FASTA', 10 | 'filepath', 11 | 'GenBank', 12 | 'nuccore', 13 | 'REGEX', 14 | 'rentrez', 15 | "rentrez's", 16 | 'retmode', 17 | 'Rettypes', 18 | 'seq', 19 | 'seqid', 20 | 'SQLite') 21 | #'uilist') 22 | devtools::spell_check(ignore = ignore, dict = 'en_GB') 23 | vignette_files <- file.path(getwd(), 'vignettes', 24 | list.files('vignettes', pattern = 'Rmd')) 25 | for (fl in vignette_files) { 26 | print(fl) 27 | print(devtools:::spell_check_file(fl, ignore = ignore, dict = 'en_GB')) 28 | } 29 | -------------------------------------------------------------------------------- /other/viral_lookup.R: -------------------------------------------------------------------------------- 1 | # How long does it take to get the definition of a single ID from viral sequences? 2 | # setup 3 | devtools::load_all('~/Coding/restez') 4 | restez_path_set('~/Desktop/viral') 5 | db_delete(everything = TRUE) 6 | restez_path_set('~/Desktop/viral') 7 | db_download(preselection = '13') 8 | restez_connect() 9 | db_create() 10 | restez_disconnect() 11 | 12 | # get def 13 | devtools::load_all('~/Coding/restez') 14 | restez_path_set('~/Desktop/viral') 15 | restez_connect() 16 | id <- sample(list_db_ids(), 1) 17 | system.time(gb_definition_get(id)) 18 | # user system elapsed 19 | # 0.048 0.000 0.024 -------------------------------------------------------------------------------- /paper/outline.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/paper/outline.png -------------------------------------------------------------------------------- /pkgdown/favicon/apple-touch-icon-120x120.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/apple-touch-icon-120x120.png -------------------------------------------------------------------------------- /pkgdown/favicon/apple-touch-icon-152x152.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/apple-touch-icon-152x152.png -------------------------------------------------------------------------------- /pkgdown/favicon/apple-touch-icon-180x180.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/apple-touch-icon-180x180.png -------------------------------------------------------------------------------- /pkgdown/favicon/apple-touch-icon-60x60.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/apple-touch-icon-60x60.png -------------------------------------------------------------------------------- /pkgdown/favicon/apple-touch-icon-76x76.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/apple-touch-icon-76x76.png -------------------------------------------------------------------------------- /pkgdown/favicon/apple-touch-icon.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/apple-touch-icon.png -------------------------------------------------------------------------------- /pkgdown/favicon/favicon-16x16.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/favicon-16x16.png -------------------------------------------------------------------------------- /pkgdown/favicon/favicon-32x32.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/favicon-32x32.png -------------------------------------------------------------------------------- /pkgdown/favicon/favicon.ico: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/pkgdown/favicon/favicon.ico -------------------------------------------------------------------------------- /revdep/.gitignore: -------------------------------------------------------------------------------- 1 | checks 2 | library 3 | checks.noindex 4 | library.noindex 5 | cloud.noindex 6 | data.sqlite 7 | *.html 8 | -------------------------------------------------------------------------------- /revdep/README.md: -------------------------------------------------------------------------------- 1 | # Platform 2 | 3 | |field |value | 4 | |:--------|:-----------------------------| 5 | |version |R version 4.4.1 (2024-06-14) | 6 | |os |macOS Sonoma 14.6.1 | 7 | |system |aarch64, darwin20 | 8 | |ui |X11 | 9 | |language |(EN) | 10 | |collate |ja_JP.UTF-8 | 11 | |ctype |ja_JP.UTF-8 | 12 | |tz |Asia/Tokyo | 13 | |date |2025-03-07 | 14 | |pandoc |3.1.2 @ /usr/local/bin/pandoc | 15 | 16 | # Dependencies 17 | 18 | |package |old |new |Δ | 19 | |:--------|:-----|:----------|:--| 20 | |restez |2.1.4 |2.1.4.9000 |* | 21 | |ape |NA |5.8-1 |* | 22 | |askpass |NA |1.2.1 |* | 23 | |cli |NA |3.6.4 |* | 24 | |curl |NA |6.2.1 |* | 25 | |digest |NA |0.6.37 |* | 26 | |fs |NA |1.6.5 |* | 27 | |jsonlite |NA |1.9.1 |* | 28 | |openssl |NA |2.3.2 |* | 29 | |R6 |NA |2.6.1 |* | 30 | |Rcpp |NA |1.0.14 |* | 31 | |sys |NA |3.4.3 |* | 32 | |XML |NA |3.99-0.18 |* | 33 | 34 | # Revdeps 35 | 36 | -------------------------------------------------------------------------------- /revdep/cran.md: -------------------------------------------------------------------------------- 1 | ## revdepcheck results 2 | 3 | We checked 0 reverse dependencies, comparing R CMD check results across CRAN and dev versions of this package. 4 | 5 | * We saw 0 new problems 6 | * We failed to check 0 packages 7 | 8 | -------------------------------------------------------------------------------- /revdep/email.yml: -------------------------------------------------------------------------------- 1 | release_date: ??? 2 | rel_release_date: ??? 3 | my_news_url: ??? 4 | release_version: ??? 5 | release_details: ??? 6 | -------------------------------------------------------------------------------- /revdep/failures.md: -------------------------------------------------------------------------------- 1 | *Wow, no problems at all. :)* -------------------------------------------------------------------------------- /revdep/problems.md: -------------------------------------------------------------------------------- 1 | *Wow, no problems at all. :)* -------------------------------------------------------------------------------- /tests/test-all.R: -------------------------------------------------------------------------------- 1 | library(testthat) 2 | test_check("restez") 3 | -------------------------------------------------------------------------------- /tests/testthat/data/records.RData: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/tests/testthat/data/records.RData -------------------------------------------------------------------------------- /tests/testthat/data/release_notes_gb224.RData: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ropensci/restez/2fca23308ad67f064e01cf7e88328a0f98540c28/tests/testthat/data/release_notes_gb224.RData -------------------------------------------------------------------------------- /tests/testthat/test-biomartr-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # RUNNING 5 | test_that("check_connection() works", { 6 | local_mocked_bindings( 7 | check_connection = function(...) TRUE 8 | ) 9 | expect_true(check_connection()) 10 | }) 11 | 12 | test_that("url_exists() works", { 13 | local_mocked_bindings( 14 | url_exists = function(...) FALSE 15 | ) 16 | expect_error( 17 | check_connection(), 18 | "Unable to connect to" 19 | ) 20 | }) 21 | -------------------------------------------------------------------------------- /tests/testthat/test-connection-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # RUNNING 5 | cleanup() 6 | context('Testing \'connection-tools\'') 7 | test_that('restez_ready() works', { 8 | on.exit(cleanup()) 9 | expect_false(restez_ready()) 10 | setup() 11 | demo_db_create() 12 | expect_true(restez_ready()) 13 | cleanup() 14 | }) 15 | test_that('restez_connect() works', { 16 | on.exit({ 17 | restez_disconnect() 18 | cleanup() 19 | }) 20 | expect_error(restez_connect()) 21 | restez_disconnect() 22 | setup() 23 | restez_connect() 24 | expect_true(is(connection_get(), 'duckdb_connection')) 25 | restez_disconnect() 26 | cleanup() 27 | }) 28 | test_that('restez_connect() works in read-only mode', { 29 | on.exit({ 30 | cleanup() 31 | restez_disconnect() 32 | }) 33 | setup() 34 | demo_db_create(n = 10) 35 | restez_connect(read_only = TRUE) 36 | expect_true(is(connection_get(), 'duckdb_connection')) 37 | restez_disconnect() 38 | cleanup() 39 | }) 40 | test_that('restez_disconnect() works', { 41 | on.exit({ 42 | cleanup() 43 | restez_disconnect() 44 | }) 45 | expect_null(restez_disconnect()) 46 | setup() 47 | restez_connect() 48 | expect_true(is(connection_get(), 'duckdb_connection')) 49 | restez_disconnect() 50 | expect_error(connection_get()) 51 | cleanup() 52 | }) 53 | test_that('connected() works', { 54 | on.exit({ 55 | cleanup() 56 | restez_disconnect() 57 | }) 58 | expect_false(connected()) 59 | setup() 60 | expect_false(connected()) 61 | restez_connect() 62 | expect_true(connected()) 63 | cleanup() 64 | restez_disconnect() 65 | }) 66 | test_that('has_data() works', { 67 | on.exit({ 68 | cleanup() 69 | restez_disconnect() 70 | }) 71 | expect_false(has_data()) 72 | setup() 73 | expect_false(has_data()) 74 | demo_db_create(n = 10) 75 | expect_true(has_data()) 76 | cleanup() 77 | restez_disconnect() 78 | }) 79 | test_that('connection_get() works', { 80 | on.exit({ 81 | cleanup() 82 | restez_disconnect() 83 | }) 84 | setup() 85 | restez_connect() 86 | expect_true(is(connection_get(), 'duckdb_connection')) 87 | restez_disconnect() 88 | cleanup() 89 | expect_error(connection_get()) 90 | }) 91 | cleanup() 92 | -------------------------------------------------------------------------------- /tests/testthat/test-entrez-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # VARS 5 | nrcrds <- 10 # how many fake records to test on? 6 | data_d <- testdatadir_get() 7 | 8 | # DATA 9 | records <- readRDS(file = file.path(data_d, 'records.RData')) 10 | 11 | # SETUP 12 | cleanup() 13 | setup() 14 | df <- gb_df_generate(records = sample(records, size = nrcrds)) 15 | df <- gb_df_generate(records = records) 16 | ids <- as.character(df[['accession']]) 17 | gb_sql_add(df = df) 18 | 19 | # RUNNING 20 | context('Testing \'entrez-tools\'') 21 | test_that('entrez_fasta_get() works', { 22 | # rentrez_fastas <- rentrez::entrez_fetch(id = id, db = 'nucleotide', 23 | # rettype = 'fasta') 24 | res <- entrez_fasta_get(id = sample(ids, 2)) 25 | expect_true(inherits(res, 'character')) 26 | mtch_obj <- gregexpr(pattern = '\n\n', text = res)[[1]] 27 | expect_true(length(mtch_obj) == 2) 28 | expect_true(grepl(pattern = '^>.*', x = res)) 29 | # if not in local, should search internet 30 | local_mocked_bindings( 31 | entrez_fetch_wrap = function(...) '>notanid\nATCG\n\n' 32 | ) 33 | res <- entrez_fasta_get(id = 'notanid') 34 | expect_true(grepl('>notanid', res)) 35 | # should be able to handle mixture 36 | res <- entrez_fasta_get(id = c(sample(ids, 2), 'notanid')) 37 | expect_true(grepl('>notanid', res)) 38 | mtch_obj <- gregexpr(pattern = '\n\n', text = res)[[1]] 39 | expect_true(length(mtch_obj) == 3) 40 | }) 41 | test_that('entrez_gb_get() works', { 42 | # id <- c('S71333', 'AY952423') 43 | # rentrez_records <- rentrez::entrez_fetch(id = id, db = 'nucleotide', 44 | # rettype = 'gb') 45 | id <- sample(ids, 2) 46 | res <- entrez_gb_get(id = id, db = 'nucleotide', 47 | rettype = 'gb') 48 | # res == rentrez_records 49 | expect_true(inherits(res, 'character')) 50 | mtch_obj <- gregexpr(pattern = '\n\n', text = res)[[1]] 51 | expect_true(length(mtch_obj) == 2) 52 | expect_true(grepl(pattern = 'LOCUS', x = res)) 53 | # if not in local, should search internet 54 | local_mocked_bindings( 55 | entrez_fetch_wrap = function(...) 'LOCUS notanid\n//\n\n' 56 | ) 57 | res <- entrez_gb_get(id = 'notanid') 58 | expect_true(grepl('notanid', res)) 59 | # should be able to handle mixture 60 | res <- entrez_gb_get(id = c(sample(ids, 2), 'notanid')) 61 | expect_true(grepl('notanid', res)) 62 | mtch_obj <- gregexpr(pattern = 'LOCUS', text = res)[[1]] 63 | expect_true(length(mtch_obj) == 3) 64 | }) 65 | cleanup() 66 | -------------------------------------------------------------------------------- /tests/testthat/test-extract-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # VARS 5 | data_d <- testdatadir_get() 6 | 7 | # DATA 8 | # TODO: get ten random records of each NCBI domain 9 | records <- readRDS(file = file.path(data_d, 'records.RData')) 10 | 11 | # RUNNING 12 | context('Testing \'extract-tools\'') 13 | test_that('extract_by_patterns() works', { 14 | record <- sample(records, 1)[[1]] 15 | res <- extract_by_patterns(record = record, 16 | start_pattern = '\nFEATURES\\s{2,}') 17 | expect_true(grepl('Location/Qualifiers', res)) 18 | }) 19 | test_that('extract_version() works', { 20 | record <- sample(records, 1)[[1]] 21 | accession_version <- extract_version(record = record) 22 | # print(accession_version) 23 | expect_true(grepl('^[a-z0-9_]+\\.[0-9]+$', accession_version, 24 | ignore.case = TRUE)) 25 | }) 26 | test_that('extract_organism() works', { 27 | record <- sample(records, 1)[[1]] 28 | organism <- extract_organism(record = record) 29 | # print(organism) 30 | expect_false(grepl('\\s{2,}', organism, 31 | ignore.case = TRUE)) 32 | expect_false(grepl('\n', organism, ignore.case = TRUE)) 33 | }) 34 | test_that('extract_definition() works', { 35 | record <- sample(records, 1)[[1]] 36 | definition <- extract_definition(record = record) 37 | #print(definition) 38 | expect_false(grepl('\\s{2,}', definition, 39 | ignore.case = TRUE)) 40 | expect_false(grepl('\n', definition, ignore.case = TRUE)) 41 | }) 42 | test_that('extract_sequence() works', { 43 | record <- sample(records, 1)[[1]] 44 | sequence <- extract_sequence(record = record) 45 | expect_false(grepl('[0-9]', sequence, 46 | ignore.case = TRUE)) 47 | }) 48 | test_that('extract_features() works', { 49 | record <- sample(records, 1)[[1]] 50 | features <- extract_features(record = record) 51 | # print(features) 52 | expect_true(inherits(features, 'list')) 53 | }) 54 | test_that('extract_locus() works', { 55 | record <- sample(records, 1)[[1]] 56 | locus <- extract_locus(record = record) 57 | # print(locus) 58 | expect_true(inherits(locus, 'character')) 59 | }) 60 | test_that('extract_keywords() works', { 61 | record <- sample(records, 1)[[1]] 62 | keywords <- extract_keywords(record = record) 63 | # print(keywords) 64 | expect_true(inherits(keywords, 'character')) 65 | }) 66 | test_that('gb_extract() works', { 67 | opts <- c('accession', 'version', 'organism','sequence', 'definition', 68 | 'locus', 'features', 'keywords') 69 | what <- sample(opts, 1) 70 | record <- sample(records, 1)[[1]] 71 | res <- gb_extract(record = record, what = what) 72 | if (what == 'features') { 73 | expect_true(inherits(res, 'list')) 74 | } else { 75 | expect_true(inherits(res, 'character')) 76 | } 77 | }) 78 | -------------------------------------------------------------------------------- /tests/testthat/test-filepath-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # RUNNING 5 | cleanup() 6 | context('Testing \'filepath-tools\'') 7 | test_that('restez_path_set() works', { 8 | setup() 9 | on.exit(cleanup()) 10 | expect_true(restez_path_get() == file.path('test_db_fldr', 'restez')) 11 | }) 12 | test_that('restez_path_unset() works', { 13 | setup() 14 | on.exit(cleanup()) 15 | restez_path_unset() 16 | expect_null(restez_path_get()) 17 | }) 18 | test_that('restez_path_get() works', { 19 | setup() 20 | on.exit(cleanup()) 21 | expect_true(grepl('test_db_fldr', restez_path_get())) 22 | }) 23 | test_that('sql_path_get() works', { 24 | setup() 25 | on.exit(cleanup()) 26 | expect_true(is.character(sql_path_get())) 27 | }) 28 | test_that('dwnld_path_get() works', { 29 | setup() 30 | on.exit(cleanup()) 31 | expect_true(is.character(dwnld_path_get())) 32 | }) 33 | test_that('restez_path_check() works', { 34 | expect_error(restez_path_check()) 35 | setup() 36 | on.exit(cleanup()) 37 | expect_null(restez_path_check()) 38 | unlink('test_db_fldr', recursive = TRUE) 39 | expect_error(restez_path_check()) 40 | }) 41 | cleanup() 42 | -------------------------------------------------------------------------------- /tests/testthat/test-gb-setup-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # VARS 5 | nrcrds <- 50 # how many fake records to test on? 6 | data_d <- testdatadir_get() 7 | 8 | # DATA 9 | records <- readRDS(file = file.path(data_d, 'records.RData')) 10 | 11 | # FUNCTIONS 12 | write_fake_records <- function(n=nrcrds) { 13 | records_text <- '' 14 | for (i in 1:n) { 15 | records_text <- paste0(records_text, 16 | mock_rec(i, sequence = 'atcg'), '\n') 17 | } 18 | cat(records_text, file = 'test_records.txt') 19 | NULL 20 | } 21 | 22 | # RUNNING 23 | context('Testing \'setup-tools\'') 24 | cleanup() 25 | test_that('gb_build() works', { 26 | local_mocked_bindings( 27 | flatfile_read = function(...) NULL 28 | ) 29 | res <- gb_build( 30 | dpth = NULL, seq_files = 1:10, max_length = NULL, min_length = NULL) 31 | expect_true(res) 32 | res <- gb_build( 33 | dpth = NULL, seq_files = NULL, max_length = NULL, min_length = NULL) 34 | expect_false(res) 35 | }) 36 | test_that('flatfile_read() works', { 37 | write_fake_records(n = nrcrds) 38 | records <- flatfile_read(flpth = 'test_records.txt') 39 | expect_true(length(records) == nrcrds) 40 | cleanup() 41 | }) 42 | test_that('gb_df_create() works', { 43 | fake_data <- rep('', nrcrds) 44 | df <- gb_df_create(accessions = fake_data, versions = fake_data, 45 | organisms = fake_data, definitions = fake_data, 46 | sequences = fake_data, records = fake_data) 47 | expect_true(nrow(df) == nrcrds) 48 | }) 49 | test_that('gb_df_generate() works', { 50 | df <- gb_df_generate(records = sample(records, size = 3)) 51 | expect_true(nrow(df) == 3) 52 | expctd_clnms <- c("accession", "version", "organism", "raw_definition", 53 | "raw_sequence", "raw_record") 54 | expect_true(all(colnames(df) %in% expctd_clnms)) 55 | }) 56 | test_that('gb_df_generate() can filter by accession', { 57 | accs_filter <- c("AC092025", "AC090116", "AC091644") 58 | df <- gb_df_generate( 59 | records = records, 60 | acc_filter = accs_filter) 61 | expect_equal(sort(accs_filter), sort(df$accession)) 62 | expect_true(nrow(df) == 3) 63 | expctd_clnms <- c("accession", "version", "organism", "raw_definition", 64 | "raw_sequence", "raw_record") 65 | expect_true(all(colnames(df) %in% expctd_clnms)) 66 | }) 67 | test_that('gb_sql_add() works', { 68 | setup() 69 | on.exit(cleanup()) 70 | df <- gb_df_generate(records = sample(records, size = 3)) 71 | gb_sql_add(df = df) 72 | expect_true(file.exists(sql_path_get())) 73 | }) 74 | cleanup() 75 | -------------------------------------------------------------------------------- /tests/testthat/test-mock-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # RUNNING 5 | context('Testing \'mock-tools\'') 6 | test_that('mock_rec() works', { 7 | record <- mock_rec(i = 1) 8 | expect_true(is.character(record)) 9 | }) 10 | test_that('mock_seq() works', { 11 | sequence <- mock_seq(i = 1, sqlngth = 100) 12 | expect_true(grepl(pattern = '[atcg]', x = sequence)) 13 | expect_true(nchar(gsub(pattern = '[^atcg]', replacement = '', 14 | x = sequence)) == 100) 15 | }) 16 | test_that('mock_def() works', { 17 | def <- mock_def(i = 1) 18 | expect_true(is.character(def)) 19 | }) 20 | test_that('mock_org() works', { 21 | org <- mock_org(i = 1) 22 | expect_true(is.character(org)) 23 | }) 24 | test_that('mock_gb_df_generate() works', { 25 | df <- mock_gb_df_generate(n = 100) 26 | expect_true(nrow(df) == 100) 27 | }) 28 | -------------------------------------------------------------------------------- /tests/testthat/test-print-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # RUNNING 5 | context('Testing \'print-tools\'') 6 | test_that('stat() works', { 7 | coloured_text <- stat('example text', 'example text') 8 | cat(coloured_text) 9 | expect_true(inherits(coloured_text, 'character')) 10 | }) 11 | test_that('char() works', { 12 | coloured_text <- char(x = 'example text') 13 | cat(coloured_text) 14 | expect_true(inherits(coloured_text, 'character')) 15 | }) 16 | test_that('cat_line() works', { 17 | nores <- cat_line('example text') 18 | expect_null(nores) 19 | }) 20 | -------------------------------------------------------------------------------- /tests/testthat/test-rentrez-wrappers.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # VARS 5 | nrcrds <- 10 # how many fake records to test on? 6 | wd <- getwd() 7 | data_d <- testdatadir_get() 8 | 9 | # DATA 10 | records <- readRDS(file = file.path(data_d, 'records.RData')) 11 | 12 | # SETUP 13 | cleanup() 14 | setup() 15 | df <- gb_df_generate(records = sample(records, size = nrcrds)) 16 | ids <- as.character(df[['accession']]) 17 | gb_sql_add(df = df) 18 | 19 | # RUNNING 20 | context('Testing \'rentrez-wrappers\'') 21 | test_that('entrez_fetch() works', { 22 | # TODO, what if an ID that is not in the local db is given? 23 | fasta_res <- entrez_fetch(db = 'nucleotide', id = sample(ids, 2), 24 | rettype = 'fasta') 25 | gb_res <- entrez_fetch(db = 'nucleotide', id = sample(ids, 2), 26 | rettype = 'gb') 27 | # xml is not supported, rentrez will be called 28 | local_mocked_bindings( 29 | entrez_fetch = function(...) TRUE, .package = "rentrez" 30 | ) 31 | res <- entrez_fetch(db = 'nucleotide', id = sample(ids, 2), 32 | rettype = 'gb', retmode = 'xml') 33 | expect_true(res) 34 | }) 35 | cleanup() 36 | -------------------------------------------------------------------------------- /tests/testthat/test-status-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # RUNNING 5 | cleanup() 6 | context('Testing \'status-tools\'') 7 | test_that('restez_status() works', { 8 | on.exit(cleanup()) 9 | expect_error(restez_status(gb_check = FALSE)) 10 | setup() 11 | status_obj <- restez_status(gb_check = FALSE) 12 | expect_true(status_obj$`Restez path`$`Does path exist?`) 13 | demo_db_create(n = 10) 14 | status_obj <- restez_status(gb_check = FALSE) 15 | expect_true(status_obj$Database$`Does path exist?`) 16 | expect_equal(as.character(status_obj$Database$`Total size`), "780K") 17 | expect_true(status_obj$Database$`Does the database have data?`) 18 | expect_equal(status_obj$Database$`Number of sequences`, 10) 19 | expect_equal(status_obj$Database$`Min. sequence length`, "0") 20 | expect_equal(status_obj$Database$`Max. sequence length`, "Inf") 21 | }) 22 | test_that('status_class() works', { 23 | cleanup() 24 | setup() 25 | expect_true(inherits(status_class(), 'status')) 26 | }) 27 | test_that('print.status() works', { 28 | cleanup() 29 | setup() 30 | expect_null(print.status(status_class())) 31 | }) 32 | cleanup() 33 | -------------------------------------------------------------------------------- /tests/testthat/test-test-tools.R: -------------------------------------------------------------------------------- 1 | # LIBS 2 | library(testthat) 3 | 4 | # RUNNING 5 | context('Testing \'test-tools\'') 6 | test_that('cleanup() works', { 7 | dir.create('test_db_fldr') 8 | cleanup() 9 | expect_false(dir.exists('test_db_fldr')) 10 | }) 11 | test_that('testdatadir_get() works', { 12 | expect_true(dir.exists(testdatadir_get())) 13 | }) 14 | -------------------------------------------------------------------------------- /vignettes/precompile.R: -------------------------------------------------------------------------------- 1 | # Load package in working state 2 | # not with library() 3 | library(devtools) 4 | library(knitr) 5 | load_all() 6 | 7 | # First five vignettes require downloaded rodent db, must be precompiled: 8 | 9 | # gen rodent db 10 | if (!dir.exists(file.path('rodents'))) { 11 | source(file.path('other', 'rodent_db.R')) 12 | } 13 | 14 | # precompile 15 | vgnts <- c('1_rodents.Rmd', '2_search_and_fetch.Rmd', '3_parsing.Rmd', 16 | '4_phylotaR.Rmd', '5_tips_and_tricks.Rmd') 17 | for (vgnt in vgnts) { 18 | knit(paste0("vignettes/", vgnt, ".orig"), paste0("vignettes/", vgnt)) 19 | } 20 | 21 | build_vignettes() 22 | --------------------------------------------------------------------------------