├── .bumpversion.cfg
├── .flake8
├── .gitignore
├── .readthedocs.yml
├── .travis.yml
├── CONTRIBUTING.rst
├── LICENSE
├── MANIFEST.in
├── README.rst
├── docs
    ├── Makefile
    └── source
    │   ├── bio2bel_repositories.rst
    │   ├── cli
    │       ├── inference.rst
    │       ├── summarize.rst
    │       └── train_and_evaluate.rst
    │   ├── conf.py
    │   ├── convert.rst
    │   ├── hyper_parameter_optimization.rst
    │   ├── images
    │       ├── batch_size.png
    │       ├── data_source.png
    │       ├── epochs.png
    │       ├── execution_mode.png
    │       ├── output_directory.png
    │       ├── preferred_device.png
    │       ├── provide_dataset.png
    │       ├── random_seed.png
    │       ├── select_database.png
    │       └── select_model.png
    │   ├── index.rst
    │   ├── installation.rst
    │   └── train_and_evaluate.rst
├── notebooks
    ├── Case Scenario ADEPTUS and HSDN.ipynb
    ├── Case Scenario ADEPTUS.ipynb
    ├── Case Scenario ComPath.ipynb
    ├── Case Scenario HSDN.ipynb
    └── README.rst
├── requirements-rtd.txt
├── setup.cfg
├── setup.py
├── src
    └── biokeen
    │   ├── __init__.py
    │   ├── __main__.py
    │   ├── cli
    │       ├── __init__.py
    │       ├── __main__.py
    │       ├── cli.py
    │       ├── messages.py
    │       └── prompts.py
    │   ├── constants.py
    │   ├── content.py
    │   └── convert
    │       ├── __init__.py
    │       ├── converters.py
    │       └── io.py
├── tests
    ├── __init__.py
    ├── test_convert.py
    └── test_simple.py
└── tox.ini


/.bumpversion.cfg:
--------------------------------------------------------------------------------
 1 | [bumpversion]
 2 | current_version = 0.0.15-dev
 3 | commit = True
 4 | tag = False
 5 | parse = (?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)(?:-(?P<release>[0-9A-Za-z-]+(?:\.[0-9A-Za-z-]+)*))?(?:\+(?P<build>[0-9A-Za-z-]+(?:\.[0-9A-Za-z-]+)*))?
 6 | serialize = 
 7 | 	{major}.{minor}.{patch}-{release}+{build}
 8 | 	{major}.{minor}.{patch}+{build}
 9 | 	{major}.{minor}.{patch}-{release}
10 | 	{major}.{minor}.{patch}
11 | 
12 | [bumpversion:part:release]
13 | optional_value = production
14 | first_value = dev
15 | values = 
16 | 	dev
17 | 	production
18 | 
19 | [bumpverion:part:build]
20 | values = [0-9A-Za-z-]+
21 | 
22 | [bumpversion:file:setup.cfg]
23 | search = version = {current_version}
24 | replace = version = {new_version}
25 | 
26 | [bumpversion:file:src/biokeen/constants.py]
27 | search = VERSION = '{current_version}'
28 | replace = VERSION = '{new_version}'
29 | 
30 | [bumpversion:file:docs/source/conf.py]
31 | search = release = '{current_version}'
32 | replace = release = '{new_version}'
33 | 
34 | 


--------------------------------------------------------------------------------
/.flake8:
--------------------------------------------------------------------------------
 1 | #########################
 2 | # Flake8 Configuration  #
 3 | # (.flake8)             #
 4 | # (formerly in tox.ini) #
 5 | #########################
 6 | [flake8]
 7 | # flake8
 8 | ignore =
 9 |     # Line too long
10 |     E501
11 |     # Line break after binary operator
12 |     W504
13 | exclude =
14 |     .tox,
15 |     .git,
16 |     __pycache__,
17 |     docs/source/conf.py,
18 |     build,
19 |     dist,
20 |     *.pyc,
21 |     *.egg-info,
22 |     .cache,
23 |     .eggs
24 | format = ${cyan}%(path)s${reset}:${yellow_bold}%(row)d${reset}:${green_bold}%(col)d${reset}: ${red_bold}%(code)s${reset} %(text)s
25 | 
26 | # flake8-import-order
27 | application-import-names =
28 | 	biokeen
29 |     pykeen
30 |     pybel
31 |     bio2bel
32 | 	tests
33 | import-order-style = pycharm
34 | 
35 | # mccabe
36 | max-complexity = 10
37 | 


--------------------------------------------------------------------------------
/.gitignore:
--------------------------------------------------------------------------------
  1 | # Byte-compiled / optimized / DLL files
  2 | __pycache__/
  3 | *.py[cod]
  4 | *$py.class
  5 | 
  6 | # C extensions
  7 | *.so
  8 | 
  9 | # Distribution / packaging
 10 | .Python
 11 | env/
 12 | build/
 13 | develop-eggs/
 14 | dist/
 15 | downloads/
 16 | eggs/
 17 | .eggs/
 18 | lib/
 19 | lib64/
 20 | parts/
 21 | sdist/
 22 | var/
 23 | *.egg-info/
 24 | .installed.cfg
 25 | *.egg
 26 | 
 27 | # PyInstaller
 28 | #  Usually these files are written by a python script from a template
 29 | #  before PyInstaller builds the exe, so as to inject date/other infos into it.
 30 | *.manifest
 31 | *.spec
 32 | 
 33 | # Installer logs
 34 | pip-log.txt
 35 | pip-delete-this-directory.txt
 36 | 
 37 | # Unit test / coverage reports
 38 | htmlcov/
 39 | .tox/
 40 | .coverage
 41 | .coverage.*
 42 | .cache
 43 | nosetests.xml
 44 | coverage.xml
 45 | *,cover
 46 | .hypothesis/
 47 | 
 48 | # Translations
 49 | *.mo
 50 | *.pot
 51 | 
 52 | # Django stuff:
 53 | *.log
 54 | local_settings.py
 55 | 
 56 | # Flask stuff:
 57 | instance/
 58 | .webassets-cache
 59 | 
 60 | # Scrapy stuff:
 61 | .scrapy
 62 | 
 63 | # Sphinx documentation
 64 | docs/_build/
 65 | 
 66 | # PyBuilder
 67 | target/
 68 | 
 69 | # IPython Notebook
 70 | .ipynb_checkpoints
 71 | 
 72 | # pyenv
 73 | .python-version
 74 | 
 75 | # celery beat schedule file
 76 | celerybeat-schedule
 77 | 
 78 | # dotenv
 79 | .env
 80 | 
 81 | # virtualenv
 82 | venv/
 83 | ENV/
 84 | 
 85 | # Spyder project settings
 86 | .spyderproject
 87 | 
 88 | # Rope project settings
 89 | .ropeproject
 90 | 
 91 | # PyCharm project settings
 92 | .idea/*
 93 | 
 94 | *.pickle
 95 | *.gpickle
 96 | 
 97 | scratch
 98 | scratch/*
 99 | 
100 | .pytest_cache
101 | 
102 | ### OSX ###
103 | # General
104 | .DS_Store
105 | .AppleDouble
106 | .LSOverride
107 | 
108 | # Icon must end with two \r
109 | Icon
110 | 
111 | # Thumbnails
112 | ._*
113 | 
114 | # Files that might appear in the root of a volume
115 | .DocumentRevisions-V100
116 | .fseventsd
117 | .Spotlight-V100
118 | .TemporaryItems
119 | .Trashes
120 | .VolumeIcon.icns
121 | .com.apple.timemachine.donotpresent
122 | 
123 | # Directories potentially created on remote AFP share
124 | .AppleDB
125 | .AppleDesktop
126 | Network Trash Folder
127 | Temporary Items
128 | .apdisk
129 | 
130 | ### Vim ###
131 | # Swap
132 | [._]*.s[a-v][a-z]
133 | [._]*.sw[a-p]
134 | [._]s[a-rt-v][a-z]
135 | [._]ss[a-gi-z]
136 | [._]sw[a-p]
137 | 
138 | # Session
139 | Session.vim
140 | 
141 | # Temporary
142 | .netrwhist
143 | *~
144 | # Auto-generated tag files
145 | tags
146 | # Persistent undo
147 | [._]*.un~
148 | 
149 | .idea/
150 | data/entities_to_embeddings.pkl
151 | data/hyper_parameters.json
152 | data/relations_to_embeddings.pkl
153 | data/losses.json
154 | data/evaluation_summary.json
155 | /data/
156 | /data_local/
157 | /data_local/*


--------------------------------------------------------------------------------
/.readthedocs.yml:
--------------------------------------------------------------------------------
1 | # See: https://docs.readthedocs.io/en/latest/yaml-config.html
2 | build:
3 |   image: latest
4 | python:
5 |   version: 3.6
6 | requirements_file: requirements-rtd.txt
7 | 


--------------------------------------------------------------------------------
/.travis.yml:
--------------------------------------------------------------------------------
 1 | sudo: false
 2 | cache: pip
 3 | language: python
 4 | python:
 5 |   - 3.6
 6 | stages:
 7 |   - lint
 8 |   - docs
 9 |   - test
10 | jobs:
11 |   include:
12 |     # lint stage
13 |     - stage: lint
14 |       env: TOXENV=manifest
15 |     - env: TOXENV=flake8
16 |     - env: TOXENV=xenon
17 |     - env: TOXENV=pyroma
18 |     # docs stage
19 |     - stage: docs
20 |       env: TOXENV=doc8
21 |     - env: TOXENV=readme
22 |     - env: TOXENV=docs
23 |     - stage: test
24 |       env: TOXENV=py
25 |       python: "3.6"
26 |     - env: TOXENV=py
27 |       sudo: true
28 |       python: "3.7"
29 |       dist: xenial
30 |   allow_failures:
31 |     - env: TOXENV=xenon
32 | install:
33 |   - sh -c 'if [ "$TOXENV" = "py" ]; then pip install tox codecov coverage; else pip install tox; fi'
34 | script:
35 |   - tox
36 | after_success:
37 |   - sh -c 'if [ "$TOXENV" = "py" ]; then tox -e coverage-report; codecov; fi'
38 | 


--------------------------------------------------------------------------------
/CONTRIBUTING.rst:
--------------------------------------------------------------------------------
 1 | Contributing
 2 | ============
 3 | BioKEEN is part of the KEEN Universe with which we aim to make knowledge graph embedding
 4 | research reproducible and transferable to other domains. Contributions, whether big or small, are appreciated!
 5 | You can get involved by submitting an issue, making a suggestion, or adding code to the project.
 6 | 
 7 | 
 8 | Having a Problem? Submit an Issue.
 9 | ----------------------------------
10 | 
11 | 1. Check that you have the latest version of :code:`BioKEEN`
12 | 2. Go here: https://github.com/SmartDataAnalytics/BioKEEN/issues
13 | 3. Check that this issue hasn't been solved
14 | 4. Click "new issue"
15 | 5. Add a short, but descriptive title
16 | 6. Add a full description of the problem, including the code that caused it and any support files related to this code
17 |    so others can reproduce your problem
18 | 7. Copy the output and error message you're getting
19 | 
20 | Have a Question or Suggestion?
21 | ------------------------------
22 | 
23 | Same drill! Submit an issue and we'll have a nice conversation in the thread.
24 | 
25 | Want to Contribute?
26 | -------------------
27 | 
28 | 1. Get the code. Fork the repository from GitHub using the big button in the top-right corner of https://github.com/SmartDataAnalytics/BioKEEN
29 | 
30 | 2. Clone your directory with
31 | 
32 |     $ git clone https://github.com/<YourUsername>/SmartDataAnalytics/BioKEEN.git
33 | 
34 | 3. Install with :code:`pip`. The flag, :code:`-e`, makes your installation editable, so your changes will be reflected
35 |    automatically in your installation.
36 | 
37 |     $ cd biokeen
38 |     $ python3 -m pip install -e .
39 | 
40 | 4. Make a branch off of develop, then make contributions! This line makes a new branch and checks it out
41 | 
42 |     $ git checkout -b feature/<YourFeatureName>
43 | 
44 | 5. This project should be well tested, so write unit tests in the :code:`tests/` directory
45 | 6. Check that all tests are passing and code coverage is good with :code:`tox` before committing.
46 | 
47 |     $ tox
48 | 
49 | 
50 | Pull Requests
51 | -------------
52 | 
53 | Once you've got your feature or bugfix finished (or if its in a partially complete state but you want to publish it
54 | for comment), push it to your fork of the repository and open a pull request against the develop branch on GitHub.
55 | 
56 | Make a descriptive comment about your pull request, perhaps referencing the issue it is meant to fix
57 | (something along the lines of "fixes issue #10" will cause GitHub to automatically link to that issue).
58 | The maintainers will review your pull request and perhaps make comments about it, request changes, or may pull it in
59 | to the develop branch! If you need to make changes to your pull request, simply push more commits to the feature branch
60 | in your fork to GitHub and they will automatically be added to the pull. You do not need to close and reissue your
61 | pull request to make changes!
62 | 
63 | 
64 | 


--------------------------------------------------------------------------------
/LICENSE:
--------------------------------------------------------------------------------
 1 | MIT License
 2 | 
 3 | Copyright (c) 2018 Smart Data Analytics
 4 | 
 5 | Permission is hereby granted, free of charge, to any person obtaining a copy
 6 | of this software and associated documentation files (the "Software"), to deal
 7 | in the Software without restriction, including without limitation the rights
 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 9 | copies of the Software, and to permit persons to whom the Software is
10 | furnished to do so, subject to the following conditions:
11 | 
12 | The above copyright notice and this permission notice shall be included in all
13 | copies or substantial portions of the Software.
14 | 
15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 | SOFTWARE.
22 | 


--------------------------------------------------------------------------------
/MANIFEST.in:
--------------------------------------------------------------------------------
 1 | graft src
 2 | graft tests
 3 | prune notebooks
 4 | prune data
 5 | 
 6 | recursive-include docs/source *.py
 7 | recursive-include docs/source *.rst
 8 | recursive-include docs/source *.png
 9 | include docs/Makefile
10 | include config/config.ini
11 | 
12 | global-exclude *.py[cod] __pycache__ *.so *.dylib .DS_Store *.gpickle
13 | 
14 | exclude .bumpversion.cfg data .travis.yml .readthedocs.yml .flake8 requirements-rtd.txt
15 | include *.rst LICENSE tox.ini *.tsv
16 | 


--------------------------------------------------------------------------------
/README.rst:
--------------------------------------------------------------------------------
  1 | WARNING: This repository is superceded by `PyKEEN <https://github.com/pykeen/pykeen>`_ and is therefore no longer maintained.
  2 | =============================================================================================================================
  3 | BioKEEN 
  4 | -------
  5 | |build| |coverage| |docs| |zenodo|
  6 | 
  7 | BioKEEN (Biological KnowlEdge EmbeddiNgs) is a package for training and evaluating biological knowledge graph
  8 | embeddings built on .
  9 | 
 10 | Because we use PyKEEN as the underlying software package, implementations of 10 knowledge graph embedding models are
 11 | currently available for BioKEEN. Furthermore, BioKEEN can be run in *training mode* in which users provide their own set
 12 | of hyper-parameter values, or in *hyper-parameter optimization mode* to find suitable hyper-parameter values from set
 13 | of user defined values.
 14 | 
 15 | Through the integration of the `Bio2BEL <https://github.com/bio2bel>`_ [2]_ software numerous biomedical databases are
 16 | directly accessible within BioKEEN.
 17 | 
 18 | BioKEEN can also be run without having experience in programing by using its interactive command line interface that can
 19 | be started with the command “biokeen” from a terminal.
 20 | 
 21 | Share Your Experimental Artifacts
 22 | ---------------------------------
 23 | You can share you trained KGE models along the other experimental artifacts through the `KEEN-Model-Zoo <https://github.com/SmartDataAnalytics/KEEN-Model-Zoo>`_.
 24 | 
 25 | Tutorials
 26 | ---------
 27 | A brief tutorial on how to get started with BioKEEN is available `here <https://vimeo.com/314252656>`_.
 28 | 
 29 | .. image:: https://i.vimeocdn.com/video/755767182.jpg?mw=1100&mh=619&q=70
 30 |     :width: 300px
 31 |     :target: https://vimeo.com/314252656
 32 | 
 33 | 
 34 | Further tutorials are can be found in the `notebooks directory <https://github.com/SmartDataAnalytics/BioKEEN/tree/master/notebooks>`_ and in our `documentation <https://biokeen.readthedocs.io/en/latest/>`_.
 35 | 
 36 | Citation
 37 | --------
 38 | If you find BioKEEN useful in your work, please consider citing:
 39 | 
 40 | .. [1] Ali, M., *et al.* (2019). `BioKEEN: A library for learning and evaluating biological knowledge graph embeddings
 41 |        <https://academic.oup.com/bioinformatics/advance-article/doi/10.1093/bioinformatics/btz117/5320556>`_. *Bioinformatics*, btz117.
 42 | 
 43 | **Note**: ComPath has been updated, for this reason we have uploaded the dataset version that we have used for
 44 | our experiments: `dataset <https://github.com/SmartDataAnalytics/KEEN-Model-Zoo/blob/master/bioinformatics/ComPath/compath.keen.tsv>`_
 45 | 
 46 | Installation |pypi_version| |python_versions| |pypi_license|
 47 | ------------------------------------------------------------
 48 | To install biokeen, Python 3.6+ is required, and we recommend to install it on Linux or Mac OS systems.
 49 | Please run following command:
 50 | 
 51 | .. code-block:: sh
 52 | 
 53 |     $ pip install git+https://github.com/SmartDataAnalytics/BioKEEN.git
 54 | 
 55 | Alternatively, it can be installed from the source for development with:
 56 | 
 57 | .. code-block:: sh
 58 | 
 59 |     $ git clone https://github.com/SmartDataAnalytics/BioKEEN.git biokeen
 60 |     $ cd biokeen
 61 |     $ pip install -e .
 62 | 
 63 | Contributing
 64 | ------------
 65 | Contributions, whether filing an issue, making a pull request, or forking, are appreciated.
 66 | See `CONTRIBUTING.rst <https://github.com/SmartDataAnalytics/BioKEEN/blob/master/CONTRIBUTING.rst>`_ for more
 67 | information on getting involved.
 68 | 
 69 | CLI Usage
 70 | ---------
 71 | To show BioKEEN's available commands, please run following command:
 72 | 
 73 | .. code-block:: sh
 74 | 
 75 |     biokeen
 76 | 
 77 | Starting the Training/HPO Pipeline - Set Up Your Experiment within 60 seconds
 78 | *****************************************************************************
 79 | To configure an experiment via the CLI, please run following command:
 80 | 
 81 | .. code-block:: sh
 82 | 
 83 |     biokeen start
 84 | 
 85 | To start BioKEEN with an existing configuration file, please run the following command:
 86 | 
 87 | .. code-block:: sh
 88 | 
 89 |     biokeen start -f /path/to/config.json
 90 | 
 91 | Starting the Prediction Pipeline
 92 | ********************************
 93 | To make prediction based on a trained model, please run following command:
 94 | 
 95 | .. code-block:: sh
 96 | 
 97 |     biokeen predict -m /path/to/model/directory -d /path/to/data/directory
 98 | 
 99 | where the value for the argument **-m** is the directory containing the model, in more detail following files must be
100 | contained in the directory:
101 | 
102 | * configuration.json
103 | * entities_to_embeddings.json
104 | * relations_to_embeddings.json
105 | * trained_model.pkl
106 | 
107 | These files are created automatically created after model is trained (and evaluated) and exported in your
108 | specified output directory.
109 | 
110 | The value for the argument **-d** is the directory containing the data for which inference should be applied, and it
111 | needs to contain following files:
112 | 
113 | * entities.tsv
114 | * relations.tsv
115 | 
116 | where *entities.tsv* contains all entities of interest, and relations.tsv all relations. Both files should contain
117 | should contain a single column containing all the entities/relations. Based on these files, PyKEEN will create all
118 | triple permutations, and computes the predictions for them, and saves them in data directory
119 | in *predictions.tsv*.
120 | 
121 | Summarize the Results of All Experiments
122 | ****************************************
123 | To summarize the results of all experiments, please run following command:
124 | 
125 | .. code-block:: sh
126 | 
127 |     biokeen summarize -d /path/to/experiments/directory -o /path/to/output/file.csv
128 | 
129 | Getting Bio2BEL Data
130 | ********************
131 | To download and structure the data from a `Bio2BEL <https://github.com/bio2bel>`_ repository, run:
132 | 
133 | .. code-block:: sh
134 | 
135 |     biokeen data get <name>
136 | 
137 | Where ``<name>`` can be any repository name in Bio2BEL such as ``hippie``, ``mirtarbase``.
138 | 
139 | References
140 | ----------
141 | 
142 | .. [2] Hoyt, C., *et al.* (2019). `Integration of Structured Biological Data Sources using Biological Expression Language
143 |     <https://doi.org/10.1101/631812>`_. *bioRxiv*, 631812.
144 | 
145 | .. |build| image:: https://travis-ci.org/SmartDataAnalytics/BioKEEN.svg?branch=master
146 |     :target: https://travis-ci.org/SmartDataAnalytics/BioKEEN
147 | 
148 | .. |zenodo| image:: https://zenodo.org/badge/150270965.svg
149 |     :target: https://zenodo.org/badge/latestdoi/150270965
150 | 
151 | .. |docs| image:: http://readthedocs.org/projects/biokeen/badge/?version=latest
152 |     :target: https://biokeen.readthedocs.io/en/latest/
153 |     :alt: Documentation Status
154 | 
155 | .. |python_versions| image:: https://img.shields.io/pypi/pyversions/biokeen.svg
156 |     :alt: Stable Supported Python Versions
157 | 
158 | .. |pypi_version| image:: https://img.shields.io/pypi/v/biokeen.svg
159 |     :alt: Current version on PyPI
160 | 
161 | .. |pypi_license| image:: https://img.shields.io/pypi/l/biokeen.svg
162 |     :alt: MIT License
163 | 
164 | .. |coverage| image:: https://codecov.io/gh/SmartDataAnalytics/BioKEEN/branch/master/graphs/badge.svg
165 |     :target: https://codecov.io/gh/SmartDataAnalytics/BioKEEN
166 |     :alt: Coverage Status on CodeCov
167 | 


--------------------------------------------------------------------------------
/docs/Makefile:
--------------------------------------------------------------------------------
 1 | # Minimal makefile for Sphinx documentation
 2 | #
 3 | 
 4 | # You can set these variables from the command line.
 5 | SPHINXOPTS    =
 6 | SPHINXBUILD   = sphinx-build
 7 | SOURCEDIR     = source
 8 | BUILDDIR      = build
 9 | 
10 | # Put it first so that "make" without argument is like "make help".
11 | help:
12 | 	@$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
13 | 
14 | .PHONY: help Makefile
15 | 
16 | # Catch-all target: route all unknown targets to Sphinx using the new
17 | # "make mode" option.  $(O) is meant as a shortcut for $(SPHINXOPTS).
18 | %: Makefile
19 | 	@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
20 | 


--------------------------------------------------------------------------------
/docs/source/bio2bel_repositories.rst:
--------------------------------------------------------------------------------
 1 | Biological Databases
 2 | ====================
 3 | The following biological databases can be used for training and evaluating knowledge graph embeddings. This is done
 4 | by using the `Bio2BEL <https://github.com/bio2bel>`_ universe.
 5 | 
 6 | +------------------------+-----------------------+
 7 | | Source                 | DOI                   |
 8 | |                        |                       |
 9 | +========================+=======================+
10 | | ADEPTUS                | |adeptus_zenodo|      |
11 | +------------------------+-----------------------+
12 | | ComPath                | |compath_zenodo|      |
13 | +------------------------+-----------------------+
14 | | DrugBank               | |drugbank_zenodo|     |
15 | +------------------------+-----------------------+
16 | | ExPASy                 | |expasy_zenodo|       |
17 | +------------------------+-----------------------+
18 | | HIPPIE                 | |hippie_zenodo|       |
19 | +------------------------+-----------------------+
20 | | HSDN                   | |hsdn_zenodo|         |
21 | +------------------------+-----------------------+
22 | | KEGG                   | |kegg_zenodo|         |
23 | +------------------------+-----------------------+
24 | | miRTarBase             | |mirtarbase_zenodo|   |
25 | +------------------------+-----------------------+
26 | | MSigDB                 | |msig_zenodo|         |
27 | +------------------------+-----------------------+
28 | | Reactome               | |reactome_zenodo|     |
29 | +------------------------+-----------------------+
30 | | SIDER                  | |sider_zenodo|        |
31 | +------------------------+-----------------------+
32 | | InterPro               | |interpro_zenodo|     |
33 | +------------------------+-----------------------+
34 | | WikiPathways           | |wikipathways_zenodo| |
35 | +------------------------+-----------------------+
36 | 
37 | 
38 | .. |interpro_zenodo| image:: https://zenodo.org/badge/98345182.svg
39 |     :target: https://zenodo.org/badge/latestdoi/98345182
40 |     :alt: InterPro Zenodo DOI
41 | 
42 | .. |hsdn_zenodo| image:: https://zenodo.org/badge/158366852.svg
43 |    :target: https://zenodo.org/badge/latestdoi/158366852
44 |     :alt: HSDN Zenodo DOI
45 | 
46 | .. |sider_zenodo| image:: https://zenodo.org/badge/129140922.svg
47 |     :target: https://zenodo.org/badge/latestdoi/129140922
48 |     :alt: SIDER Zenodo DOI
49 | 
50 | .. |expasy_zenodo| image:: https://zenodo.org/badge/100023822.svg
51 |     :target: https://zenodo.org/badge/latestdoi/100023822
52 |     :alt: ExPASy Zenodo DOI
53 | 
54 | .. |adeptus_zenodo| image:: https://zenodo.org/badge/158358036.svg
55 |    :target: https://zenodo.org/badge/latestdoi/158358036
56 |     :alt: ADEPTUS Zenodo DOI
57 | 
58 | .. |kegg_zenodo| image:: https://zenodo.org/badge/105248163.svg
59 |     :target: https://zenodo.org/badge/latestdoi/105248163
60 |     :alt: KEGG Zenodo DOI
61 | 
62 | .. |compath_zenodo| image:: https://zenodo.org/badge/132792765.svg
63 |    :target: https://zenodo.org/badge/latestdoi/132792765
64 | 
65 | .. |reactome_zenodo| image:: https://zenodo.org/badge/103138323.svg
66 |     :target: https://zenodo.org/badge/latestdoi/103138323
67 |     :alt: Reactome Zenodo DOI
68 | 
69 | .. |wikipathways_zenodo| image:: https://zenodo.org/badge/118924155.svg
70 |     :target: https://zenodo.org/badge/latestdoi/118924155
71 |     :alt: WikiPathways Zenodo DOI
72 | 
73 | .. |msig_zenodo| image:: https://zenodo.org/badge/123948554.svg
74 |     :target: https://zenodo.org/badge/latestdoi/123948554
75 |     :alt: MSigDB Zenodo DOI
76 | 
77 | .. |drugbank_zenodo| image:: https://zenodo.org/badge/DOI/10.5281/zenodo.1243727.svg
78 |     :target: https://doi.org/10.5281/zenodo.1243727
79 |     :alt: Zenodo DOI
80 | 
81 | .. |hippie_zenodo| image:: https://zenodo.org/badge/DOI/10.5281/zenodo.1435930.svg
82 |     :target: https://doi.org/10.5281/zenodo.1435930
83 |     :alt: HIPPIE Zenodo DOI
84 | 
85 | .. |mirtarbase_zenodo| image:: https://zenodo.org/badge/95350968.svg
86 |     :target: https://zenodo.org/badge/latestdoi/95350968
87 |     :alt: Zenodo DOI
88 | 


--------------------------------------------------------------------------------
/docs/source/cli/inference.rst:
--------------------------------------------------------------------------------
 1 | Perform Inference
 2 | =================
 3 | Starting the Prediction Pipeline
 4 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 5 | .. code-block:: sh
 6 | 
 7 |    biokeen predict -m /path/to/model/directory -d /path/to/data/directory
 8 | 
 9 | where the value for the argument **-m** is the directory containing the model, in more detail following files must be
10 | contained in the directory:
11 | 
12 | * configuration.json
13 | * entities_to_embeddings.json
14 | * relations_to_embeddings.json
15 | * trained_model.pkl
16 | 
17 | These files are created automatically created when an experiment is configured through the CLI.
18 | 
19 | The value for the argument **-d** is the directory containing the data for which inference should be applied, and it
20 | needs to contain following files:
21 | 
22 | * entities.tsv
23 | * relations.tsv
24 | 
25 | where *entities.tsv* contains all entities of interest, and relations.tsv all relations. PyKEEN will create all possible
26 | combinations of triples, and computes the predictions for them, and saves them in data directory in *predictions.tsv*.
27 | 
28 | Optionally, a set of triples can be provided that should be exluded from the prediction, e.g. all the triples
29 | contained in the training set:
30 | 
31 | .. code-block:: sh
32 | 
33 |    pykeen-predict -m /path/to/model/directory -d /path/to/data/directory -t /path/to/triples.tsv
34 | 
35 | Hence, it is easily possible to compute plausibility scores forr all triples that are not contained in the training set.
36 | 
37 | CLI Manual
38 | ~~~~~~~~~~
39 | .. click:: biokeen.cli.cli:predict
40 |    :prog: biokeen predict
41 |    :show-nested:
42 | 


--------------------------------------------------------------------------------
/docs/source/cli/summarize.rst:
--------------------------------------------------------------------------------
 1 | Summarize all Experiments
 2 | =========================
 3 | Here, we describe how to summarize all experiments into a single csv-file.
 4 | To get the summary, please provide the path to parent directory containing all the experiments as sub-directories,
 5 | and the path to the output file:
 6 | 
 7 | .. code-block:: python
 8 | 
 9 |     biokeen summarize -d /path/to/experiments/directory -o /path/to/output/file.csv
10 | 


--------------------------------------------------------------------------------
/docs/source/cli/train_and_evaluate.rst:
--------------------------------------------------------------------------------
 1 | Train and Evaluate
 2 | ==================
 3 | Tutorial
 4 | --------
 5 | Step 1: Start CLI
 6 | ~~~~~~~~~~~~~~~~~
 7 | .. code-block:: sh
 8 | 
 9 |    biokeen start
10 | 
11 | Step 2: Select data source
12 | ~~~~~~~~~~~~~~~~~~~~~~~~~~
13 | .. image:: ../images/data_source.png
14 | 
15 | Step 3: Select database
16 | ~~~~~~~~~~~~~~~~~~~~~~~
17 | .. image:: ../images/select_database.png
18 | 
19 | Step 4: Specify execution mode
20 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
21 | .. image:: ../images/execution_mode.png
22 | 
23 | Step 5: Select KGE model
24 | ~~~~~~~~~~~~~~~~~~~~~~~~
25 | .. image:: ../images/select_model.png
26 | 
27 | Step 6: Specify model dependent hyper-parameters
28 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
29 | 
30 | Step 7: Specify the batch-size
31 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
32 | .. image:: ../images/batch_size.png
33 | 
34 | Step 8: Specify the number of training epochs
35 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
36 | .. image:: ../images/epochs.png
37 | 
38 | Step 9: Specify whether to evaluate the model
39 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
40 | .. image:: ../images/epochs.png
41 | 
42 | Step 10: Specify whether to evaluate the model
43 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
44 | .. image:: ../images/epochs.png
45 | 
46 | Step 11: Provide a random seed
47 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
48 | .. image:: ../images/random_seed.png
49 | 
50 | Step 12: Specify preferred device
51 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
52 | .. image:: ../images/preferred_device.png
53 | 
54 | Step 13: Specify the path to the output directory
55 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
56 | .. image:: ../images/output_directory.png
57 | 
58 | Reference
59 | ---------
60 | .. click:: biokeen.cli.cli:start
61 |    :prog: biokeen
62 |    :show-nested:
63 | 


--------------------------------------------------------------------------------
/docs/source/conf.py:
--------------------------------------------------------------------------------
  1 | # -*- coding: utf-8 -*-
  2 | #
  3 | # Configuration file for the Sphinx documentation builder.
  4 | #
  5 | # This file does only contain a selection of the most common options. For a
  6 | # full list see the documentation:
  7 | # http://www.sphinx-doc.org/en/master/config
  8 | 
  9 | # -- Path setup --------------------------------------------------------------
 10 | 
 11 | # If extensions (or modules to document with autodoc) are in another directory,
 12 | # add these directories to sys.path here. If the directory is relative to the
 13 | # documentation root, use os.path.abspath to make it absolute, like shown here.
 14 | #
 15 | import os
 16 | import re
 17 | import sys
 18 | 
 19 | # -- Mockup PyTorch to exclude it while compiling the docs--------------------------------------------------------------
 20 | autodoc_mock_imports = ['pykeen']
 21 | 
 22 | sys.path.insert(0, os.path.abspath('../../src'))
 23 | 
 24 | # -- Project information -----------------------------------------------------
 25 | 
 26 | project = 'BioKEEN'
 27 | copyright = '2018, Mehdi Ali, Charles Tapley Hoyt, and Daniel Domingo-Fernández'
 28 | author = 'Mehdi Ali, Charles Tapley Hoyt, and Daniel Domingo-Fernández'
 29 | 
 30 | # The full version, including alpha/beta/rc tags.
 31 | release = '0.0.15-dev'
 32 | 
 33 | # The short X.Y version.
 34 | parsed_version = re.match(
 35 |     '(?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)(?:-(?P<release>[0-9A-Za-z-]+(?:\.[0-9A-Za-z-]+)*))?(?:\+(?P<build>[0-9A-Za-z-]+(?:\.[0-9A-Za-z-]+)*))?',
 36 |     release
 37 | )
 38 | version = parsed_version.expand('\g<major>.\g<minor>.\g<patch>')
 39 | 
 40 | if parsed_version.group('release'):
 41 |     tags.add('prerelease')
 42 | 
 43 | # -- General configuration ---------------------------------------------------
 44 | 
 45 | # If your documentation needs a minimal Sphinx version, state it here.
 46 | #
 47 | # needs_sphinx = '1.0'
 48 | 
 49 | # Add any Sphinx extension module names here, as strings. They can be
 50 | # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom
 51 | # ones.
 52 | extensions = [
 53 |     'sphinx.ext.autodoc',
 54 |     'sphinx.ext.intersphinx',
 55 |     'sphinx.ext.coverage',
 56 |     'sphinx.ext.viewcode',
 57 |     'sphinx_autodoc_typehints',
 58 |     'sphinx_click.ext',
 59 | ]
 60 | 
 61 | # Add any paths that contain templates here, relative to this directory.
 62 | # templates_path = ['_templates']
 63 | 
 64 | # The suffix(es) of source filenames.
 65 | # You can specify multiple suffix as a list of string:
 66 | #
 67 | # source_suffix = ['.rst', '.md']
 68 | source_suffix = '.rst'
 69 | 
 70 | # The master toctree document.
 71 | master_doc = 'index'
 72 | 
 73 | # The language for content autogenerated by Sphinx. Refer to documentation
 74 | # for a list of supported languages.
 75 | #
 76 | # This is also used if you do content translation via gettext catalogs.
 77 | # Usually you set "language" from the command line for these cases.
 78 | language = None
 79 | 
 80 | # List of patterns, relative to source directory, that match files and
 81 | # directories to ignore when looking for source files.
 82 | # This pattern also affects html_static_path and html_extra_path.
 83 | exclude_patterns = []
 84 | 
 85 | # The name of the Pygments (syntax highlighting) style to use.
 86 | pygments_style = None
 87 | 
 88 | # -- Options for HTML output -------------------------------------------------
 89 | 
 90 | # The theme to use for HTML and HTML Help pages.  See the documentation for
 91 | # a list of builtin themes.
 92 | #
 93 | html_theme = 'sphinx_rtd_theme'
 94 | 
 95 | # Theme options are theme-specific and customize the look and feel of a theme
 96 | # further.  For a list of options available for each theme, see the
 97 | # documentation.
 98 | #
 99 | # html_theme_options = {}
100 | 
101 | # Add any paths that contain custom static files (such as style sheets) here,
102 | # relative to this directory. They are copied after the builtin static files,
103 | # so a file named "default.css" will overwrite the builtin "default.css".
104 | # html_static_path = ['_static']
105 | 
106 | # Custom sidebar templates, must be a dictionary that maps document names
107 | # to template names.
108 | #
109 | # The default sidebars (for documents that don't match any pattern) are
110 | # defined by theme itself.  Builtin themes are using these templates by
111 | # default: ``['localtoc.html', 'relations.html', 'sourcelink.html',
112 | # 'searchbox.html']``.
113 | #
114 | # html_sidebars = {}
115 | 
116 | 
117 | # -- Options for HTMLHelp output ---------------------------------------------
118 | 
119 | # Output file base name for HTML help builder.
120 | htmlhelp_basename = 'BioKEENdoc'
121 | 
122 | # -- Options for LaTeX output ------------------------------------------------
123 | 
124 | latex_elements = {
125 |     # The paper size ('letterpaper' or 'a4paper').
126 |     #
127 |     # 'papersize': 'letterpaper',
128 | 
129 |     # The font size ('10pt', '11pt' or '12pt').
130 |     #
131 |     # 'pointsize': '10pt',
132 | 
133 |     # Additional stuff for the LaTeX preamble.
134 |     #
135 |     # 'preamble': '',
136 | 
137 |     # Latex figure (float) alignment
138 |     #
139 |     # 'figure_align': 'htbp',
140 | }
141 | 
142 | # Grouping the document tree into LaTeX files. List of tuples
143 | # (source start file, target name, title,
144 | #  author, documentclass [howto, manual, or own class]).
145 | latex_documents = [
146 |     (master_doc, 'BioKEEN.tex', 'BioKEEN Documentation',
147 |      'Mehdi Ali, Charles Tapley Hoyt, and Daniel Domingo-Fernández', 'manual'),
148 | ]
149 | 
150 | # -- Options for manual page output ------------------------------------------
151 | 
152 | # One entry per manual page. List of tuples
153 | # (source start file, name, description, authors, manual section).
154 | man_pages = [
155 |     (master_doc, 'biokeen', 'BioKEEN Documentation',
156 |      [author], 1)
157 | ]
158 | 
159 | # -- Options for Texinfo output ----------------------------------------------
160 | 
161 | # Grouping the document tree into Texinfo files. List of tuples
162 | # (source start file, target name, title, author,
163 | #  dir menu entry, description, category)
164 | texinfo_documents = [
165 |     (master_doc, 'BioKEEN', 'BioKEEN Documentation',
166 |      author, 'BioKEEN', 'One line description of project.',
167 |      'Miscellaneous'),
168 | ]
169 | 
170 | # -- Options for Epub output -------------------------------------------------
171 | 
172 | # Bibliographic Dublin Core info.
173 | epub_title = project
174 | 
175 | # The unique identifier of the text. This can be a ISBN number
176 | # or the project homepage.
177 | #
178 | # epub_identifier = ''
179 | 
180 | # A unique identification for the text.
181 | #
182 | # epub_uid = ''
183 | 
184 | # A list of files that should not be packed into the epub file.
185 | epub_exclude_files = ['search.html']
186 | 
187 | # -- Extension configuration -------------------------------------------------
188 | 
189 | # -- Options for intersphinx extension ---------------------------------------
190 | 
191 | # Example configuration for intersphinx: refer to the Python standard library.
192 | intersphinx_mapping = {
193 |     'https://docs.python.org/3': None,
194 | }
195 | 


--------------------------------------------------------------------------------
/docs/source/convert.rst:
--------------------------------------------------------------------------------
1 | Handling BEL
2 | ============
3 | .. autofunction:: biokeen.convert.to_pykeen_path
4 | .. autofunction:: biokeen.convert.to_pykeen_df
5 | 


--------------------------------------------------------------------------------
/docs/source/hyper_parameter_optimization.rst:
--------------------------------------------------------------------------------
 1 | Apply a Hyper-Parameter Optimization
 2 | ====================================
 3 | 
 4 | Here, we describe how to define an experiment that should perform a hyper-parameter optimization mode.
 5 | 
 6 | 
 7 | Configure your experiment
 8 | ~~~~~~~~~~~~~~~~~~~~~~~~~
 9 | To run experiments programmatically, the core software library PyKEEN should be used. To run PyKEEN in
10 | hyper-parameter optimization (HPO) mode, please set **execution_mode**  to **HPO_mode**.In HPO mode several values
11 | can be provided for the hyper-parameters from which different settings will be tested based
12 | on the hyper-parameter optimization algorithm. The possible values for a single hyper-parameter need to be provided as
13 | a list. The **maximum_number_of_hpo_iters** defines how many HPO iterations should be performed.
14 | 
15 | .. code-block:: python
16 | 
17 |     config = dict(
18 |         training_set_path           = 'data/corpora/compath.tsv',
19 |         test_set_ratio              = 0.1,
20 |         execution_mode              = 'HPO_mode',
21 |         kg_embedding_model_name     = 'TransE',
22 |         embedding_dim               = [50,100,150]
23 |         normalization_of_entities   =  2,  # corresponds to L2
24 |         scoring_function            = [1,2],  # corresponds to L1
25 |         margin_loss                 = [1,1.5,2],
26 |         learning_rate               = [0.1,0.01],
27 |         batch_size                  = [32,128],
28 |         num_epochs                  = 1000,
29 |         maximum_number_of_hpo_iters = 5,
30 |         filter_negative_triples     = True,
31 |         random_seed                 = 2,
32 |         preferred_device            = 'cpu',
33 |     )
34 | 
35 | 
36 | Run your experiment
37 | ~~~~~~~~~~~~~~~~~~~
38 | The experiment will be started with the *run* function, and in the output directory the exported results will be saved.
39 | 
40 | .. code-block:: python
41 | 
42 |     results = pykeen.run(
43 |         config=config,
44 |         output_directory=output_directory,
45 |     )
46 | 
47 | Access your results
48 | ~~~~~~~~~~~~~~~~~~~
49 | Show all keys contained in ``results``:
50 | 
51 | .. code-block:: python
52 | 
53 |     print('Keys:', *sorted(results.results.keys()), sep='\n  ')
54 | 
55 | 
56 | Access trained KGE model
57 | ~~~~~~~~~~~~~~~~~~~~~~~~
58 | .. code-block:: python
59 | 
60 |     results.results['trained_model']
61 | 
62 | Access the losses
63 | ~~~~~~~~~~~~~~~~~~
64 | 
65 | .. code-block:: python
66 | 
67 |     results.results['losses']
68 | 
69 | Access evaluation results
70 | ~~~~~~~~~~~~~~~~~~~~~~~~~
71 | 
72 | .. code-block:: python
73 | 
74 |     results.results['eval_summary']
75 | 
76 | 


--------------------------------------------------------------------------------
/docs/source/images/batch_size.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/batch_size.png


--------------------------------------------------------------------------------
/docs/source/images/data_source.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/data_source.png


--------------------------------------------------------------------------------
/docs/source/images/epochs.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/epochs.png


--------------------------------------------------------------------------------
/docs/source/images/execution_mode.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/execution_mode.png


--------------------------------------------------------------------------------
/docs/source/images/output_directory.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/output_directory.png


--------------------------------------------------------------------------------
/docs/source/images/preferred_device.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/preferred_device.png


--------------------------------------------------------------------------------
/docs/source/images/provide_dataset.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/provide_dataset.png


--------------------------------------------------------------------------------
/docs/source/images/random_seed.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/random_seed.png


--------------------------------------------------------------------------------
/docs/source/images/select_database.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/select_database.png


--------------------------------------------------------------------------------
/docs/source/images/select_model.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/SmartDataAnalytics/BioKEEN/9d401e7a22f5ae5fe3aa57e4bf3f8bc1dfa812bc/docs/source/images/select_model.png


--------------------------------------------------------------------------------
/docs/source/index.rst:
--------------------------------------------------------------------------------
 1 | BioKEEN
 2 | =======
 3 | BioKEEN (Biological KnowlEdge EmbeddiNgs) is a package for training and evaluating **biological** knowledge graph
 4 | embeddings built on `PyKEEN <https://github.com/SmartDataAnalytics/PyKEEN>`_. Within BioKEEN several biomedical
 5 | databases are directly accessible for training and evaluating biological knowledge graph embeddings
 6 | (see :ref:`bio2bel_repositories`).
 7 | 
 8 | Because we use PyKEEN as the core underlying framework, currently, implementations of 10
 9 | knowledge graph emebddings models are avaialble for BioKEEN. Furthermore, it can be run in training mode in
10 | which users provide their own set of hyper-parameter values, or in hyper-parameter optimization mode to find suitable
11 | hyper-parameter values from set of user defined values. BioKEEN can also be run without having experience in programing
12 | by using its interactive command line interface that can be started with the command "biokeen" from a terminal.
13 | 
14 | Installation is as easy as getting the code from `PyPI <https://pypi.python.org/pypi/biokeen>`_ with
15 | :code:`python3 -m pip install biokeen`.
16 | 
17 | Citation
18 | --------
19 | If you use BioKEEN in your work, please cite [1]_:
20 | 
21 | .. [1] Ali, M., *et al.* (2018). `BioKEEN: A library for learning and evaluating biological knowledge graph embeddings
22 |        <https://zenodo.org/badge/latestdoi/136345023>`_.
23 | 
24 | .. toctree::
25 |    :maxdepth: 2
26 |    :caption: Getting Started
27 |    :name: start
28 | 
29 |    installation
30 | 
31 | .. toctree::
32 |    :maxdepth: 2
33 |    :caption: CLI Usage
34 |    :name: cli
35 | 
36 |    cli/train_and_evaluate
37 |    cli/inference
38 |    cli/summarize
39 | 
40 | .. toctree::
41 |    :maxdepth: 2
42 |    :caption: Running PyKEEN programmatically
43 |    :name: prog
44 | 
45 |    train_and_evaluate
46 |    hyper_parameter_optimization
47 | 
48 | 
49 | .. toctree::
50 |    :maxdepth: 2
51 |    :caption: Reference
52 |    :name: reference
53 | 
54 |    convert
55 | 
56 | .. toctree::
57 |    :maxdepth: 2
58 |    :caption: Biological Databases
59 |    :name: bio2bel_repositories
60 | 
61 |    bio2bel_repositories
62 | 
63 | Indices and tables
64 | ==================
65 | * :ref:`genindex`
66 | * :ref:`modindex`
67 | * :ref:`search`
68 | 


--------------------------------------------------------------------------------
/docs/source/installation.rst:
--------------------------------------------------------------------------------
 1 | Installation
 2 | ============
 3 | There are several ways to download and install BioKEEN.
 4 | 
 5 | .. warning:: BioKEEN requires Python 3.6+
 6 | 
 7 | Easiest
 8 | ~~~~~~~
 9 | Download the latest stable code from `PyPI <https://pypi.python.org/pypi/biokeen>`_ with:
10 | 
11 | .. code-block:: sh
12 | 
13 |    $ pip install biokeen
14 | 
15 | Get the Latest
16 | ~~~~~~~~~~~~~~~
17 | Download the most recent code from `GitHub <https://github.com/SmartDataAnalytics/BioKEEN>`_ with:
18 | 
19 | .. code-block:: sh
20 | 
21 |    $ pip install git+https://github.com/SmartDataAnalytics/BioKEEN.git
22 | 


--------------------------------------------------------------------------------
/docs/source/train_and_evaluate.rst:
--------------------------------------------------------------------------------
 1 | Train and Evaluate
 2 | ==================
 3 | Here, we explain how to define and run experiments programmatically. This should be done using PyKEEN.
 4 | 
 5 | Configure your experiment
 6 | ~~~~~~~~~~~~~~~~~~~~~~~~~
 7 | To programmatically train (and evaluate) a KGE model, a python dictionary must be created specifying the experiment:
 8 | 
 9 | .. code-block:: python
10 | 
11 |     config = dict(
12 |         training_set_path           = 'data/corpora/fb15k/compath.tsv',
13 |         test_set_ratio              = 0.1,
14 |         execution_mode              = 'Training_mode',
15 |         kg_embedding_model_name     = 'TransE',
16 |         embedding_dim               = 50,
17 |         normalization_of_entities   = 2,  # corresponds to L2
18 |         scoring_function            = 1,  # corresponds to L1
19 |         margin_loss                 = 1,
20 |         learning_rate               = 0.01,
21 |         batch_size                  = 32,
22 |         num_epochs                  = 1000,
23 |         filter_negative_triples     = True,
24 |         random_seed                 = 2,
25 |         preferred_device            = 'cpu',
26 |     )
27 | 
28 | Run your experiment
29 | ~~~~~~~~~~~~~~~~~~~
30 | .. code-block:: python
31 | 
32 |     results = pykeen.run(
33 |         config=config,
34 |         output_directory=output_directory,
35 |     )
36 | 
37 | Access your results
38 | ~~~~~~~~~~~~~~~~~~~
39 | Show all keys contained in ``results``:
40 | 
41 | .. code-block:: python
42 | 
43 |     print('Keys:', *sorted(results.results.keys()), sep='\n  ')
44 | 
45 | 
46 | Access trained KGE model
47 | ~~~~~~~~~~~~~~~~~~~~~~~~
48 | .. code-block:: python
49 | 
50 |     results.results['trained_model']
51 | 
52 | Access the losses
53 | ~~~~~~~~~~~~~~~~~
54 | .. code-block:: python
55 | 
56 |     results.results['losses']
57 | 
58 | Access evaluation results
59 | ~~~~~~~~~~~~~~~~~~~~~~~~~
60 | .. code-block:: python
61 | 
62 |     results.results['eval_summary']
63 | 
64 | 


--------------------------------------------------------------------------------
/notebooks/Case Scenario ADEPTUS and HSDN.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "# Using BioKEEN to Train and Evaluate a KGE Model on ADEPTUS"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 1,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "import json\n",
 17 |     "import logging\n",
 18 |     "import os\n",
 19 |     "import sys\n",
 20 |     "import time\n",
 21 |     "import warnings\n",
 22 |     "\n",
 23 |     "import matplotlib\n",
 24 |     "import matplotlib.pyplot as plt\n",
 25 |     "import numpy as np\n",
 26 |     "\n",
 27 |     "import biokeen\n",
 28 |     "import pykeen"
 29 |    ]
 30 |   },
 31 |   {
 32 |    "cell_type": "code",
 33 |    "execution_count": 2,
 34 |    "metadata": {},
 35 |    "outputs": [],
 36 |    "source": [
 37 |     "warnings.filterwarnings('ignore', category=UserWarning)"
 38 |    ]
 39 |   },
 40 |   {
 41 |    "cell_type": "code",
 42 |    "execution_count": 3,
 43 |    "metadata": {},
 44 |    "outputs": [],
 45 |    "source": [
 46 |     "logging.basicConfig(level=logging.INFO)\n",
 47 |     "logging.getLogger('biokeen').setLevel(logging.INFO)"
 48 |    ]
 49 |   },
 50 |   {
 51 |    "cell_type": "code",
 52 |    "execution_count": 4,
 53 |    "metadata": {},
 54 |    "outputs": [
 55 |     {
 56 |      "name": "stdout",
 57 |      "output_type": "stream",
 58 |      "text": [
 59 |       "3.7.0 (default, Jul 23 2018, 20:22:55) \n",
 60 |       "[Clang 9.1.0 (clang-902.0.39.2)]\n"
 61 |      ]
 62 |     }
 63 |    ],
 64 |    "source": [
 65 |     "print(sys.version)"
 66 |    ]
 67 |   },
 68 |   {
 69 |    "cell_type": "code",
 70 |    "execution_count": 5,
 71 |    "metadata": {},
 72 |    "outputs": [
 73 |     {
 74 |      "name": "stdout",
 75 |      "output_type": "stream",
 76 |      "text": [
 77 |       "Sun Jan 20 21:23:06 2019\n"
 78 |      ]
 79 |     }
 80 |    ],
 81 |    "source": [
 82 |     "print(time.asctime())"
 83 |    ]
 84 |   },
 85 |   {
 86 |    "cell_type": "code",
 87 |    "execution_count": 6,
 88 |    "metadata": {},
 89 |    "outputs": [
 90 |     {
 91 |      "name": "stdout",
 92 |      "output_type": "stream",
 93 |      "text": [
 94 |       "PyKEEN Version: 0.0.19-dev\n",
 95 |       "BioKEEN Version: 0.0.12-dev\n"
 96 |      ]
 97 |     }
 98 |    ],
 99 |    "source": [
100 |     "print(f'PyKEEN Version: {pykeen.constants.VERSION}')\n",
101 |     "print(f'BioKEEN Version: {biokeen.constants.VERSION}')"
102 |    ]
103 |   },
104 |   {
105 |    "cell_type": "code",
106 |    "execution_count": 7,
107 |    "metadata": {},
108 |    "outputs": [],
109 |    "source": [
110 |     "output_directory = os.path.join(\n",
111 |     "    os.path.expanduser('~'), \n",
112 |     "    'Desktop', \n",
113 |     "    'biokeen_test'\n",
114 |     ")"
115 |    ]
116 |   },
117 |   {
118 |    "cell_type": "markdown",
119 |    "metadata": {},
120 |    "source": [
121 |     "## Step 1: Configure your experiment"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "code",
126 |    "execution_count": 8,
127 |    "metadata": {},
128 |    "outputs": [],
129 |    "source": [
130 |     "config = dict(\n",
131 |     "    training_set_path           = [\n",
132 |     "                                    'bio2bel:adeptus', \n",
133 |     "                                    'bio2bel:hsdn',\n",
134 |     "                                  ],\n",
135 |     "    execution_mode              = 'Training_mode',  \n",
136 |     "    kg_embedding_model_name     = 'TransE',\n",
137 |     "    embedding_dim               = 50, \n",
138 |     "    normalization_of_entities   = 2,  # corresponds to L2\n",
139 |     "    scoring_function            = 1,  # corresponds to L1\n",
140 |     "    margin_loss                 = 1,\n",
141 |     "    learning_rate               = 0.01,\n",
142 |     "    batch_size                  = 128,\n",
143 |     "    num_epochs                  = 1000,  \n",
144 |     "    test_set_ratio              = 0.1,\n",
145 |     "    filter_negative_triples     = True,\n",
146 |     "    random_seed                 = 2,\n",
147 |     "    preferred_device            = 'cpu',\n",
148 |     ")"
149 |    ]
150 |   },
151 |   {
152 |    "cell_type": "markdown",
153 |    "metadata": {},
154 |    "source": [
155 |     "## Step 2: Run BioKEEN to Train and Evaluate the Model"
156 |    ]
157 |   },
158 |   {
159 |    "cell_type": "code",
160 |    "execution_count": 9,
161 |    "metadata": {},
162 |    "outputs": [
163 |     {
164 |      "name": "stderr",
165 |      "output_type": "stream",
166 |      "text": [
167 |       "INFO:pykeen.utilities.pipeline:-------------Train KG Embeddings-------------\n",
168 |       "Training epoch: 100%|██████████| 1000/1000 [10:19<00:00,  1.61it/s]\n",
169 |       "INFO:pykeen.utilities.pipeline:-------------Start Evaluation-------------\n",
170 |       "INFO:pykeen.utilities.evaluation_utils.metrics_computations:Evaluation took 121.18s seconds\n"
171 |      ]
172 |     }
173 |    ],
174 |    "source": [
175 |     "results = pykeen.run(\n",
176 |     "    config=config,\n",
177 |     "    output_directory=output_directory,\n",
178 |     ")"
179 |    ]
180 |   },
181 |   {
182 |    "cell_type": "markdown",
183 |    "metadata": {},
184 |    "source": [
185 |     "## Step 3: Show Exported Results"
186 |    ]
187 |   },
188 |   {
189 |    "cell_type": "markdown",
190 |    "metadata": {},
191 |    "source": [
192 |     "### 3.1: Show Trained Model"
193 |    ]
194 |   },
195 |   {
196 |    "cell_type": "code",
197 |    "execution_count": 10,
198 |    "metadata": {},
199 |    "outputs": [
200 |     {
201 |      "data": {
202 |       "text/plain": [
203 |        "TransE(\n",
204 |        "  (criterion): MarginRankingLoss()\n",
205 |        "  (entity_embeddings): Embedding(4127, 50)\n",
206 |        "  (relation_embeddings): Embedding(3, 50)\n",
207 |        ")"
208 |       ]
209 |      },
210 |      "execution_count": 10,
211 |      "metadata": {},
212 |      "output_type": "execute_result"
213 |     }
214 |    ],
215 |    "source": [
216 |     "results.results['trained_model']"
217 |    ]
218 |   },
219 |   {
220 |    "cell_type": "markdown",
221 |    "metadata": {},
222 |    "source": [
223 |     "### 3.2: Plot losses"
224 |    ]
225 |   },
226 |   {
227 |    "cell_type": "code",
228 |    "execution_count": 11,
229 |    "metadata": {},
230 |    "outputs": [],
231 |    "source": [
232 |     "losses = results.results['losses']"
233 |    ]
234 |   },
235 |   {
236 |    "cell_type": "code",
237 |    "execution_count": 12,
238 |    "metadata": {},
239 |    "outputs": [
240 |     {
241 |      "data": {
242 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzt3XmcXFWd9/HPr7q6el+Tzr50gIQQEsISdh7kJQgh44ArgoKoKOM8MoMO6oALA8yMyziD6MgoqDxuCIPIEgEFWQRRICQkZCUkZO1OSDpJd3pNr7/nj7pdVDq9pdPV1d33+369+kXde0/dOrcvqW+fc+4919wdERERgEi6KyAiIsOHQkFERBIUCiIikqBQEBGRBIWCiIgkKBRERCRBoSAyyplZuZm5mUXTXRcZ/hQKMmyZ2RYzuyANn/sJM2s3s3ozqzWzFWb23kHcv5tZQ7D/zp8vD9b+RY6E/nIQ6d5L7n6OmUWAzwEPmNlkd6/u7w7MLOrubT1snu/uGwelpiKDSC0FGZHM7DNmttHM9pnZYjObFKw3M/uume0O/spfZWZzg22LzGytmdWZWaWZfbGvz3H3DuAeIAc4OtjPe4PWQ42Z/dXMTkiq1xYz+2czWwk0HG6XjZndYmYPmtn/BvV8zczmJ20/zsz+FHz2GjO7JGlbjpn9l5ltNbP9ZvaimeUk7f5jZrbNzPaY2VcPp14SHgoFGXHM7N3AN4HLgInAVuD+YPOFwLnALKAoKLM32PZT4O/cvQCYCzzbj8+KAp8G6oENZnYS8ZD4O2AMcBew2Myykt52BfA3QHEvLYXeXAr8BigFfg08YmaZZpYJ/A54ChgH/ANwr5kdG7zvP4FTgLOC934Z6Eja7znAscD5wM1mdtwA6iajnEJBRqKPAfe4+2vu3gzcBJxpZuVAK1AAzAbM3de5+87gfa3AHDMrdPdqd3+tl884w8xqgLeJf8m/3933A9cCd7n7K+7e7u4/B5qBM5Le+3133+7uTb3s/7Xgr/3On4uSti1z9wfdvRW4HcgO9n8GkA98y91b3P1Z4DHgiqCb61PA9e5eGdTtr8Hvp9Ot7t7k7q8DrwPzEelCoSAj0STirQMA3L2eeGtgcvBF+QPgTmC3md1tZoVB0Q8Ci4CtZva8mZ3Zy2e87O7F7j7W3c9w96eD9dOBG5K/0IGpQZ06be/HMZwc7L/z58nu3h90X1UE+58EbA/WddoKTAbGEg+Pt3r5zLeTXjcSDxiRgygUZCTaQfzLGQAzyyPelVMJ4O7fd/dTgDnEu5G+FKx/1d0vJd718gjwwAA+ezvw712+0HPd/b6kMkc69fDUzhdBC2AK8WPeAUwN1nWaRvy49wAHCMY9RAZKoSDDXaaZZSf9RIH7gE+a2YlBX/43gFfcfYuZnWpmpwf97w3Evyg7zCxmZh8zs6KgW6aWg/vb++vHwGeDzzAzyzOzvzGzgkE6XoBTzOwDwbF+nnj31MvAK8T/wv9yMMZwHvC3wP1JA+K3m9kkM8swszO7jHWI9EmhIMPdE0BT0s8tQVfO14HfAjuJ/3V8eVC+kPgXdzXxrpW9wHeCbVcBW8ysFvgs8bGJw+LuS4HPEO+iqgY2Ap8YwHG93uU+hTuStj0KfCTY/1XAB9y91d1biIfAxcRbBv8DfNzd3wje90VgFfAqsA/4Nvo3LofJ9JAdkeHDzG4BjnH3K9NdFwkn/RUhIiIJCgUREUlQ95GIiCSopSAiIgkjbkK8sWPHenl5ebqrISIyoixbtmyPu5f1VW7EhUJ5eTlLly5NdzVEREYUM9vadyl1H4mISBKFgoiIJCgUREQkQaEgIiIJCgUREUlQKIiISIJCQUREEkITCq9u2cd/PbWe1vaBTKEvIhIOoQmF5duq+e9nN9LcplAQEelJaEIhMyN+qG1qKYiI9Ch0odCiUBAR6VFoQiEWhEJru6YKFxHpSWhCIZphALRqTEFEpEehCYXMREtBoSAi0pMQhoK6j0REehKaUIhFg+4jtRRERHoUmlBQ95GISN9CEwrRiC5JFRHpS2hC4Z3uI40piIj0JDShkOg+0iWpIiI9Cl0otHUoFEREehK6UGhR95GISI9CEwoxdR+JiPQpNKGQmOZCVx+JiPQoNKGg+xRERPoWmlDQLKkiIn0LTShkapoLEZE+hScU1H0kItKn0IRCNBJvKeiSVBGRnqUsFMzsHjPbbWare9huZvZ9M9toZivN7ORU1SX4PDIzTC0FEZFepLKl8DNgYS/bLwZmBj/XAj9MYV2AeBdSm0JBRKRHKQsFd38B2NdLkUuBX3jcy0CxmU1MVX0gHgq6+khEpGfpHFOYDGxPWq4I1qVMZkZEU2eLiPRiRAw0m9m1ZrbUzJZWVVUNeD+ZGaZpLkREepHOUKgEpiYtTwnWHcLd73b3Be6+oKysbMAfqJaCiEjv0hkKi4GPB1chnQHsd/edqfzAWDRCi1oKIiI9iqZqx2Z2H3AeMNbMKoB/ATIB3P1HwBPAImAj0Ah8MlV16ZSTmcGB1vZUf4yIyIiVslBw9yv62O7A51L1+d3JzoxwoFUtBRGRnoyIgebBkp2ZwYE2tRRERHoSqlDIimaopSAi0otQhUJ2ZoRmjSmIiPQoZKGQQZNCQUSkR6EKBV19JCLSu1CFgq4+EhHpXchCIX71UfxqWBER6Sp0oeCOproQEelBqEIhKxo/3AMtCgURke6EKhSyMzMAdAObiEgPQhUKOZ2hoCuQRES6FapQSLQUdAWSiEi3QhYKwZiCWgoiIt0KWSjEWwq6q1lEpHshCwW1FEREehOqUMiKakxBRKQ3oQqFnFg8FJp1SaqISLdCFQrZuiRVRKRX4QqFzjua1X0kItKtcIWCrj4SEelVKENB3UciIt0LVShkRIzMDFP3kYhID0IVChCf/6ippS3d1RARGZZCFwrFuTFqmlrTXQ0RkWEpdKFQkhdjX0NLuqshIjIshS4USnMzqW5UKIiIdCd0oVCSF6O6Qd1HIiLdCV0olObG1FIQEelB6EKhJC9GY0u77lUQEelGSkPBzBaa2Xoz22hmN3azfZqZPWdmy81spZktSmV9AEpyYwBqLYiIdCNloWBmGcCdwMXAHOAKM5vTpdjXgAfc/STgcuB/UlWfTiW5mQAaVxAR6UYqWwqnARvdfZO7twD3A5d2KeNAYfC6CNiRwvoAkJcVBaBRN7CJiBwimsJ9Twa2Jy1XAKd3KXML8JSZ/QOQB1yQwvoAkJcVn/+ooUVjCiIiXaV7oPkK4GfuPgVYBPzSzA6pk5lda2ZLzWxpVVXVEX1gbiyegw3NaimIiHSVylCoBKYmLU8J1iW7BngAwN1fArKBsV135O53u/sCd19QVlZ2RJXKz1IoiIj0JJWh8Cow08xmmFmM+EDy4i5ltgHnA5jZccRD4ciaAn3IDR7J2ajuIxGRQ6QsFNy9DbgOeBJYR/wqozVmdpuZXRIUuwH4jJm9DtwHfMLdPVV1gncGmuvVUhAROUQqB5px9yeAJ7qsuznp9Vrg7FTWoausaISMiOnqIxGRbqR7oHnImRm5sQwamtV9JCLSVehCAeKDzRpoFhE5VChDITeWoYFmEZFuhDIU8rKiNGhMQUTkEOEMhZi6j0REuhPKUCjOzdQjOUVEuhHKUBhfmM2u2uZ0V0NEZNgJZShMKMqmvrmNugOaPltEJFkoQ6E4J/5MhboDGlcQEUkWylDICeY/atIjOUVEDhLKUMjODEJB9yqIiBwklKGQk6mWgohId0IZCp3TZ6ulICJysFCGQrZaCiIi3QplKHQONB9QKIiIHCSUodDZfaTps0VEDhbKUCjNiwGwp153NYuIJAtlKGRFMyjOzWR33YF0V0VEZFgJZSgAjC/IZrfmPxIROUhoQ6E4N5OaJs19JCKSLLShUJiTSa1CQUTkIKENhSKFgojIIUIdCvsVCiIiBwltKBRmZ9LQ0k5be0e6qyIiMmyENhSKcqIA1OqZCiIiCf0KBTO73swKLe6nZvaamV2Y6sqlUmHwoB11IYmIvKO/LYVPuXstcCFQAlwFfCtltRoCRQoFEZFD9DcULPjvIuCX7r4mad2I1BkKugJJROQd/Q2FZWb2FPFQeNLMCoARPUKr7iMRkUP1NxSuAW4ETnX3RiAT+GRfbzKzhWa23sw2mtmNPZS5zMzWmtkaM/t1v2t+hNR9JCJyqGg/y50JrHD3BjO7EjgZ+F5vbzCzDOBO4D1ABfCqmS1297VJZWYCNwFnu3u1mY0byEEMRGF20H10QKEgItKpvy2FHwKNZjYfuAF4C/hFH+85Ddjo7pvcvQW4H7i0S5nPAHe6ezWAu+/ud82PUHZmhFhGRC0FEZEk/Q2FNnd34l/qP3D3O4GCPt4zGdietFwRrEs2C5hlZn8xs5fNbGE/63PEzIyi3Ex27df02SIinfrbfVRnZjcRvxT1/5hZhPi4wmB8/kzgPGAK8IKZzXP3muRCZnYtcC3AtGnTBuFj4848agx/3rBn0PYnIjLS9bel8BGgmfj9Cm8T/wL/Th/vqQSmJi1PCdYlqwAWu3uru28G3iQeEgdx97vdfYG7LygrK+tnlftWPjaPfY0ttHf4oO1TRGQk61coBEFwL1BkZu8FDrh7X2MKrwIzzWyGmcWAy4HFXco8QryVgJmNJd6dtKn/1T8yJbmZuOteBRGRTv2d5uIyYAnwYeAy4BUz+1Bv73H3NuA64ElgHfCAu68xs9vM7JKg2JPAXjNbCzwHfMnd9w7sUA5fSW78Wc3VjS1D9ZEiIsNaf8cUvkr8HoXdAGZWBjwNPNjbm9z9CeCJLutuTnrtwD8FP0OuODc+LKJQEBGJ6++YQqTL5aJ7D+O9w1aipdCg7iMREeh/S+EPZvYkcF+w/BG6tABGotI8dR+JiCTrVyi4+5fM7IPA2cGqu9394dRVa2h0dh/VNKqlICIC/W8p4O6/BX6bwroMufysKLGMCFX1zemuiojIsNBrKJhZHdDdRfxGfJy4MCW1GiJmxtHj8ln/dl26qyIiMiz0Ggru3tdUFiPeUWV5rN1Rm+5qiIgMCyP+CqIjVZSTSZ1mShURARQKFGZnUtvUlu5qiIgMC6EPhYLsKC3tHRxobU93VURE0i70odD5WM66A2otiIiEPhRKgnsV9jboslQRkdCHwpSSXAC272tKc01ERNIv9KEwtSQHgO37GtNcExGR9At9KJTmxciNZVBRrZaCiEjoQ8HMmFqSyza1FEREFAoAx4zPZ91O3dUsIqJQAOZNLqKypon9eiyniIScQgGYWJQNQFXdgTTXREQkvRQKwLiCeCjsrtW9CiISbgoFYELQUqis0RVIIhJuCgVgWmkuOZkZrNVgs4iEnEIByIgYcyYVsqZSoSAi4aZQCMyZWKjLUkUk9BQKgYnF2dQ1t9HYotlSRSS8FAqBsXlZAOytb0lzTURE0kehECjNiwGwr0GhICLhpVAIlBXEWwq763SvgoiEl0IhMKk4PoX2Dt2rICIhplAIjMmLEYtGFAoiEmopDQUzW2hm681so5nd2Eu5D5qZm9mCVNanN5GIMakomwqFgoiEWMpCwcwygDuBi4E5wBVmNqebcgXA9cArqapLfzW1tvP4yp3UHtBsqSISTqlsKZwGbHT3Te7eAtwPXNpNuX8Fvg2kfYrSeZOLAXh67a4010REJD1SGQqTge1JyxXBugQzOxmY6u6P97YjM7vWzJaa2dKqqqrBr2ngR1eeDMCbu+pT9hkiIsNZ2gaazSwC3A7c0FdZd7/b3Re4+4KysrKU1SmaEWFiUTZ76nVZqoiEUypDoRKYmrQ8JVjXqQCYC/zJzLYAZwCL0znYDFCQHeXBZRXprIKISNqkMhReBWaa2QwziwGXA4s7N7r7fncf6+7l7l4OvAxc4u5LU1inPsWi8V/Jtr2N6ayGiEhapCwU3L0NuA54ElgHPODua8zsNjO7JFWfe6S+cvFxALxVpXEFEQmfaCp37u5PAE90WXdzD2XPS2Vd+uuY8fkAbNunloKIhI/uaO5iTDBb6n1LtqW5JiIiQ0+h0EVGxAB44+06Gpr1bAURCReFQi8++pO032QtIjKkFArduP2y+QC8vr0mzTURERlaCoVuvP+k+I3XC6aXpLkmIiJDS6HQDTNjwfQSlm6t1uR4IhIqCoUeLN1aDcC/PbY2zTURERk6CoUe/OjKUwB4YGkFm3Qjm4iEhEKhBwvnTuDXnzkdgJ++uDnNtRERGRoKhV6cdfRYAO59ZRt1GlsQkRBQKPTT+rfr0l0FEZGUUyj0ofMO5w/96KU010REJPUUCn146O/PSrxube9IY01ERFJPodCHeZOLEq8vv/tl2js8jbUREUkthUIfIhHjjX9dCMCyrdU8uebtNNdIRCR1FAr9kJ2ZkXj95i4NOIvI6KVQ6KcPnzIFgDue3sDG3bqZTURGJ4VCP33nw/P5xFnlAFxw+/PprYyISIooFA7DeceWJV7rATwiMhopFA5D56M6Ac79j+dw15VIIjK6KBQOw7ETChKv9za0sGP/gTTWRkRk8CkUDkMsGmHtbRcxc1w+AGd/61k272lIc61ERAaPQuEw5caiPPy5sxPLNzywQt1IIjJqKBQGID8ryq2XHE8sGuG1bTX8829XsnN/U7qrJSJyxBQKA3T1WeV8+aJjgfiDeM785rPcsnhNmmslInJkFApH4FNnz+CWv52TWP7ZX7ewu1aDzyIycikUjkAkYnzi7Bn87rpzEutO+8YzfO2RVfx5QxX1updBREYYhcIgmDeliMf/8Z1g+NXL27jqp0v4/P0r0lgrEZHDZyPtypkFCxb40qVL012NHrW2d3Dr79bwq5e3AXD9+TMpyI4CcPG8iUwuzkln9UQkpMxsmbsv6LNcKkPBzBYC3wMygJ+4+7e6bP8n4NNAG1AFfMrdt/a2z+EeCp1WV+7n/977Gtv2NR60ft1tC8mJZfTwLhGR1OhvKKSs+8jMMoA7gYuBOcAVZjanS7HlwAJ3PwF4EPiPVNVnqM2dXMR/XTb/kPWPrqjkhTerWLNjvwalRWTYiaZw36cBG919E4CZ3Q9cCqztLODuzyWVfxm4MoX1GXKnlpfyvctP5PqksYUbH1p1UJnH/uEc/uPJ9dy4cDbTx+SSl5XKUyIi0rtUfgNNBrYnLVcAp/dS/hrg991tMLNrgWsBpk2bNlj1GxKXnjiZS+ZPormtg98s3c7XHz34Xob3/veLALzwZhUAy752AaV5MdbtrGP59mouOn4CY/OzDtmviEgqpGxMwcw+BCx0908Hy1cBp7v7dd2UvRK4DniXuzf3tt+RMqbQk4bmNvKyosy46XF6+tWPL8xiV+07v4YlXzmfX7y0lVNnlPKuWWUcaG0nMyNCRsSGqNYiMtKlfUwBqASmJi1PCdYdxMwuAL4KXNJXIIwGnd1D915zOh88eQprbr2In33y1IPKJAcCxO99+MFzG7n6niV0dDizv/4H/ukBXe4qIoMvlS2FKPAmcD7xMHgV+Ki7r0kqcxLxAeaF7r6hP/sd6S2F3vzkz5v4zdIK9tQ3s7ehpc/y33j/PG57bA2/v/5cJhRmk50ZwSzeenD3xGsRkeFySeoi4A7il6Te4+7/bma3AUvdfbGZPQ3MA3YGb9nm7pf0ts/RHAqdDrS2c8fTG/jsu45iyeZ9XPvLZf1630nTijnr6DFs3F3Pm7vqee6L59HR4dzxzAYuOn48Jbkx8rKibN7TQFt7BwvKS1N8JCIyXAyLUEiFMIRCV+vfruPosjxWVu7n6bW7uG/JNqobWwGYP6WI1yv2D2i/j37ubO5bso2vvXcO+brqSWRUUyiMcjWNLTyyvJKPn1kOwA+e28h9S7axM3ga3PQxuWzd29jLHg729+cdzc6apvgT5Wqa+MFHT2ZPfTO3LF7DZQum8s3fvwHEg2T+1OJBPx4RSS2FQkjtqGkiPztKYXYmNY0t/OKlrfxl4x7eqqrn/SdNprmtg1+81OtN4326/NSp3P9q/Grjvzv3KD77rqNZt7OW59+s4syjx5CdmcHY/BhZ0QymluYC8TGOtg4nw4xI0lVTjS1tZEczDlonIoNPoSDdOtDazuyv/4F/XjibbfsayIpm8LO/bkls//ApUzhlegk3PrSKsfkx3nfiZO75y2Y6Bvi/yYVzxjOlJJdfvbKVlrYOAE4rL+XN3XW0tzt1zW2cd2wZX110HMu31/CuWWXUNrVyzLh8DZSLDCKFgvRbU0s7EH8Gdee9Dx0djhmJL+Y/rN7Jhl31NLd18MPn3+LWS47nb0+YxFceXsXjq3b2uO+B+tf3zSWWYdzz4hbW76rj3Fll3PCeWcyeWMD3n9lAbizK1WeVk5OZweOrdrJo7gS27G2kfEwu0QxN/ivSlUJBhsy+hhaq6pqpb27juIkF/HZZBafOKGVVxX4eeq2SlzbtTcnnTi7OoawgixXbaxLrrj9/JidOLSY/O8qMsXmU5sYwg+a2Dv60vorXK2r4wgWz+MmLm7ji1GmU5MVwd2oPtFGUk0l7h3PXC29x+anTKM2LJfa7fFs1u+uauej4CSk5FpFUUyjIsODu3POXLZx3bBlHl+VTWdN00PTh2/Y2Ut3YQiwaobmtg01V9exvauXW3609aD9Z0QjTx+Ty5q76lNX12PEFrN9Vl1j+8ccXMH1MLt97ZgOPr4y3hpZ//T0U52YC9Ni91d7huttchh2FgoxYbe0dPLpiB8dOKKClvYM5EwvJzoxPN76pqp6xBVmsqtjPqsr9/Oj5t6hpbOX4SYUcMy6fR1fsAKAwO0p9c9uAx0J6EzGIRiK0tMfHSMrH5PLu2eO55y+bOXlaMasra1l+83v43es7eGnTXhqa21lQXkJpXowPnDS5x+6t9g7nqw+v4sozpjN3ctFB2z5//3IWzZvIhWqpyAApFESA+uY2Hn6tgoVzJ1JWkMWyrfuImHH/ku3Mn1rM2PwYP3z+Lc4+eixNre1UN7Tw0PJK5k4upCQ3xp837BnU+syfWkxeLIMdNU184OQp5MYy+FZwue+EomwqqpsAWHnLhVRWN/HKpr3srD3AXc9vAmDTNxYlrtR6cFkFX/zN63z6nBl87b1dZ6UXOZhCQWSQ/HlDFRMKszlmXD4PL6/k3FllFGZn8npFDY8sr+TeV7b1+v4LjhtHVmYGT6za2e0kiLGMCIU5meyp7//UXzPH5bNh98FdaUU5mWRFI8wcn8+nzzmKO57ZwO7aA7S2d7D4unPYVNVARXUjl58Wn2l4deV+bv3dGv7tffOYVppLdmYEdxKh097h7KhpSlxWLCObQkFkiHT+GzIzGlviXVZ76pqZUJRNc2sHRcEYBMQH5WPRCP/55Hoeeq2CGWPz+PHVC8iNRbn50dW8smkflTVNKa1vQVaUBeUlPLe+6pBtsWiEq86Yzl827uGNt+PjKzdePJu8rCinlZeyYns1k4pzOKosn6fX7uLDC6bQ2ua0dXQwJj+L1ZX7mViUzZj8LHbXHmDNzlrOm1Wmy4uHAYWCyAj16IpKrr9/BZ8+ZwZfWXQcT63dxbtnj+ORFZX8ftVOvrxwNht21/PKpr20tTvXvfsYXt2yj/rmNm5+dA1TSnIS3VAzxuaxeU9DyupqRqL1kxvLoDG4vLmr4yYWcsLkIi6YM54X3qxiamkOFxw3nmVbq/nSgyv59gfn8af1VeTEMrjguPEsmjeRmsYWbnjgdW5aNJun1u5i/dt1/Nv75pIVzSAWfWdcpu5AKy1t8VCSnikUREKoc3bcFdtryM6MMHtCIdv3NXLdr1/jmx84gQlF2ewNZuEtyI7y9Nrd/Oj5t5g1oYBPnzOD0rwYNzzwOlecNo0TpxVz9T1LAJgzsZC1O2sP+qyS3EwmFOWwrsv6oXDlGdP41csHd9tlRSN87PTp3POXzZw2o5Q5Ewv52V+3cHRZHufOKmPdzlq++YETWLG9mvlTiolGIpQVZNHS1kFzWzvjCrPZU9/MmLzYqGzZKBRE5Ii9vGkvJ00rJiuawYZddUwfk8fvV+9kxtg8TpjyzhxYlTVNTCjMpsOdHzy7kSklObx79jjW7qwlGolw+x/X8+qWar688FgmF+fw4z9vYnVlLTPH5ZMRMRpa2ti+L7XdZn05d1ZZ4gmIX7hgFjmxCN944g0uOG48cyYVMn9KEdf8/J3vns7w+MJ7ZlKUk8kf1+7iY6dP57Vt1Zw7s4xNe+opzonR2NLG6TPGsGN/E5v3NLBuZy2fv2AWBkM6vYtCQURGjNb2Dp5Zt5tzZo7l8ZU7eHh5JXd85CQKsqM8sqKSugNtLDx+AlmZEVZsq+E7T62nOCeTvQ0tXH1mOTPK8rj50dXsrW9h0byJPLisIt2H1KszjxrDS5v2MqUkh+MnFTJ9TB5t7c7qyv3MnVxEZtS4cM4ENlXV8+qWfSwoL2X2hAKOn1Q04HtgFAoiEiqNLW00NLdTVpDFC29WMb4wm9b2Do6dUMDb+w/w8PJK5k0u4sd/3kTEjNsvm8+4wmzaO5xn39jNZ36xlLOOHsN75oznzV117K5tZkpJDmPzs9i2r5Hn1ldx8dwJrN9Vx5LN+9JyjF9ZNJtrzz16QO9VKIiIpMjmPQ2U5sXY19DCjLF5uDsvbNjDEyt3ckp5CVNLcvndyh38+/vm8lZVffyu/lllXHj8BO55cTPrdtZy3rHjWDRvAjNuegKAhcdP4PLTprJsazV3Pb8pcXPk/ClF7GtsYfu+Jv74hXOZOb5gQHVWKIiIjAArK2p4edPeQ1oAHR3OgbZ2MiKGO6yq3M+pR/C0xP6Ggh63JSKSRidMKT5o0L5TJGLkxt75ij6SQDgcmmNYREQSFAoiIpKgUBARkQSFgoiIJCgUREQkQaEgIiIJCgUREUlQKIiISMKIu6PZzKqArQN8+1hgcJ+vOPzpmMNBxxwOR3LM0929rK9CIy4UjoSZLe3Pbd6jiY45HHTM4TAUx6zuIxERSVAoiIhIQthC4e50VyANdMzhoGMOh5Qfc6jGFEREpHdhaymIiEgvFAoiIpIQmlAws4Vmtt7MNprZjemuz2Axs6lm9pyZrTWzNWZ2fbC+1Mz+aGYbgv+WBOvNzL4f/B5WmtnJ6T2CgTGzDDNbbmaPBcszzOyV4LjdXs/kAAAFOElEQVT+18xiwfqsYHljsL08nfUeKDMrNrMHzewNM1tnZmeG4Bx/Ifh/erWZ3Wdm2aPxPJvZPWa228xWJ6077HNrZlcH5TeY2dUDrU8oQsHMMoA7gYuBOcAVZjYnvbUaNG3ADe4+BzgD+FxwbDcCz7j7TOCZYBniv4OZwc+1wA+HvsqD4npgXdLyt4HvuvsxQDVwTbD+GqA6WP/doNxI9D3gD+4+G5hP/NhH7Tk2s8nAPwIL3H0ukAFczug8zz8DFnZZd1jn1sxKgX8BTgdOA/6lM0gOm7uP+h/gTODJpOWbgJvSXa8UHeujwHuA9cDEYN1EYH3w+i7giqTyiXIj5QeYEvxDeTfwGGDE7/KMdj3fwJPAmcHraFDO0n0Mh3m8RcDmrvUe5ed4MrAdKA3O22PARaP1PAPlwOqBnlvgCuCupPUHlTucn1C0FHjnf7BOFcG6USVoMp8EvAKMd/edwaa3gfHB69Hwu7gD+DLQESyPAWrcvS1YTj6mxPEG2/cH5UeSGUAV8P+CLrOfmFkeo/gcu3sl8J/ANmAn8fO2jNF9npMd7rkdtHMellAY9cwsH/gt8Hl3r03e5vE/HUbFtcdm9l5gt7svS3ddhlAUOBn4obufBDTwTncCMLrOMUDQ9XEp8UCcBORxaBdLKAz1uQ1LKFQCU5OWpwTrRgUzyyQeCPe6+0PB6l1mNjHYPhHYHawf6b+Ls4FLzGwLcD/xLqTvAcVmFg3KJB9T4niD7UXA3qGs8CCoACrc/ZVg+UHiITFazzHABcBmd69y91bgIeLnfjSf52SHe24H7ZyHJRReBWYGVy7EiA9YLU5znQaFmRnwU2Cdu9+etGkx0HkFwtXExxo61388uIrhDGB/UjN12HP3m9x9iruXEz+Pz7r7x4DngA8Fxboeb+fv4UNB+RH1F7W7vw1sN7Njg1XnA2sZpec4sA04w8xyg//HO4951J7nLg733D4JXGhmJUEr68Jg3eFL9wDLEA7kLALeBN4Cvpru+gzicZ1DvGm5ElgR/Cwi3p/6DLABeBooDcob8Sux3gJWEb+6I+3HMcBjPw94LHh9FLAE2Aj8BsgK1mcHyxuD7Uelu94DPNYTgaXBeX4EKBnt5xi4FXgDWA38EsgajecZuI/4uEkr8VbhNQM5t8CnguPfCHxyoPXRNBciIpIQlu4jERHpB4WCiIgkKBRERCRBoSAiIgkKBRERSVAoiAwhMzuvc2ZXkeFIoSAiIgkKBZFumNmVZrbEzFaY2V3B8xvqzey7wRz/z5hZWVD2RDN7OZjf/uGkue+PMbOnzex1M3vNzI4Odp+f9GyEe4M7dkWGBYWCSBdmdhzwEeBsdz8RaAc+RnxStqXufjzwPPH56wF+Afyzu59A/C7TzvX3Ane6+3zgLOJ3rUJ8JtvPE3+2x1HE5/QRGRaifRcRCZ3zgVOAV4M/4nOIT0jWAfxvUOZXwENmVgQUu/vzwfqfA78xswJgsrs/DODuBwCC/S1x94pgeQXxufRfTP1hifRNoSByKAN+7u43HbTS7Otdyg10jpjmpNft6N+hDCPqPhI51DPAh8xsHCSelzud+L+Xzhk6Pwq86O77gWoz+z/B+quA5929Dqgws/cF+8gys9whPQqRAdBfKCJduPtaM/sa8JSZRYjPXvk54g+3OS3Ytpv4uAPEpzb+UfClvwn4ZLD+KuAuM7st2MeHh/AwRAZEs6SK9JOZ1bt7frrrIZJK6j4SEZEEtRRERCRBLQUREUlQKIiISIJCQUREEhQKIiKSoFAQEZGE/w8z8IKFdlTYAgAAAABJRU5ErkJggg==\n",
243 |       "text/plain": [
244 |        "<Figure size 432x288 with 1 Axes>"
245 |       ]
246 |      },
247 |      "metadata": {},
248 |      "output_type": "display_data"
249 |     }
250 |    ],
251 |    "source": [
252 |     "epochs = np.arange(len(losses))\n",
253 |     "plt.title(r'Loss Per Epoch')\n",
254 |     "plt.xlabel('epoch')\n",
255 |     "plt.ylabel('loss')\n",
256 |     "plt.plot(epochs, losses)\n",
257 |     "plt.show()"
258 |    ]
259 |   },
260 |   {
261 |    "cell_type": "code",
262 |    "execution_count": 13,
263 |    "metadata": {},
264 |    "outputs": [
265 |     {
266 |      "name": "stdout",
267 |      "output_type": "stream",
268 |      "text": [
269 |       "{\n",
270 |       "  \"mean_rank\": 315.866030283081,\n",
271 |       "  \"hits@k\": {\n",
272 |       "    \"1\": 0.10368663594470046,\n",
273 |       "    \"3\": 0.21329822251481237,\n",
274 |       "    \"5\": 0.293614219881501,\n",
275 |       "    \"10\": 0.4206714944042133\n",
276 |       "  }\n",
277 |       "}\n"
278 |      ]
279 |     }
280 |    ],
281 |    "source": [
282 |     "print(json.dumps(results.results['eval_summary'], indent=2))"
283 |    ]
284 |   }
285 |  ],
286 |  "metadata": {
287 |   "kernelspec": {
288 |    "display_name": "Python 3",
289 |    "language": "python",
290 |    "name": "python3"
291 |   },
292 |   "language_info": {
293 |    "codemirror_mode": {
294 |     "name": "ipython",
295 |     "version": 3
296 |    },
297 |    "file_extension": ".py",
298 |    "mimetype": "text/x-python",
299 |    "name": "python",
300 |    "nbconvert_exporter": "python",
301 |    "pygments_lexer": "ipython3",
302 |    "version": "3.7.0"
303 |   }
304 |  },
305 |  "nbformat": 4,
306 |  "nbformat_minor": 2
307 | }
308 | 


--------------------------------------------------------------------------------
/notebooks/Case Scenario ADEPTUS.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "# Using BioKEEN to Train and Evaluate a KGE Model on ADEPTUS"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 1,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "import json\n",
 17 |     "import logging\n",
 18 |     "import os\n",
 19 |     "import sys\n",
 20 |     "import time\n",
 21 |     "import warnings\n",
 22 |     "\n",
 23 |     "import matplotlib\n",
 24 |     "import matplotlib.pyplot as plt\n",
 25 |     "import numpy as np\n",
 26 |     "\n",
 27 |     "import biokeen\n",
 28 |     "import pykeen"
 29 |    ]
 30 |   },
 31 |   {
 32 |    "cell_type": "code",
 33 |    "execution_count": 2,
 34 |    "metadata": {},
 35 |    "outputs": [],
 36 |    "source": [
 37 |     "warnings.filterwarnings('ignore', category=UserWarning)"
 38 |    ]
 39 |   },
 40 |   {
 41 |    "cell_type": "code",
 42 |    "execution_count": 3,
 43 |    "metadata": {},
 44 |    "outputs": [],
 45 |    "source": [
 46 |     "logging.basicConfig(level=logging.INFO)\n",
 47 |     "logging.getLogger('biokeen').setLevel(logging.INFO)"
 48 |    ]
 49 |   },
 50 |   {
 51 |    "cell_type": "code",
 52 |    "execution_count": 4,
 53 |    "metadata": {},
 54 |    "outputs": [
 55 |     {
 56 |      "name": "stdout",
 57 |      "output_type": "stream",
 58 |      "text": [
 59 |       "3.7.0 (default, Jul 23 2018, 20:22:55) \n",
 60 |       "[Clang 9.1.0 (clang-902.0.39.2)]\n"
 61 |      ]
 62 |     }
 63 |    ],
 64 |    "source": [
 65 |     "print(sys.version)"
 66 |    ]
 67 |   },
 68 |   {
 69 |    "cell_type": "code",
 70 |    "execution_count": 5,
 71 |    "metadata": {},
 72 |    "outputs": [
 73 |     {
 74 |      "name": "stdout",
 75 |      "output_type": "stream",
 76 |      "text": [
 77 |       "Sun Jan 20 21:07:16 2019\n"
 78 |      ]
 79 |     }
 80 |    ],
 81 |    "source": [
 82 |     "print(time.asctime())"
 83 |    ]
 84 |   },
 85 |   {
 86 |    "cell_type": "code",
 87 |    "execution_count": 6,
 88 |    "metadata": {},
 89 |    "outputs": [
 90 |     {
 91 |      "name": "stdout",
 92 |      "output_type": "stream",
 93 |      "text": [
 94 |       "PyKEEN Version: 0.0.19-dev\n",
 95 |       "BioKEEN Version: 0.0.12-dev\n"
 96 |      ]
 97 |     }
 98 |    ],
 99 |    "source": [
100 |     "print(f'PyKEEN Version: {pykeen.constants.VERSION}')\n",
101 |     "print(f'BioKEEN Version: {biokeen.constants.VERSION}')"
102 |    ]
103 |   },
104 |   {
105 |    "cell_type": "code",
106 |    "execution_count": 7,
107 |    "metadata": {},
108 |    "outputs": [],
109 |    "source": [
110 |     "output_directory = os.path.join(\n",
111 |     "    os.path.expanduser('~'), \n",
112 |     "    'Desktop', \n",
113 |     "    'biokeen_test'\n",
114 |     ")"
115 |    ]
116 |   },
117 |   {
118 |    "cell_type": "markdown",
119 |    "metadata": {},
120 |    "source": [
121 |     "## Step 1: Configure your experiment"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "code",
126 |    "execution_count": 8,
127 |    "metadata": {},
128 |    "outputs": [],
129 |    "source": [
130 |     "config = dict(\n",
131 |     "    training_set_path           = 'bio2bel:adeptus',\n",
132 |     "    execution_mode              = 'Training_mode',  \n",
133 |     "    kg_embedding_model_name     = 'TransE',\n",
134 |     "    embedding_dim               = 50, \n",
135 |     "    normalization_of_entities   = 2,  # corresponds to L2\n",
136 |     "    scoring_function            = 1,  # corresponds to L1\n",
137 |     "    margin_loss                 = 1,\n",
138 |     "    learning_rate               = 0.01,\n",
139 |     "    batch_size                  = 128,\n",
140 |     "    num_epochs                  = 1000,  \n",
141 |     "    test_set_ratio              = 0.1,\n",
142 |     "    filter_negative_triples     = True,\n",
143 |     "    random_seed                 = 2,\n",
144 |     "    preferred_device            = 'cpu',\n",
145 |     ")"
146 |    ]
147 |   },
148 |   {
149 |    "cell_type": "markdown",
150 |    "metadata": {},
151 |    "source": [
152 |     "## Step 2: Run BioKEEN to Train and Evaluate the Model"
153 |    ]
154 |   },
155 |   {
156 |    "cell_type": "code",
157 |    "execution_count": 10,
158 |    "metadata": {},
159 |    "outputs": [
160 |     {
161 |      "name": "stderr",
162 |      "output_type": "stream",
163 |      "text": [
164 |       "INFO:pykeen.utilities.pipeline:-------------Train KG Embeddings-------------\n",
165 |       "Training epoch: 100%|██████████| 1000/1000 [02:57<00:00,  5.62it/s]\n",
166 |       "INFO:pykeen.utilities.pipeline:-------------Start Evaluation-------------\n",
167 |       "INFO:pykeen.utilities.evaluation_utils.metrics_computations:Evaluation took 32.19s seconds\n"
168 |      ]
169 |     }
170 |    ],
171 |    "source": [
172 |     "results = pykeen.run(\n",
173 |     "    config=config,\n",
174 |     "    output_directory=output_directory,\n",
175 |     ")"
176 |    ]
177 |   },
178 |   {
179 |    "cell_type": "code",
180 |    "execution_count": 11,
181 |    "metadata": {},
182 |    "outputs": [
183 |     {
184 |      "name": "stdout",
185 |      "output_type": "stream",
186 |      "text": [
187 |       "Keys:\n",
188 |       "  entity_to_embedding\n",
189 |       "  entity_to_id\n",
190 |       "  eval_summary\n",
191 |       "  final_configuration\n",
192 |       "  losses\n",
193 |       "  relation_to_embedding\n",
194 |       "  relation_to_id\n",
195 |       "  trained_model\n"
196 |      ]
197 |     }
198 |    ],
199 |    "source": [
200 |     "print('Keys:', *sorted(results.results.keys()), sep='\\n  ')"
201 |    ]
202 |   },
203 |   {
204 |    "cell_type": "markdown",
205 |    "metadata": {},
206 |    "source": [
207 |     "## Step 3: Show Exported Results"
208 |    ]
209 |   },
210 |   {
211 |    "cell_type": "markdown",
212 |    "metadata": {},
213 |    "source": [
214 |     "### 3.1: Show Trained Model"
215 |    ]
216 |   },
217 |   {
218 |    "cell_type": "code",
219 |    "execution_count": 12,
220 |    "metadata": {},
221 |    "outputs": [
222 |     {
223 |      "data": {
224 |       "text/plain": [
225 |        "TransE(\n",
226 |        "  (criterion): MarginRankingLoss()\n",
227 |        "  (entity_embeddings): Embedding(3696, 50)\n",
228 |        "  (relation_embeddings): Embedding(2, 50)\n",
229 |        ")"
230 |       ]
231 |      },
232 |      "execution_count": 12,
233 |      "metadata": {},
234 |      "output_type": "execute_result"
235 |     }
236 |    ],
237 |    "source": [
238 |     "results.results['trained_model']"
239 |    ]
240 |   },
241 |   {
242 |    "cell_type": "markdown",
243 |    "metadata": {},
244 |    "source": [
245 |     "### 3.2: Plot losses"
246 |    ]
247 |   },
248 |   {
249 |    "cell_type": "code",
250 |    "execution_count": 13,
251 |    "metadata": {},
252 |    "outputs": [],
253 |    "source": [
254 |     "losses = results.results['losses']"
255 |    ]
256 |   },
257 |   {
258 |    "cell_type": "code",
259 |    "execution_count": 14,
260 |    "metadata": {},
261 |    "outputs": [
262 |     {
263 |      "data": {
264 |       "image/png": "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\n",
265 |       "text/plain": [
266 |        "<Figure size 432x288 with 1 Axes>"
267 |       ]
268 |      },
269 |      "metadata": {},
270 |      "output_type": "display_data"
271 |     }
272 |    ],
273 |    "source": [
274 |     "epochs = np.arange(len(losses))\n",
275 |     "plt.title(r'Loss Per Epoch')\n",
276 |     "plt.xlabel('epoch')\n",
277 |     "plt.ylabel('loss')\n",
278 |     "plt.plot(epochs, losses)\n",
279 |     "plt.show()"
280 |    ]
281 |   },
282 |   {
283 |    "cell_type": "code",
284 |    "execution_count": 15,
285 |    "metadata": {},
286 |    "outputs": [
287 |     {
288 |      "name": "stdout",
289 |      "output_type": "stream",
290 |      "text": [
291 |       "{\n",
292 |       "  \"mean_rank\": 836.7464646464647,\n",
293 |       "  \"hits@k\": {\n",
294 |       "    \"1\": 0.048484848484848485,\n",
295 |       "    \"3\": 0.1414141414141414,\n",
296 |       "    \"5\": 0.20707070707070707,\n",
297 |       "    \"10\": 0.32626262626262625\n",
298 |       "  }\n",
299 |       "}\n"
300 |      ]
301 |     }
302 |    ],
303 |    "source": [
304 |     "print(json.dumps(results.results['eval_summary'], indent=2))"
305 |    ]
306 |   }
307 |  ],
308 |  "metadata": {
309 |   "kernelspec": {
310 |    "display_name": "Python 3",
311 |    "language": "python",
312 |    "name": "python3"
313 |   },
314 |   "language_info": {
315 |    "codemirror_mode": {
316 |     "name": "ipython",
317 |     "version": 3
318 |    },
319 |    "file_extension": ".py",
320 |    "mimetype": "text/x-python",
321 |    "name": "python",
322 |    "nbconvert_exporter": "python",
323 |    "pygments_lexer": "ipython3",
324 |    "version": "3.7.0"
325 |   }
326 |  },
327 |  "nbformat": 4,
328 |  "nbformat_minor": 2
329 | }
330 | 


--------------------------------------------------------------------------------
/notebooks/Case Scenario ComPath.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "# Using BioKEEN to Train and Evaluate a KGE Model on ComPath"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 1,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "import json\n",
 17 |     "import logging\n",
 18 |     "import os\n",
 19 |     "import sys\n",
 20 |     "import time\n",
 21 |     "import warnings\n",
 22 |     "\n",
 23 |     "import matplotlib\n",
 24 |     "import matplotlib.pyplot as plt\n",
 25 |     "import numpy as np\n",
 26 |     "\n",
 27 |     "import biokeen\n",
 28 |     "import pykeen"
 29 |    ]
 30 |   },
 31 |   {
 32 |    "cell_type": "code",
 33 |    "execution_count": 2,
 34 |    "metadata": {},
 35 |    "outputs": [],
 36 |    "source": [
 37 |     "warnings.filterwarnings('ignore', category=UserWarning)"
 38 |    ]
 39 |   },
 40 |   {
 41 |    "cell_type": "code",
 42 |    "execution_count": 3,
 43 |    "metadata": {},
 44 |    "outputs": [],
 45 |    "source": [
 46 |     "logging.basicConfig(level=logging.INFO)\n",
 47 |     "logging.getLogger('biokeen').setLevel(logging.INFO)"
 48 |    ]
 49 |   },
 50 |   {
 51 |    "cell_type": "code",
 52 |    "execution_count": 4,
 53 |    "metadata": {},
 54 |    "outputs": [
 55 |     {
 56 |      "name": "stdout",
 57 |      "output_type": "stream",
 58 |      "text": [
 59 |       "3.7.0 (default, Jul 23 2018, 20:22:55) \n",
 60 |       "[Clang 9.1.0 (clang-902.0.39.2)]\n"
 61 |      ]
 62 |     }
 63 |    ],
 64 |    "source": [
 65 |     "print(sys.version)"
 66 |    ]
 67 |   },
 68 |   {
 69 |    "cell_type": "code",
 70 |    "execution_count": 5,
 71 |    "metadata": {},
 72 |    "outputs": [
 73 |     {
 74 |      "name": "stdout",
 75 |      "output_type": "stream",
 76 |      "text": [
 77 |       "Sun Jan 20 20:57:46 2019\n"
 78 |      ]
 79 |     }
 80 |    ],
 81 |    "source": [
 82 |     "print(time.asctime())"
 83 |    ]
 84 |   },
 85 |   {
 86 |    "cell_type": "code",
 87 |    "execution_count": 6,
 88 |    "metadata": {},
 89 |    "outputs": [
 90 |     {
 91 |      "name": "stdout",
 92 |      "output_type": "stream",
 93 |      "text": [
 94 |       "PyKEEN Version: 0.0.19-dev\n",
 95 |       "BioKEEN Version: 0.0.12-dev\n"
 96 |      ]
 97 |     }
 98 |    ],
 99 |    "source": [
100 |     "print(f'PyKEEN Version: {pykeen.constants.VERSION}')\n",
101 |     "print(f'BioKEEN Version: {biokeen.constants.VERSION}')"
102 |    ]
103 |   },
104 |   {
105 |    "cell_type": "code",
106 |    "execution_count": 7,
107 |    "metadata": {},
108 |    "outputs": [],
109 |    "source": [
110 |     "output_directory = os.path.join(\n",
111 |     "    os.path.expanduser('~'), \n",
112 |     "    'Desktop', \n",
113 |     "    'biokeen_test'\n",
114 |     ")"
115 |    ]
116 |   },
117 |   {
118 |    "cell_type": "markdown",
119 |    "metadata": {},
120 |    "source": [
121 |     "## Step 1: Configure Your Experiment"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "code",
126 |    "execution_count": 8,
127 |    "metadata": {},
128 |    "outputs": [],
129 |    "source": [
130 |     "config = dict(\n",
131 |     "    training_set_path           = 'bio2bel:compath',\n",
132 |     "    execution_mode              = 'Training_mode',  \n",
133 |     "    kg_embedding_model_name     = 'TransE',\n",
134 |     "    embedding_dim               = 50, # 150 is better for real\n",
135 |     "    normalization_of_entities   = 2,  # corresponds to L2\n",
136 |     "    scoring_function            = 1,  # corresponds to L1\n",
137 |     "    margin_loss                 = 5,\n",
138 |     "    learning_rate               = 0.01,\n",
139 |     "    batch_size                  = 32,\n",
140 |     "    num_epochs                  = 1000,  # 2500 is better for real\n",
141 |     "    test_set_ratio              = 0.1,\n",
142 |     "    filter_negative_triples     = True,\n",
143 |     "    random_seed                 = 2,\n",
144 |     "    preferred_device            = 'cpu',\n",
145 |     ")"
146 |    ]
147 |   },
148 |   {
149 |    "cell_type": "markdown",
150 |    "metadata": {},
151 |    "source": [
152 |     "## Step 2: Run BioKEEN to Train and Evaluate the Model"
153 |    ]
154 |   },
155 |   {
156 |    "cell_type": "code",
157 |    "execution_count": 10,
158 |    "metadata": {},
159 |    "outputs": [
160 |     {
161 |      "name": "stderr",
162 |      "output_type": "stream",
163 |      "text": [
164 |       "INFO:pykeen.utilities.pipeline:-------------Train KG Embeddings-------------\n",
165 |       "Training epoch: 100%|██████████| 1000/1000 [01:22<00:00, 12.12it/s]\n",
166 |       "INFO:pykeen.utilities.pipeline:-------------Start Evaluation-------------\n",
167 |       "INFO:pykeen.utilities.evaluation_utils.metrics_computations:Evaluation took 2.77s seconds\n"
168 |      ]
169 |     }
170 |    ],
171 |    "source": [
172 |     "results = pykeen.run(\n",
173 |     "    config=config,\n",
174 |     "    output_directory=output_directory,\n",
175 |     ")"
176 |    ]
177 |   },
178 |   {
179 |    "cell_type": "code",
180 |    "execution_count": 11,
181 |    "metadata": {},
182 |    "outputs": [
183 |     {
184 |      "name": "stdout",
185 |      "output_type": "stream",
186 |      "text": [
187 |       "Keys:\n",
188 |       "  entity_to_embedding\n",
189 |       "  entity_to_id\n",
190 |       "  eval_summary\n",
191 |       "  final_configuration\n",
192 |       "  losses\n",
193 |       "  relation_to_embedding\n",
194 |       "  relation_to_id\n",
195 |       "  trained_model\n"
196 |      ]
197 |     }
198 |    ],
199 |    "source": [
200 |     "print('Keys:', *sorted(results.results.keys()), sep='\\n  ')"
201 |    ]
202 |   },
203 |   {
204 |    "cell_type": "markdown",
205 |    "metadata": {},
206 |    "source": [
207 |     "## Step 3: Show Exported Results"
208 |    ]
209 |   },
210 |   {
211 |    "cell_type": "markdown",
212 |    "metadata": {},
213 |    "source": [
214 |     "### 3.1: Show the Trained Model"
215 |    ]
216 |   },
217 |   {
218 |    "cell_type": "code",
219 |    "execution_count": 12,
220 |    "metadata": {},
221 |    "outputs": [
222 |     {
223 |      "data": {
224 |       "text/plain": [
225 |        "TransE(\n",
226 |        "  (criterion): MarginRankingLoss()\n",
227 |        "  (entity_embeddings): Embedding(997, 50)\n",
228 |        "  (relation_embeddings): Embedding(1, 50)\n",
229 |        ")"
230 |       ]
231 |      },
232 |      "execution_count": 12,
233 |      "metadata": {},
234 |      "output_type": "execute_result"
235 |     }
236 |    ],
237 |    "source": [
238 |     "results.results['trained_model']"
239 |    ]
240 |   },
241 |   {
242 |    "cell_type": "markdown",
243 |    "metadata": {},
244 |    "source": [
245 |     "## 3.2: Plot the Losses"
246 |    ]
247 |   },
248 |   {
249 |    "cell_type": "code",
250 |    "execution_count": 13,
251 |    "metadata": {},
252 |    "outputs": [],
253 |    "source": [
254 |     "losses = results.results['losses']"
255 |    ]
256 |   },
257 |   {
258 |    "cell_type": "code",
259 |    "execution_count": 14,
260 |    "metadata": {},
261 |    "outputs": [
262 |     {
263 |      "data": {
264 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAEWCAYAAABliCz2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzt3Xd8VfX9x/HXJ5ssCCEBZIUwZToCghMQlaLVuqXuurXW0dqfVluttc62WlfrtlqLo46qKC5AQBANCMiUsGQEElZCGCHj+/vjnoSEkJCE3Nybe9/Px+M+uPecc8/5nBx95+R7vud7zDmHiIiEvohAFyAiIs1DgS8iEiYU+CIiYUKBLyISJhT4IiJhQoEvIhImFPgiLZyZvWxm9wW6Dgl+CnwJCDNbZWajA11HUzOze8ysxMyKqry2BbouEVDgizSamUXVMusN51xilVebZi1MpBYKfAk6ZnaVmeWY2RYze9/MDvGmm5k9amZ5ZlZoZt+b2QBv3lgzW2Rm281snZn9ppZ1X2ZmX5nZk2ZWYGZLzOzEKvNbm9kLZpbrrec+M4vc57uPmtlm4J5G7Jszs1+Z2Qoz22Rmj5hZhDcvwszuMrPV3j6+Ymatq3z3WDObYWbbzGyNmV1WZdUpZjbB2/9ZZtajobVJ6FPgS1Axs1HAA8B5QEdgNfC6N/tk4HigN9DaW2azN+8F4BrnXBIwAJhUx2aOApYD7YC7gXfMrK0372WgFOgJHO5t88p9vrsCaA/8uZG7eSaQBRwBnAH8wpt+mfcaCWQCicCTAGbWDfgYeAJIAw4D5lZZ5wXAH4EUIOcgapMQpsCXYHMh8KJzbo5zrhi4AxhuZhlACZAE9AXMObfYOZfrfa8E6Gdmyc65rc65OXVsIw94zDlX4px7A1gKnGpm7YGxwM3OuR3OuTzgUXxhWmG9c+4J51ypc25XLes/zzsLr3hN3mf+Q865Lc65H4HHgHFV9v1vzrkVzrkib98v8JqOfg587pwb79W92TlXNfDfdc5945wrBV7D9wtBpBoFvgSbQ/Cd1QPgBd9moJNzbhK+M96ngDwze9bMkr1Fz8YX1qvN7EszG17HNta56qMGrva22w2IBnIrwhp4BkivsuyaeuzDm865NlVeI/eZX3UdFduGffbdex+F76+JLvj+KqnNhirvd+L760CkGgW+BJv1+IIXADNLAFKBdQDOucedc0cC/fA17dzmTf/WOXcGvnB+D3izjm10MjOr8rmrt901QDHQrkpYJzvn+ldZtimGl+2yn23DPvvuzSsFNnq1qV1eDooCXwIp2sziqryigPHA5WZ2mJnFAvcDs5xzq8xsiJkdZWbRwA5gN1BuZjFmdqGZtXbOlQCFQHkd200HfmVm0WZ2LnAo8JHXPPQp8FczS/YuovYwsxOaeL9vM7MUM+sC3AS84U0fD9xiZt3NLNHb9zeqNNOMNrPzzCzKzFLNTM020iAKfAmkj4BdVV73OOc+B34PvA3k4jurrWhDTwaeA7bia+7YDDzizbsYWGVmhcC1+NrDazML6AVswndx8xznXMXF30uAGGCRt53/4rt43BDn79MPv8jMqjYL/Q+Yje+i6wR8F5wBXgReBaYCK/H9QrsRwGvvHwv8GtjifXdwA+uSMGd6AIqEE68r45XOuWMDtH0H9HLO5QRi+xLedIYvIhImFPgiImFCTToiImFCZ/giImGitsGfAqJdu3YuIyMj0GWIiLQYs2fP3uScS6vPskEV+BkZGWRnZwe6DBGRFsPMVh94KR816YiIhAkFvohImFDgi4iECQW+iEiYUOCLiIQJv/bSMbNVwHagDCh1zmX5c3siIlK75uiWOdI5t6kZtiMiInVo8U06ZeWOpybnMPWH/ECXIiIS1Pwd+A741Mxmm9nV+1vAzK42s2wzy87Pb3hoR0YYz3y5nM8XbzzYWkVEQpq/A/9Y59wRwE+AG8zs+H0XcM4965zLcs5lpaXV6+7gGjqnxLNmy86DLFVEJLT5NfCdcxXPIc0D3gWG+mM7nVNasW7bLn+sWkQkZPgt8M0swcySKt4DJwML/LGt1MQYtu4s8ceqRURChj976bQH3jWziu38xzk30R8bSm4VTcEuBb6ISF38FvjOuRU000OWW7eKZk9pObtLyoiLjmyOTYqItDgtvlsm+AIf0Fm+iEgdQiLw28bHALCpqDjAlYiIBK+QCPxOKa0AWLdVPXVERGoTEoHfOSUegB/VF19EpFYhEfgp8dG0S4xlUW5hoEsREQlaIRH4ZsahHZPIySsKdCkiIkErJAIfoGPrODYU7A50GSIiQStkAr9D61bkFxVTUlYe6FJERIJS6AR+chzOQf52dc0UEdmf0An81rEAbChUs46IyP6ETOC3T44D4O3ZawNciYhIcAqZwO+VngTAyk07AlyJiEhwCpnAj4mK4NSBHclVTx0Rkf0KmcAHX9fM9dt24ZwLdCkiIkEntAK/TSuKS8t5YfrKQJciIhJ0Qirwk2J9w/vfN2FxgCsREQk+IRX4/Q5JDnQJIiJBy5+POGx2Azq15oTeaXqguYjIfoTUGT5ARmq87rYVEdmPkAv8tKRYCnaVUFxaFuhSRESCSkgGPsCmoj0BrkREJLiEXOB38Z5+9d2PWwNciYhIcAm5wD8qM5Xu7RL4++fLKNVQySIilUIu8CMjjF8c251leUUaV0dEpIqQC3yA3umJgIZKFhGpKiQDv0Nr31DJGkhNRGSvkAz8Tm1aAfDb/87ni8UbA1yNiEhwCMnAj4rcu1vjv1kTwEpERIJHSAY+QFy0b9dKy9VTR0QEQjjwX73iKACmLM1nyw7dhCUiErKBPySjbeX76TmbAliJiEhwCNnAB3j6wiMA9AQsERFCPPCzuqUAsEHdM0VEQjvwk1tFA/DAx0t0li8iYc/vgW9mkWb2nZl96O9t7SsuOrLy/eLc7c29eRGRoNIcZ/g3AQF7yOzD5wwCYOqy/ECVICISFPwa+GbWGTgVeN6f26nLeVldAHjw4yWUaPRMEQlj/j7Dfwz4LVBr0prZ1WaWbWbZ+fn+PQv/cP56v65fRCSY+S3wzew0IM85N7uu5ZxzzzrnspxzWWlpaX6p5V+/GArAO3PW+WX9IiItgT/P8I8BTjezVcDrwCgz+7cft1erE3qncXK/9qzevDMQmxcRCQp+C3zn3B3Ouc7OuQzgAmCSc+4if23vQHq3T2Lt1p16CpaIhK2Q7odfVfvkWModGldHRMJWswS+c26Kc+605thWbdKTfQ9FeWjiUp3li0hYCpsz/IqHorw9Zy2vf6sx8kUk/IRN4Pc/JLny/ZINhQGsREQkMMIm8M2s8v2/v/6R7btLAliNiEjzC5vA39etb84LdAkiIs0qrAJ/3NCule8XrisIYCUiIs0vrAL/gbMGclR335Ow8rYXB7gaEZHmFVaBD9A2IQaA0nJHwS6144tI+Ai7wL//zIEMz0wFYEV+UYCrERFpPmEX+CkJMZVj5J/59Ax+2KgHo4hIeAi7wAfo0jaeIRm+592e8tjUAFcjItI8wjLwAY7p2Q4APepWRMJF2Aa+gl5Ewk3YBv5Qr3smwKaiYvXYEZGQF7aBf0zPdlw0zHcjVtZ9n3Pknz4LcEUiIv4VtoEP8NsxfYmN8v0ISsvVxiMioS2sAz85LpqxAztWfi5T6ItICAvrwAfo1T6x8v2Gwt0BrERExL/CPvDbJcRWvp+3ZlsAKxER8a+wD/z42MjK918szgtgJSIi/qXAj9kb+G/PWcurM1cFrBYREX8K+8CPiqj+I/j9/xYGqBIREf9S4EdYjWnrtu0KQCUiIv4V9oEfE1XzR3DMg5MoKSsPQDUiIv4T9oF/eNcUrjkhk/duOIZDOyZXTn9/7nq27NgTwMpERJqWuSAaRSwrK8tlZ2cHbPs795TS7w+fVH7u2DqOmXecGLB6REQOxMxmO+ey6rNs2J/hVxUfE8Xzl+z9ueUW6EYsEQkdCvx9ZLSLD3QJIiJ+ocDfR8/0pMoB1QCKS8sCWI2ISNNR4O9Hh9Zxle/zCosDWImISNNR4O/Huq17++Ff9tI3rNy0I4DViIg0DQX+fjx94RH0SveNork8fwc3jp8T4IpERA6eAn8/Tu7fgXdvOKby84J1hWTeMSGAFYmIHDwFfi0SY6OqfS53MH3ZpgBVIyJy8BT4dejatnoXzYtemBWgSkREDp7fAt/M4szsGzObZ2YLzeyP/tqWv4zqmx7oEkREmkzUgRdptGJglHOuyMyigelm9rFz7ms/brNJ3XXqoVwyvBvbdpVw1tMzOLxrm0CXJCLSaH4LfOcbpKfI+xjtvYJn4J56iIqMIDPN11tnSEYK367ayuLcwmqDrImItBR+bcM3s0gzmwvkAZ8551psI3iXFF97/k/+Po2JC3IDXI2ISMP5NfCdc2XOucOAzsBQMxuw7zJmdrWZZZtZdn5+vj/LOSh3n96f5DjfH0QPf7I0wNWIiDRcs/TScc5tAyYDY/Yz71nnXJZzListLa05ymmU1q2ieed6X9/8Ffm681ZEWh5/9tJJM7M23vtWwEnAEn9trzn0TE/k3CM7AzBz+eYAVyMi0jD+PMPvCEw2s/nAt/ja8D/04/aaxfUjewIw7rmvmbRkY4CrERGpv3r10jGzHsBa51yxmY0ABgGveE01++Wcmw8c3iRVBpHu7RIq3z85KYdW0VEM75EawIpEROqnvmf4bwNlZtYTeBboAvzHb1UFuV+d2AuAOT9uY9xzX3PaE9P4bJHO9kUkuNU38Mudc6XAmcATzrnb8DXZhKVbT+rNxcO6VX5esK6Qq14J3LN4RUTqo76BX2Jm44BLgYp2+Gj/lNQy3HtG/xrTyspb1H1lIhJm6hv4lwPDgT8751aaWXfgVf+VFfzMrMa0vO166LmIBK96Bb5zbpFz7lfOufFmlgIkOece8nNtQW/Jn8bw8DmDKj8Pf2AS89bUeh1bRCSg6hX4ZjbFzJLNrC0wB3jOzP7m39KCX1x0JEfv00Pnyck5AapGRKRu9W3Sae2cKwTOwtcd8yhgtP/Kajlat6p+KSMhJjJAlYiI1K2+gR9lZh2B89h70Vao+WSsLTtLKt9v312y7+IiIgFT38C/F/gEWO6c+9bMMoFl/iur5TAzxl81jCfG+e4xm/pDPg9PXMKgez5h4D2f8r+56wJcoYiIT30v2r7lnBvknLvO+7zCOXe2f0trOYb3SOWngw+p/Pz0lOUU7i4FYNKSvECVJSJSTX0v2nY2s3fNLM97vW1mnf1dXEszsk/N0T5rdt4UEQmM+jbpvAS8DxzivT7wpkkVL10+lN+f1q/atAgzpi3LZ+oPwTvWv4iEB/M9ifAAC5nN9R5kUue0g5WVleWys0NjiILnp63gvgmLSYyNoqjY17wz4/ZRJLeKrnGhV0SkscxstnMuqz7L1vcMf7OZXeQ9sjDSzC4CNCB8Ha48LpNBnVtXhj3A0Q9O4px/zAhgVSISzuob+L/A1yVzA5ALnANc5qeaQsbmoj01pi3ZsD0AlYiI1L+Xzmrn3OnOuTTnXLpz7meAeukcwLptu/Y7fcG6gmauRETk4J54dWuTVRGi7jr1UAAyUuOrTZ+2bBM7qjT1iIg0h4MJfPU4PIArj8tk1YOnkhhX/SLtQxOXMObvUwNUlYiEq4MJfA3+Xk+PnDO4xrQ1W/bf3CMi4i91Br6ZbTezwv28tuPrjy/1cGjHZP585oAa04uKS3l15iqcc3y/toD6dJEVEWmsOjuEO+eSmquQUHfhUd24890F1aYNuPsTAFZv3snz01fy0NkDOX9I10CUJyJh4GCadKSBXrp8CD87rOYfRtmrtwKwOFddNkXEfxT4zWhkn3Qeu+BwVj4wlttO6VM5PSevCICXZ6wKUGUiEg4U+AFgZtwwsidf3jYCoNrduCIi/qLAD6BuqQk87o2jX2HnHoW/iPiHAj/ATh9cvU3/htfmcNoT0yjU07JEpIkp8IPA/HtO5u3rhgMweWk+C9YVMuieT9VNU0SalAI/CCTHRXNkt7Y1pudtLw5ANSISqhT4QWzask088NHiWgdhExFpCAV+EHnm4iPJTEuo/Pybt+bxzNQV3PrGXMA3yubukrJAlSciLZwCP4ic0r8D955ecwiG9QW72FxUzGlPTOf2t+cHoDIRCQUK/CCTFFdztItIM3ILdgPw3tz1PPrZD5SWleuirog0iAI/yAzu0oa3rh1e+fmYnqms2ryT056YXjnt718so+edH/PoZz8EokQRaaEU+EFoSMbeHjsFu2rvj//4pBwybp/ANyu3APDqzFV8sXijv8sTkRaqztEyJXCevySLVZt3MLxHKqc+Pr3OZb9YvJE28dH8/n8LAVj14KnNUaKItDB+O8M3sy5mNtnMFpnZQjO7yV/bCkWj+7XnyuMy6X9Iaz688dgDLn/yo3qClojUzZ9NOqXAr51z/YBhwA1m1s+P2wtZAzq15tie7Wqd/8zUFc1YjYi0VH4LfOdcrnNujvd+O7AY6OSv7YW6R84dxJXHdgfADC4d3i3AFYlIS2PN0bXPzDKAqcAA51zhPvOuBq4G6Nq165GrV6/2ez0tWV7hbuJiIkmOi2bdtl3c8vpcvlm1pdoyt57Um7OP7Eyr6EjaJsQEqFIRaQ5mNts5l1WvZf0d+GaWCHwJ/Nk5905dy2ZlZbns7Gy/1hNqnp26nPs/WlLr/E9uPp7MtASiI9UhSyQUNSTw/ZoCZhYNvA28dqCwl8Y5+4jOjOiTxs+P6sqvT+pdY/4pj03ltrfmUVbu+NunS/l6xeYAVCkiwcBv3TLNzIAXgMXOub/5azvhLjUxlpcvHwr4mns+XrCBRbnVWs14b+56YqIieDN7LY9PylG3TZEw5c8z/GOAi4FRZjbXe4314/bCXnpyHB/ddBzXntCjxrw3s9cGoCIRCSb+7KUz3TlnzrlBzrnDvNdH/tqe7DWiTxoZqfGMPjSdHlVG36xw7weLKC0rD0BlIhJIupIXgoZlpjLltpE8f+kQJvzquMrpN4z0nfm/+NVK7v1wEeXljj2lNYN/9uqtZNw+ge/XFjRbzSLifwr8EBcXHcmFR3UF4LKju/PAWQMBmLe2gDvfW8CR931GSVk5u0vKmLdmG+c/M5MP5q0HYFpOfsDqFpGmp7F0wsA9p/fnquMySUuKZdzQrsxYvpkP5q1n3pptAPS68+Nqy5v5/o2KsOYuVUT8SGf4YSA6MoKMdnvb8pdt3F7n8l+v8N3IFWEKfJFQosAPQ6MPbV+v5abnbOLUx6fxyCdLyLh9Aq/N0l3QIi2ZmnTC0C0n9SYxLoqnJuewfXcpbRNi2LJjT43lpiz1teEvXO/r13/nuws4pHUrRvZNb9Z6RaRp6Aw/DEVGGNee0IPPbjkB8DXdZHVLoVV05AG/e/nL3/Kzp76qNk1dPEVaBp3hh7G0pFhGH5rO1cf3YGh331O2tu3cw2H3flbn9+au2UZxaRmG8eUP+Vz1Sjaf33o8PdOTmqNsEWmkZhkts740eFpwmL5sE5t3FNMmPoYb/zOHwt2lNZZJjI2iqHjv9IfPGURaUiwj+6i5R6Q5BdVomQ2hwA8+O/eUctt/5zNhfm69lp/3h5NpHR9d6/w5P26lZ3oiyXG1LyMi9Rc0o2VKyxcfE0V8Pdr2K7w3dx0/bt7JVzmbuHH8d1Q9oSguLeOsp2dw7auz/VGqiByAAl8O6PCuKQDcPLrXAZe9+/2FHP/IZC58fhYfzFvPrpKyynmfL8oDfNcARKT56aKtHNC4oV3Iykihd/skDu+awqUvflPv7z748RL6dUymbUIMN/xnDgCRuqFLJCAU+HJAZkbv9r4eOCf0TmNgp9Z8v843sNrQjLaVj1hMT4olb3txte++MrPmzVrKe5HAUJOONNirVwzlT2f0JykuipcuH1I5/Zs7RzPrdyce8PuFu0vJLdjFnB+3+rNMEdmHAl8arE18DBcPz+D7e04hITaKcUO7MjwzFYD2yXH85dzBB1zH8AcmcdbTM3jgo8WUlQdPTzGRUKbAl4P2wFkDGX/1sMrP5xzZufJ99l2jGdipda3ffWbqCi5+YVaN6Tv3lLJtZ83hHkSk8dSGL37xwqVZJMVF0y4xlp8O7sj36woYf9UwLn3xG/bsMxTDjOWbOe+fM2mXFMOIPun0PySZC5+fxbadJXr+rkgTUuCLX5xYZUTOK4/N5GeHdyI9KY4ubVuxPH9HjeUrLvx+9P2GatO/ytlE7/ZJbNmxhz4dNHSDyMHQnbbSrJZt3M7TU5Zz46iePPDxEj5btLHe3531uxOZv7aAEX3SiI5Ua6QIaGgFaUGmLcvn4hfq368f4O8XHMZPBx3Cj1t2Eh8Tyfhv1nDioekMqONagUioUuBLi1Fe7nhycg5/++wHAC4e1o1uqfHcN2Fxg9YTExXBD/f9hN0lZXyxOI+xAztg6vAvYaAhga82fAmoiAjjVyf24oIhXYiIMNrGx2BGtcAf2r0t36zcUud69pSW8/y0FTw7dQV524v5vzF9uW5ED3+XL9KiqCFUgkJ6chztEmOJiDDMjI9vOs43PSmWf19xFADDMtvWuY77JiyuvNP3oYm+xzJeUsswEIvWFxJMf92KNAcFvgSl1MQY799YYqIieP+Xx/DsJfX6q7WaqT/kk3H7BLJXbakM+MlL8hj7+DTembOuSWsWCXYKfAlK6Ulx3HtGf1641Bfygzq3ITkuunIcnvOzujRofef8cyY3/GcOGwt3c/nL3wLw67fmsXNPzYe7iIQqXbSVFiUnbztLNxRx6qCOgO+RjH/5dCn//vrHymVO6te+1u6ePdISqt0HcM0JmQzp1pY9ZeVc/9ocnvr5ESzKLeC2U/r6d0dEmoh66UjYWZ5fxKcLN3Llcd3ZVVLGoHs+rZx37xn9+cP/FjZofdeP6MFvx/T1XQyevoJfHNOduAY8CEakueiJVxJ2eqQlct2IHkRHRpAcF83TFx4BwF/PHcwlwzO4+vjMast3atOqzvU9PWU5/5qxit53fczDE5fS9/cT2bpDY/tIy6bAl5A0dmBH3r3+aM48vBMAURF7++QnxkYx6TcnHLDb5t3vV/+r4PA/fQbAjuJS/jt7rXr5SIujwJeQdXjXFCK8oL9gSFcqMt8MYqMiOap73d089+dnT33FsQ9N4jdvzWNRbiEA78xZywvTVzZZ3SL+osCXsNA1NZ6cP4/lzMM78dJlvoe2HNnN96zefcfvv3FUz1rXM3fNNrbuLAFgy449lJaVc+ub8/jTh4t4buoKXpm5yi/1izQFXbQVATJun8BlR2dwz+n9Afjux62c+fQMAP577XDO+efM/X6vfXIsGwurP9Zx8b1juPO97/nlyJ6syN/Bla9kM+8PJ7Mwt4ANBbs564jO+12XSGOol45IE3jmy+VEmHH+0C7Vev00xog+aUxZmg+gMf6lSQXFWDpm9iJwGpDnnBvgr+2I+Ms1J/gu6paXOzq1acWtJ/XG4evnP/iPDfsFUBH2FeuLiKh9YLevV2zmkNat6Joa36i6RWrjtzN8MzseKAJeqW/g6wxfWooV+UW8MH0lr83y3fB1/5kD+d2739fru2P6d2B9wS4iI4zuqQkc0S2FyAjjH1OW88g5gzj/2a8BWHrfGGKj1Pdf6hY0TTpmlgF8qMCXUFRW7ujxu49IT4plym0j+OV/vmPSkrxqy/zzoiO49t9zGr2NDslxzLxjlIZ6llq1qMA3s6uBqwG6du165OrVq/1Wj0hTm7F8E13bxtM5JZ7yckfm7z6qnPftnaNJS4rlq5xN/PXTpXy/roCSssb9/3ZSv/Y814jB4yT0tajAr0pn+NLSfb1iMxmpCcRGRZCSEFNj/iszV5GTV8QrM2ue2GS2S2DFpprP+62qbUIMme0SuOWk3hzTs11TlS0tmAJfJMgV7CzhycnLGNi5DdN+yOet2Wu5fkQP0pJi+eMHiw74/YzUeP595VFc+uI3pCbGcnyvdvxyVK86v1NaVk658z0dTEKHAl+khdlQsJv0JN8DYErKyul158cNXsfpgw/h0qMzmLVyMxFmXHvC3qEjtu7Yw3EPT6aouLTWbqF//GAhpWWOP/1MnepakmDpljkeGAG0M7O1wN3OuRf8tT2RlqxD67jK99GREax8YCxmxs49pURGGG9mr+X37y2ocx3vz1vP+/PWV37+yYAOvPTVKvKLipm/dhtFxfsf+3/Ljj0c4Y0TBCjwQ5jf/rZzzo1zznV0zkU75zor7EXqr6JXTnxMFLFRkVw8rFvlvDt+snes/ofPHsRxvfbfln/CI1N4ecYqJszPZc2WXZXTb379O8rKHVOW5vHBvPWc+vg0P+2FBBs9xFykhZh398lERRgJsVHMXbONmKgIzhvShcS4KKYt21Tv9bw3dz3vzV1f6/yi4lI2FOxizdZdtEuIZWDn1k1RvgQBDa0gEiImLtjA1ys2c+bhnejVPpEnJ+Xw6szVbK+lKae+su8azS1vzGVYZio3jKx9YDkJjKC5aNtQCnyRplVaVs6Vr2QzZWk+b14znKxuKXwwfz03vT630et89uIj6ZGeyJdL87n8mAwKd5fyVc4mxg7s2ISVS30p8EWkTtmrtjB5aR5PTV4OwLw/nMzgew9ugDjwDTl91XHdGdEnfb+PhPx80UYiImBU3/bsKS1XF9EmoMAXkXpxzuEcREQYKzftYORfptS5fJe2rapdAK7NuUd25pFzB1Ne7sgt3M0PG7bTsU0cYx7zXSB+5uIjuebV2Yy/ahjDe6Q2xa6ELQW+iDTKpqJilm0somPrOCYu3MCDHy8B4PlLshjeI5WSsnIOu/ezA6zF57yszryZvfaAy2m46IOjh5iLSKO0S4xleI9UMtolVLtxa2TfdBJio2gTH8P7vzyGeX84udr3Prn5+GpdR4F6hT34LjYv27id61+bzRvf/sgJj0zmoudn8erXq3ns8x+46hXfSeDjXywj4/YJlJUHz0lqS6MzfBGp1fdrC5jz41YuPTqjxrxde8pYlFvIpqJiTunfgY2Fu7no+VksyyuqXOaq47pz4qHtucAb8rmxvr1zNEP+/DkA/zemL9eN6MGE+bm0T46lpMwxvEcq67ftIrdgd+WjK+v8exOPAAAJpUlEQVQycUEuvdsnkZmWeFB1BQM16YhIwHyycAPXvDqbUX3TefGyITjnmLliMynxMaQlxXL6E9NZX7C7Sbf5z4uO5Np/zwbg6uMz2bpjDw+ePYjI/TxoxjlH9zs+IirCyLl/bJPWEQhBMbSCiISnHt5Z80n92gO+u4aP7rH3buAZd5xY+b6s3HHlv75l8tJ8MlLjWbV5Z6O2WRH2AM9OXQHAR9/nsuCPp7Aot5ApS/Pp3T6JAZ2SSYz1xV5pGDYN6QxfRJrclh17SImPrveDW9Zu3Ul6UhyLcws546mvSIiJ5KvbR/HiV6t4/ItlNZb/67mD+fVb8w643rjoCHaXlNe5zMg+adwx9lB6pSdiZuwoLmXnnjLSkmKZvXorh3VpQ7lzREcG5yVPneGLSEC13c+zAOrSOcX3/N7BXdqw4v6xlc/8jarSJPP2dcM5+x8zATjjsENYsqGQt+esY+vOPVQ9b22fHMvGwmKAA4Y9wOSl+Uxemk9cdAQPnjWIm9/Y/01pVxzbnTEDOjAkoy0VJ8ot7UlkOsMXkaBVVFzKXz5Zym/H9CE+JopXZq6iX8dksjLaVltu3LNfk5mWwI2jetEuMYbnpq3koYlL/F5f3w5JHNalDelJsbRvHceFR3WrddkdxaVM/SGfMQM6NOkvCl20FZGwVlbueGH6Cu7/aAlXH59Z2a5/2yl9uGBIF8yMX785l5P7d2BIRgqj/zaVTm1asW7bgW8qq8tZR3Ri8pI8Th98CPlFxbSKjuKBswYyc8Vm7nrve9Zs2cVdpx5Kj7REerVPpHNKPN/9uJXBndtU/lXTUAp8EZEqcvKKuObVbF6/ejhpSbE15u8oLqVVdCQTF27g+tfqfuh8q+hIrh/Rg5dmrGLLjj0H3PZVx3XnuWkra0xPT4rl96f148bx3/HbMX24fkTjBqZT4IuIHIRde8ood45W0ZG8MH0lPdIT+MXLvmyaePNx9O2QzO6SMh78eAkvz1jVJNts7B3HCnwRkWZQUlbO7NVbGZaZSnFpGXNWb+PpKTn89dzBbNtVwsmPTq1c9pvfnUhOfhEbC3dzyxs1exg1R+Crl46ISCNFR0YwLNM3+FtsVCTDe6RWDgaXnhzHUz8/gtyCXVwyPIOYqAjSk32Psvx04UY+XrCBtgkx9WoWaioKfBERPzl10P6fEVBx89f9Zw6kuLSs8mY1f1Pgi4g0s9t/0peUhBhG9U1v1mcCKPBFRJpZamIsvxt7aLNvNzjvFRYRkSanwBcRCRMKfBGRMKHAFxEJEwp8EZEwocAXEQkTCnwRkTChwBcRCRNBNXiameUDqxv59XbApiYspyXQPocH7XPoO5j97eacS6vPgkEV+AfDzLLrO2JcqNA+hwftc+hrrv1Vk46ISJhQ4IuIhIlQCvxnA11AAGifw4P2OfQ1y/6GTBu+iIjULZTO8EVEpA4KfBGRMNHiA9/MxpjZUjPLMbPbA11PUzGzLmY22cwWmdlCM7vJm97WzD4zs2XevynedDOzx72fw3wzOyKwe9B4ZhZpZt+Z2Yfe5+5mNsvbtzfMLMabHut9zvHmZwSy7sYyszZm9l8zW2Jmi81seKgfZzO7xfvveoGZjTezuFA7zmb2opnlmdmCKtMafFzN7FJv+WVmdunB1NSiA9/MIoGngJ8A/YBxZtYvsFU1mVLg1865fsAw4AZv324HvnDO9QK+8D6D72fQy3tdDfyj+UtuMjcBi6t8fgh41DnXE9gKXOFNvwLY6k1/1FuuJfo7MNE51xcYjG/fQ/Y4m1kn4FdAlnNuABAJXEDoHeeXgTH7TGvQcTWztsDdwFHAUODuil8SjeKca7EvYDjwSZXPdwB3BLouP+3r/4CTgKVAR29aR2Cp9/4ZYFyV5SuXa0kvoLP3P8Io4EPA8N2BGLXvMQc+AYZ776O85SzQ+9DA/W0NrNy37lA+zkAnYA3Q1jtuHwKnhOJxBjKABY09rsA44Jkq06st19BXiz7DZ+9/OBXWetNCivcn7OHALKC9cy7Xm7UBaO+9D5WfxWPAb4Fy73MqsM05V+p9rrpflfvszS/wlm9JugP5wEteM9bzZpZACB9n59w64C/Aj0AuvuM2m9A+zhUaelyb9Hi39MAPeWaWCLwN3OycK6w6z/l+5YdMv1ozOw3Ic87NDnQtzSgKOAL4h3PucGAHe//MB0LyOKcAZ+D7ZXcIkEDNpo+QF4jj2tIDfx3Qpcrnzt60kGBm0fjC/jXn3Dve5I1m1tGb3xHI86aHws/iGOB0M1sFvI6vWefvQBszi/KWqbpflfvszW8NbG7OgpvAWmCtc26W9/m/+H4BhPJxHg2sdM7lO+dKgHfwHftQPs4VGnpcm/R4t/TA/xbo5V3dj8F34ef9ANfUJMzMgBeAxc65v1WZ9T5QcaX+Unxt+xXTL/Gu9g8DCqr86dgiOOfucM51ds5l4DuWk5xzFwKTgXO8xfbd54qfxTne8i3qTNg5twFYY2Z9vEknAosI4eOMrylnmJnFe/+dV+xzyB7nKhp6XD8BTjazFO8vo5O9aY0T6IsaTXBRZCzwA7AcuDPQ9TThfh2L78+9+cBc7zUWX9vlF8Ay4HOgrbe84euxtBz4Hl8PiIDvx0Hs/wjgQ+99JvANkAO8BcR60+O8zzne/MxA193IfT0MyPaO9XtASqgfZ+CPwBJgAfAqEBtqxxkYj+8aRQm+v+SuaMxxBX7h7XsOcPnB1KShFUREwkRLb9IREZF6UuCLiIQJBb6ISJhQ4IuIhAkFvohImFDgS1gxszIzm1vl1WQjrJpZRtWREUWCTdSBFxEJKbucc4cFugiRQNAZvghgZqvM7GEz+97MvjGznt70DDOb5I1R/oWZdfWmtzezd81snvc62ltVpJk95431/qmZtQrYTonsQ4Ev4abVPk0651eZV+CcGwg8iW/UToAngH855wYBrwGPe9MfB750zg3GN/bNQm96L+Ap51x/YBtwtp/3R6TedKethBUzK3LOJe5n+ipglHNuhTdo3QbnXKqZbcI3fnmJNz3XOdfOzPKBzs654irryAA+c76HW2Bm/wdEO+fu8/+eiRyYzvBF9nK1vG+I4irvy9B1MgkiCnyRvc6v8u9M7/0MfCN3AlwITPPefwFcB5XP4G3dXEWKNJbOPiTctDKzuVU+T3TOVXTNTDGz+fjO0sd5027E9zSq2/A9mepyb/pNwLNmdgW+M/nr8I2MKBK01IYvQmUbfpZzblOgaxHxFzXpiIiECZ3hi4iECZ3hi4iECQW+iEiYUOCLiIQJBb6ISJhQ4IuIhIn/B0Q+fvnX4jrdAAAAAElFTkSuQmCC\n",
265 |       "text/plain": [
266 |        "<Figure size 432x288 with 1 Axes>"
267 |       ]
268 |      },
269 |      "metadata": {},
270 |      "output_type": "display_data"
271 |     }
272 |    ],
273 |    "source": [
274 |     "epochs = np.arange(len(losses))\n",
275 |     "plt.title(r'Loss per Epoch')\n",
276 |     "plt.xlabel('Epoch')\n",
277 |     "plt.ylabel('Loss')\n",
278 |     "plt.plot(epochs, losses)\n",
279 |     "plt.show()"
280 |    ]
281 |   },
282 |   {
283 |    "cell_type": "markdown",
284 |    "metadata": {},
285 |    "source": [
286 |     "### 3.3: Show Evaluation Results"
287 |    ]
288 |   },
289 |   {
290 |    "cell_type": "code",
291 |    "execution_count": 15,
292 |    "metadata": {},
293 |    "outputs": [
294 |     {
295 |      "name": "stdout",
296 |      "output_type": "stream",
297 |      "text": [
298 |       "{\n",
299 |       "  \"mean_rank\": 188.53191489361703,\n",
300 |       "  \"hits@k\": {\n",
301 |       "    \"1\": 0.18085106382978725,\n",
302 |       "    \"3\": 0.2765957446808511,\n",
303 |       "    \"5\": 0.3049645390070922,\n",
304 |       "    \"10\": 0.39361702127659576\n",
305 |       "  }\n",
306 |       "}\n"
307 |      ]
308 |     }
309 |    ],
310 |    "source": [
311 |     "print(json.dumps(results.results['eval_summary'], indent=2))"
312 |    ]
313 |   }
314 |  ],
315 |  "metadata": {
316 |   "kernelspec": {
317 |    "display_name": "Python 3",
318 |    "language": "python",
319 |    "name": "python3"
320 |   },
321 |   "language_info": {
322 |    "codemirror_mode": {
323 |     "name": "ipython",
324 |     "version": 3
325 |    },
326 |    "file_extension": ".py",
327 |    "mimetype": "text/x-python",
328 |    "name": "python",
329 |    "nbconvert_exporter": "python",
330 |    "pygments_lexer": "ipython3",
331 |    "version": "3.7.0"
332 |   }
333 |  },
334 |  "nbformat": 4,
335 |  "nbformat_minor": 2
336 | }
337 | 


--------------------------------------------------------------------------------
/notebooks/Case Scenario HSDN.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "## Use BioKEEN Programmatically to Train and Evalaute a KGE Model on HSDN"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 1,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "import json\n",
 17 |     "import logging\n",
 18 |     "import os\n",
 19 |     "import sys\n",
 20 |     "import time\n",
 21 |     "import warnings\n",
 22 |     "\n",
 23 |     "import matplotlib\n",
 24 |     "import matplotlib.pyplot as plt\n",
 25 |     "import numpy as np\n",
 26 |     "\n",
 27 |     "import biokeen\n",
 28 |     "import pykeen"
 29 |    ]
 30 |   },
 31 |   {
 32 |    "cell_type": "code",
 33 |    "execution_count": 2,
 34 |    "metadata": {},
 35 |    "outputs": [],
 36 |    "source": [
 37 |     "warnings.filterwarnings('ignore', category=UserWarning)"
 38 |    ]
 39 |   },
 40 |   {
 41 |    "cell_type": "code",
 42 |    "execution_count": 3,
 43 |    "metadata": {},
 44 |    "outputs": [],
 45 |    "source": [
 46 |     "logging.basicConfig(level=logging.INFO)\n",
 47 |     "logging.getLogger('biokeen').setLevel(logging.INFO)"
 48 |    ]
 49 |   },
 50 |   {
 51 |    "cell_type": "code",
 52 |    "execution_count": 4,
 53 |    "metadata": {},
 54 |    "outputs": [
 55 |     {
 56 |      "name": "stdout",
 57 |      "output_type": "stream",
 58 |      "text": [
 59 |       "3.7.0 (default, Jul 23 2018, 20:22:55) \n",
 60 |       "[Clang 9.1.0 (clang-902.0.39.2)]\n"
 61 |      ]
 62 |     }
 63 |    ],
 64 |    "source": [
 65 |     "print(sys.version)"
 66 |    ]
 67 |   },
 68 |   {
 69 |    "cell_type": "code",
 70 |    "execution_count": 5,
 71 |    "metadata": {},
 72 |    "outputs": [
 73 |     {
 74 |      "name": "stdout",
 75 |      "output_type": "stream",
 76 |      "text": [
 77 |       "Sun Jan 20 21:11:49 2019\n"
 78 |      ]
 79 |     }
 80 |    ],
 81 |    "source": [
 82 |     "print(time.asctime())"
 83 |    ]
 84 |   },
 85 |   {
 86 |    "cell_type": "code",
 87 |    "execution_count": 6,
 88 |    "metadata": {},
 89 |    "outputs": [
 90 |     {
 91 |      "name": "stdout",
 92 |      "output_type": "stream",
 93 |      "text": [
 94 |       "PyKEEN Version: 0.0.19-dev\n",
 95 |       "BioKEEN Version: 0.0.12-dev\n"
 96 |      ]
 97 |     }
 98 |    ],
 99 |    "source": [
100 |     "print(f'PyKEEN Version: {pykeen.constants.VERSION}')\n",
101 |     "print(f'BioKEEN Version: {biokeen.constants.VERSION}')"
102 |    ]
103 |   },
104 |   {
105 |    "cell_type": "code",
106 |    "execution_count": 7,
107 |    "metadata": {},
108 |    "outputs": [],
109 |    "source": [
110 |     "output_directory = os.path.join(\n",
111 |     "    os.path.expanduser('~'), \n",
112 |     "    'Desktop', \n",
113 |     "    'biokeen_test'\n",
114 |     ")"
115 |    ]
116 |   },
117 |   {
118 |    "cell_type": "markdown",
119 |    "metadata": {},
120 |    "source": [
121 |     "## Step 1: Configure your experiment"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "code",
126 |    "execution_count": 8,
127 |    "metadata": {},
128 |    "outputs": [],
129 |    "source": [
130 |     "config = dict(\n",
131 |     "    training_set_path           = 'bio2bel:hsdn',\n",
132 |     "    execution_mode              = 'Training_mode',  \n",
133 |     "    kg_embedding_model_name     = 'TransE',\n",
134 |     "    embedding_dim               = 50, \n",
135 |     "    normalization_of_entities   = 2,  # corresponds to L2\n",
136 |     "    scoring_function            = 1,  # corresponds to L1\n",
137 |     "    margin_loss                 = 1,\n",
138 |     "    learning_rate               = 0.01,\n",
139 |     "    batch_size                  = 128,\n",
140 |     "    num_epochs                  = 1000,  \n",
141 |     "    test_set_ratio              = 0.1,\n",
142 |     "    filter_negative_triples     = True,\n",
143 |     "    random_seed                 = 2,\n",
144 |     "    preferred_device            = 'cpu',\n",
145 |     ")"
146 |    ]
147 |   },
148 |   {
149 |    "cell_type": "markdown",
150 |    "metadata": {},
151 |    "source": [
152 |     "## Step 2: Run BioKEEN to Train and Evaluate the Model"
153 |    ]
154 |   },
155 |   {
156 |    "cell_type": "code",
157 |    "execution_count": 9,
158 |    "metadata": {},
159 |    "outputs": [
160 |     {
161 |      "name": "stderr",
162 |      "output_type": "stream",
163 |      "text": [
164 |       "INFO:pykeen.utilities.pipeline:-------------Train KG Embeddings-------------\n",
165 |       "Training epoch: 100%|██████████| 1000/1000 [04:36<00:00,  3.61it/s]\n",
166 |       "INFO:pykeen.utilities.pipeline:-------------Start Evaluation-------------\n",
167 |       "INFO:pykeen.utilities.evaluation_utils.metrics_computations:Evaluation took 10.69s seconds\n"
168 |      ]
169 |     }
170 |    ],
171 |    "source": [
172 |     "results = pykeen.run(\n",
173 |     "    config=config,\n",
174 |     "    output_directory=output_directory,\n",
175 |     ")"
176 |    ]
177 |   },
178 |   {
179 |    "cell_type": "code",
180 |    "execution_count": 10,
181 |    "metadata": {},
182 |    "outputs": [
183 |     {
184 |      "name": "stdout",
185 |      "output_type": "stream",
186 |      "text": [
187 |       "Keys:\n",
188 |       "  entity_to_embedding\n",
189 |       "  entity_to_id\n",
190 |       "  eval_summary\n",
191 |       "  final_configuration\n",
192 |       "  losses\n",
193 |       "  relation_to_embedding\n",
194 |       "  relation_to_id\n",
195 |       "  trained_model\n"
196 |      ]
197 |     }
198 |    ],
199 |    "source": [
200 |     "print('Keys:', *sorted(results.results.keys()), sep='\\n  ')"
201 |    ]
202 |   },
203 |   {
204 |    "cell_type": "markdown",
205 |    "metadata": {},
206 |    "source": [
207 |     "## Step 3: Show Exported Results"
208 |    ]
209 |   },
210 |   {
211 |    "cell_type": "markdown",
212 |    "metadata": {},
213 |    "source": [
214 |     "### 3.1: Show Trained Model"
215 |    ]
216 |   },
217 |   {
218 |    "cell_type": "code",
219 |    "execution_count": 11,
220 |    "metadata": {},
221 |    "outputs": [
222 |     {
223 |      "data": {
224 |       "text/plain": [
225 |        "TransE(\n",
226 |        "  (criterion): MarginRankingLoss()\n",
227 |        "  (entity_embeddings): Embedding(433, 50)\n",
228 |        "  (relation_embeddings): Embedding(1, 50)\n",
229 |        ")"
230 |       ]
231 |      },
232 |      "execution_count": 11,
233 |      "metadata": {},
234 |      "output_type": "execute_result"
235 |     }
236 |    ],
237 |    "source": [
238 |     "results.results['trained_model']"
239 |    ]
240 |   },
241 |   {
242 |    "cell_type": "markdown",
243 |    "metadata": {},
244 |    "source": [
245 |     "### 3.2: Plot losses"
246 |    ]
247 |   },
248 |   {
249 |    "cell_type": "code",
250 |    "execution_count": 12,
251 |    "metadata": {},
252 |    "outputs": [],
253 |    "source": [
254 |     "losses = results.results['losses']"
255 |    ]
256 |   },
257 |   {
258 |    "cell_type": "code",
259 |    "execution_count": 13,
260 |    "metadata": {},
261 |    "outputs": [
262 |     {
263 |      "data": {
264 |       "image/png": "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\n",
265 |       "text/plain": [
266 |        "<Figure size 432x288 with 1 Axes>"
267 |       ]
268 |      },
269 |      "metadata": {},
270 |      "output_type": "display_data"
271 |     }
272 |    ],
273 |    "source": [
274 |     "epochs = np.arange(len(losses))\n",
275 |     "plt.title(r'Loss Per Epoch')\n",
276 |     "plt.xlabel('epoch')\n",
277 |     "plt.ylabel('loss')\n",
278 |     "plt.plot(epochs, losses)\n",
279 |     "plt.show()"
280 |    ]
281 |   },
282 |   {
283 |    "cell_type": "markdown",
284 |    "metadata": {},
285 |    "source": [
286 |     "### 3.3: Show Evaluation Results"
287 |    ]
288 |   },
289 |   {
290 |    "cell_type": "code",
291 |    "execution_count": 14,
292 |    "metadata": {},
293 |    "outputs": [
294 |     {
295 |      "name": "stdout",
296 |      "output_type": "stream",
297 |      "text": [
298 |       "{\n",
299 |       "  \"mean_rank\": 24.59951219512195,\n",
300 |       "  \"hits@k\": {\n",
301 |       "    \"1\": 0.12097560975609756,\n",
302 |       "    \"3\": 0.2531707317073171,\n",
303 |       "    \"5\": 0.3297560975609756,\n",
304 |       "    \"10\": 0.4678048780487805\n",
305 |       "  }\n",
306 |       "}\n"
307 |      ]
308 |     }
309 |    ],
310 |    "source": [
311 |     "print(json.dumps(results.results['eval_summary'], indent=2))"
312 |    ]
313 |   }
314 |  ],
315 |  "metadata": {
316 |   "kernelspec": {
317 |    "display_name": "Python 3",
318 |    "language": "python",
319 |    "name": "python3"
320 |   },
321 |   "language_info": {
322 |    "codemirror_mode": {
323 |     "name": "ipython",
324 |     "version": 3
325 |    },
326 |    "file_extension": ".py",
327 |    "mimetype": "text/x-python",
328 |    "name": "python",
329 |    "nbconvert_exporter": "python",
330 |    "pygments_lexer": "ipython3",
331 |    "version": "3.7.0"
332 |   }
333 |  },
334 |  "nbformat": 4,
335 |  "nbformat_minor": 2
336 | }
337 | 


--------------------------------------------------------------------------------
/notebooks/README.rst:
--------------------------------------------------------------------------------
 1 | Notebooks
 2 | =========
 3 | This folder contains the following notebooks:
 4 | 
 5 | Tutorials
 6 | ---------
 7 | 1. How to use BioKEEN
 8 | 
 9 | Application Scenarios
10 | ---------------------
11 | There are several tasks that can be completed with biological knowledge graph embeddings.
12 | This section links notebooks that show several of those scenarios.
13 | 
14 | Link Prediction
15 | ~~~~~~~~~~~~~~~
16 | 1. Case Scenario applied to the ComPath dataset.
17 | 2. Case Scenario applied to the ADEPTUS dataset.
18 | 3. Case Scenario applied to the HSDN dataset.
19 | 4. Case Scenario applied to union of the HSDN and ADEPTUS datasets.
20 | 
21 | Entity Disambiguation
22 | ---------------------
23 | This is the task of identifying when several nodes from different KGs that have been
24 | merged refer to the same entity. We haven't got a use case for this yet.
25 | 
26 | Link-Based Clustering
27 | ---------------------
28 | This is the task of clustering nodes based on their embeddings. We haven't got a use
29 | case for this yet.
30 | 


--------------------------------------------------------------------------------
/requirements-rtd.txt:
--------------------------------------------------------------------------------
1 | pybel<0.14.0
2 | bio2bel<0.3.0
3 | tqdm
4 | pandas
5 | sphinx
6 | sphinx-rtd-theme
7 | sphinx-click
8 | sphinx-autodoc-typehints
9 | 


--------------------------------------------------------------------------------
/setup.cfg:
--------------------------------------------------------------------------------
  1 | ##########################
  2 | # Setup.py Configuration #
  3 | ##########################
  4 | # Configuring setup()
  5 | [metadata]
  6 | name = biokeen
  7 | version = 0.0.15-dev
  8 | description = A package for training and evaluating biological knowledge graph embeddings
  9 | long_description = file: README.rst
 10 | 
 11 | url = https://github.com/SmartDataAnalytics/BioKEEN
 12 | download_url = https://github.com/SmartDataAnalytics/BioKEEN/releases
 13 | project_urls =
 14 |     Bug Tracker = https://github.com/SmartDataAnalytics/BioKEEN/issues
 15 |     Documentation = https://biokeen.readthedocs.io
 16 | 
 17 | author = Mehdi Ali
 18 | author_email = mehdi.ali@cs.uni-bonn.de
 19 | maintainer = Mehdi Ali
 20 | maintainer_email = mehdi.ali@cs.uni-bonn.de
 21 | 
 22 | license = MIT
 23 | license_file = LICENSE
 24 | 
 25 | classifiers =
 26 |     Development Status :: 1 - Planning
 27 |     Intended Audience :: Developers
 28 |     License :: OSI Approved :: MIT License
 29 |     Programming Language :: Python
 30 |     Programming Language :: Python :: 3.7
 31 |     Programming Language :: Python :: 3.6
 32 |     Programming Language :: Python :: 3 :: Only
 33 | keywords =
 34 |     KEEN
 35 |     knowledge embedding
 36 | 
 37 | [options]
 38 | install_requires =
 39 |     mock
 40 |     pykeen<1.0.0
 41 |     pybel<0.14.0
 42 |     bio2bel<0.3.0
 43 |     tqdm
 44 |     pandas
 45 |     click-default-group
 46 |     click
 47 | 
 48 | zip_safe = false
 49 | python_requires = >=3.6
 50 | packages = find:
 51 | package_dir =
 52 |     = src
 53 | 
 54 | [options.extras_require]
 55 | docs =
 56 |     sphinx
 57 |     sphinx-rtd-theme
 58 |     sphinx-click
 59 |     sphinx_autodoc_typehints
 60 | 
 61 | [options.packages.find]
 62 | where = src
 63 | 
 64 | [options.entry_points]
 65 | console_scripts =
 66 |     biokeen = biokeen.cli:main
 67 | pykeen.data.importer =
 68 | 	bio2bel = biokeen.content:handle_bio2bel
 69 | 	bel_commons = biokeen.content:handle_bel_commons
 70 | 
 71 | ######################
 72 | # Doc8 Configuration #
 73 | # (doc8.ini)         #
 74 | ######################
 75 | [doc8]
 76 | max-line-length = 120
 77 | 
 78 | ##########################
 79 | # Coverage Configuration #
 80 | # (.coveragerc)          #
 81 | ##########################
 82 | [coverage:run]
 83 | branch = True
 84 | source = biokeen
 85 | omit =
 86 |     tests/*
 87 |     docs/*
 88 |     src/biokeen/cli/*
 89 |     src/biokeen/__main__.py
 90 | 
 91 | [coverage:paths]
 92 | source =
 93 |     src/biokeen
 94 |     .tox/*/lib/python*/site-packages/biokeen
 95 | 
 96 | [coverage:report]
 97 | show_missing = True
 98 | exclude_lines =
 99 |     def __str__
100 |     def __repr__
101 | 


--------------------------------------------------------------------------------
/setup.py:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | 
3 | """Setup module for BioKEEN."""
4 | 
5 | import setuptools
6 | 
7 | if __name__ == '__main__':
8 |     setuptools.setup()
9 | 


--------------------------------------------------------------------------------
/src/biokeen/__init__.py:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | 
3 | """A package for training and evaluating knowledge graph embedding models on biological knowledge graphs."""
4 | 
5 | from . import convert  # noqa: 401
6 | 


--------------------------------------------------------------------------------
/src/biokeen/__main__.py:
--------------------------------------------------------------------------------
 1 | # -*- coding: utf-8 -*-
 2 | 
 3 | """Entrypoint module, in case you use ``python -m biokeen``.
 4 | 
 5 | Why does this file exist, and why ``__main__``? For more info, read:
 6 |  - https://www.python.org/dev/peps/pep-0338/
 7 |  - https://docs.python.org/3/using/cmdline.html#cmdoption-m
 8 | """
 9 | 
10 | from .cli import main
11 | 
12 | if __name__ == '__main__':
13 |     main()
14 | 


--------------------------------------------------------------------------------
/src/biokeen/cli/__init__.py:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | 
3 | """The command line interface for BioKEEN."""
4 | 
5 | from .cli import main  # noqa: 401
6 | 


--------------------------------------------------------------------------------
/src/biokeen/cli/__main__.py:
--------------------------------------------------------------------------------
 1 | # -*- coding: utf-8 -*-
 2 | 
 3 | """Entrypoint module, in case you use ``python -m biokeen.cli``.
 4 | 
 5 | Why does this file exist, and why ``__main__``? For more info, read:
 6 |  - https://www.python.org/dev/peps/pep-0338/
 7 |  - https://docs.python.org/3/using/cmdline.html#cmdoption-m
 8 | """
 9 | 
10 | from .cli import main
11 | 
12 | if __name__ == '__main__':
13 |     main()
14 | 


--------------------------------------------------------------------------------
/src/biokeen/cli/cli.py:
--------------------------------------------------------------------------------
  1 | # -*- coding: utf-8 -*-
  2 | 
  3 | """A command line interface for BioKEEN.
  4 | 
  5 | Why does this file exist, and why not put this in ``__main__``? You might be tempted to import things from ``__main__``
  6 | later, but that will cause problems--the code will get executed twice:
  7 | - When you run ``python3 -m biokeen`` python will execute``__main__.py`` as a script. That means there won't be any
  8 |   ``biokeen.__main__`` in ``sys.modules``.
  9 | - When you import __main__ it will get executed again (as a module) because
 10 |   there's no ``biokeen.__main__`` in ``sys.modules``.
 11 | .. seealso:: http://click.pocoo.org/5/setuptools/#setuptools-integration
 12 | """
 13 | 
 14 | import json
 15 | import logging
 16 | import os
 17 | from typing import List, Optional, TextIO
 18 | 
 19 | import click
 20 | from click_default_group import DefaultGroup
 21 | 
 22 | from bio2bel.constants import get_global_connection
 23 | from biokeen.constants import EMOJI, VERSION, biokeen_config
 24 | from pykeen.constants import VERSION as PYKEEN_VERSION
 25 | 
 26 | connection_option = click.option(
 27 |     '-c',
 28 |     '--connection',
 29 |     default=get_global_connection(),
 30 |     show_default=True,
 31 |     help='Bio2BEL database connection string',
 32 | )
 33 | 
 34 | 
 35 | @click.group(cls=DefaultGroup, default_if_no_args=False)
 36 | @click.version_option()
 37 | def main():  # noqa: D401
 38 |     """A command line interface for BioKEEN."""
 39 | 
 40 | 
 41 | @main.command()
 42 | @connection_option
 43 | @click.option('-f', '--config', type=click.File())
 44 | @click.option('-r', '--rebuild', is_flag=True)
 45 | @click.option('-x', '--no-prompt-bio2bel', is_flag=True)
 46 | def start(config: Optional[TextIO], connection: str, rebuild: bool, no_prompt_bio2bel: bool):
 47 |     """Start the BioKEEN training pipeline."""
 48 |     import pykeen
 49 | 
 50 |     if config is not None:
 51 |         config = json.load(config)
 52 |     else:
 53 |         from .prompts import prompt_biokeen_config
 54 |         config = prompt_biokeen_config(
 55 |             connection=connection,
 56 |             rebuild=rebuild,
 57 |             do_prompt_bio2bel=(not no_prompt_bio2bel),
 58 |         )
 59 | 
 60 |     config['pykeen-version'] = PYKEEN_VERSION
 61 |     config['biokeen-version'] = VERSION
 62 |     pykeen.run(config)
 63 | 
 64 | 
 65 | @main.command()
 66 | @click.option('-m', '--model-directory', type=click.Path(file_okay=False, dir_okay=True, exists=True), required=True)
 67 | @click.option('-d', '--data-directory', type=click.Path(file_okay=False, dir_okay=True, exists=True), required=True)
 68 | def predict(model_directory: str, data_directory: str):
 69 |     """Use a trained model to make predictions."""
 70 |     from pykeen.predict import start_predictions_pipeline
 71 |     start_predictions_pipeline(model_directory, data_directory)
 72 | 
 73 | 
 74 | @main.group()
 75 | def data():
 76 |     """Commands for data acquisition."""
 77 | 
 78 | 
 79 | @data.command(help=f'Data stored in {biokeen_config.data_directory}')
 80 | def ls():
 81 |     """List built data."""
 82 |     for path in biokeen_config.iterate_source_paths():
 83 |         click.echo(path)
 84 | 
 85 | 
 86 | @data.command()
 87 | @click.confirmation_option()
 88 | def clear():
 89 |     """Remove all built data."""
 90 |     for path in biokeen_config.iterate_source_paths():
 91 |         os.remove(path)
 92 | 
 93 | 
 94 | @data.command()
 95 | @click.argument('names', nargs=-1)
 96 | @connection_option
 97 | @click.option('-r', '--rebuild', is_flag=True)
 98 | @click.option('-v', '--verbose', count=True)
 99 | def get(names: List[str], connection: str, rebuild: bool, verbose: bool):
100 |     """Install, populate, and build Bio2BEL repository."""
101 |     if verbose == 1:
102 |         logging.basicConfig(level=logging.INFO)
103 |     elif verbose == 2:
104 |         logging.basicConfig(level=logging.DEBUG)
105 | 
106 |     from biokeen.content import install_bio2bel_module
107 | 
108 |     for name in names:
109 |         click.secho(f'{EMOJI} Getting {name}', fg='cyan')
110 |         install_bio2bel_module(name, connection, rebuild)
111 | 
112 | 
113 | if __name__ == '__main__':
114 |     main()
115 | 


--------------------------------------------------------------------------------
/src/biokeen/cli/messages.py:
--------------------------------------------------------------------------------
 1 | # -*- coding: utf-8 -*-
 2 | 
 3 | """BioKEEN's command line interface helper."""
 4 | 
 5 | import click
 6 | 
 7 | 
 8 | def print_welcome_message():
 9 |     """Print welcome message."""
10 |     click.secho('#################################################')
11 |     click.secho("#\t\tWelcome to " + click.style("BioKEEN", bold=True) + "\t\t#")
12 |     click.secho('#################################################')
13 | 
14 | 
15 | def print_intro():
16 |     """Print intro."""
17 |     click.secho("This interface will assist you to configure your experiment.")
18 |     click.secho("")
19 |     click.secho(
20 |         "BioKEEN can be run in two modes: \n"
21 |         "1.) Training mode: BioKEEN trains a model based on a set of user-defined hyper-parameters.\n"
22 |         "2.) Hyper-parameter optimization mode: "
23 |         "Apply Random Search to determine the most appropriate set of hyper-parameter values"
24 |     )
25 | 


--------------------------------------------------------------------------------
/src/biokeen/cli/prompts.py:
--------------------------------------------------------------------------------
  1 | # -*- coding: utf-8 -*-
  2 | 
  3 | """Prompts for the BioKEEN command line interface."""
  4 | 
  5 | import os
  6 | import re
  7 | from collections import OrderedDict
  8 | from typing import Dict, Iterable, Optional
  9 | 
 10 | import click
 11 | 
 12 | from pykeen.cli.prompt import prompt_config
 13 | from pykeen.cli.utils.cli_print_msg_helper import print_section_divider
 14 | from pykeen.constants import TRAINING_SET_PATH
 15 | from .messages import print_intro, print_welcome_message
 16 | from ..constants import ID_TO_DATABASE_MAPPING
 17 | from ..content import install_bio2bel_module
 18 | 
 19 | __all__ = [
 20 |     'prompt_biokeen_config',
 21 | ]
 22 | 
 23 | 
 24 | def prompt_biokeen_config(*, connection: str, rebuild: bool, do_prompt_bio2bel: Optional[bool] = None) -> Dict:
 25 |     """Configure experiments."""
 26 |     config = OrderedDict()
 27 | 
 28 |     # Step 1: Welcome + Intro
 29 |     print_welcome_message()
 30 |     print_section_divider()
 31 |     print_intro()
 32 |     print_section_divider()
 33 | 
 34 |     # Step 2: Ask for data source
 35 |     if do_prompt_bio2bel is None:
 36 |         do_prompt_bio2bel = click.confirm('Do you want to use one of the databases provided by BioKEEN?', default=True)
 37 |         print_section_divider()
 38 | 
 39 |     do_prompt_training = True
 40 | 
 41 |     if do_prompt_bio2bel:
 42 |         do_prompt_training = False
 43 |         config[TRAINING_SET_PATH] = []
 44 |         for name in select_bio2bel_repository():
 45 |             try:
 46 |                 path = install_bio2bel_module(name=name, connection=connection, rebuild=rebuild)
 47 |             except Exception:
 48 |                 click.secho(f'failed: {name}', fg='red')
 49 |             else:
 50 |                 if os.path.exists(path):
 51 |                     config[TRAINING_SET_PATH].append(f'bio2bel:{name}')
 52 |                 else:
 53 |                     click.secho(f'failed: {name}: {path}', fg='red')
 54 | 
 55 |         # TODO replace this with less safe code that assumes everything installs no problemo
 56 |         """
 57 |         config[TRAINING_SET_PATH] = [
 58 |             f'bio2bel:{name}'
 59 |             for name in select_bio2bel_repository()
 60 |         ]
 61 |         """
 62 | 
 63 |     print_section_divider()
 64 | 
 65 |     return prompt_config(
 66 |         config=config,
 67 |         show_welcome=False,
 68 |         do_prompt_training=do_prompt_training,
 69 |     )
 70 | 
 71 | 
 72 | def select_bio2bel_repository() -> Iterable[str]:
 73 |     """Prompt the user for a Bio2BEL database."""
 74 |     click.secho("Current Step: Please select the database(s) you want to train on:", fg='blue')
 75 | 
 76 |     number_width = 1 + round(len(ID_TO_DATABASE_MAPPING) / 10)
 77 |     for model, model_id in sorted(ID_TO_DATABASE_MAPPING.items()):
 78 |         click.echo(f'{model: >{number_width}}: {model_id}')
 79 | 
 80 |     while True:
 81 |         try:
 82 |             user_input = click.prompt('> Please select one or more of the options', value_proc=process_selection)
 83 |         except ValueError as e:
 84 |             click.secho(str(e), fg='red')
 85 |         else:
 86 |             return user_input
 87 | 
 88 | 
 89 | def process_selection(values: str) -> Iterable[str]:
 90 |     for value in re.split(r'\s|,', values):
 91 |         value = value.strip()
 92 |         if not value:
 93 |             continue
 94 | 
 95 |         try:
 96 |             value = int(value)
 97 |         except ValueError:
 98 |             if value not in ID_TO_DATABASE_MAPPING.values():
 99 |                 raise ValueError(f'{value} is an invalid database')
100 |             yield value
101 |         else:
102 |             if value not in ID_TO_DATABASE_MAPPING:
103 |                 raise ValueError(f'{value} is an invalid index')
104 |             yield ID_TO_DATABASE_MAPPING[value]
105 | 


--------------------------------------------------------------------------------
/src/biokeen/constants.py:
--------------------------------------------------------------------------------
 1 | # -*- coding: utf-8 -*-
 2 | 
 3 | """Constants for BioKEEN."""
 4 | 
 5 | import os
 6 | from typing import Iterable
 7 | 
 8 | import easy_config
 9 | 
10 | HERE = os.path.abspath(os.path.dirname(__file__))
11 | HOME = os.path.expanduser('~')
12 | 
13 | 
14 | class BiokeenConfig(easy_config.EasyConfig):
15 |     """Configuration for BioKEEN."""
16 | 
17 |     NAME = 'biokeen'
18 |     FILES = [
19 |         os.path.join(HOME, '.config', 'biokeen.cfg'),
20 |         os.path.join(HOME, '.config', 'config.ini'),
21 |     ]
22 | 
23 |     #: the data directory where TSVs get exported
24 |     data_directory: str = os.path.abspath(os.path.join(HOME, '.keen', 'biokeen'))
25 | 
26 |     #: The file extension of pre-processed Bio2BEL databases
27 |     keen_tsv_ext: str = 'keen.tsv'
28 | 
29 |     def iterate_source_paths(self) -> Iterable[str]:
30 |         """Iterate over the source paths."""
31 |         for file_name in os.listdir(self.data_directory):
32 |             if file_name.endswith(self.keen_tsv_ext):
33 |                 yield os.path.join(self.data_directory, file_name)
34 | 
35 | 
36 | biokeen_config = BiokeenConfig.load()
37 | os.makedirs(biokeen_config.data_directory, exist_ok=True)
38 | 
39 | VERSION = '0.0.15-dev'
40 | EMOJI = '🍩'
41 | 
42 | # Available databases
43 | COMPATH_NAME = 'compath'
44 | HIPPIE_NAME = 'hippie'
45 | KEGG_NAME = 'kegg'
46 | MIRTARBASE_NAME = 'mirtarbase'
47 | MSIG_NAME = 'msig'
48 | REACTOME_NAME = 'reactome'
49 | SIDER_NAME = 'sider'
50 | WIKIPATHWAYS_NAME = 'wikipathways'
51 | DRUGBANK_NAME = 'drugbank'
52 | ADEPTUS_NAME = 'adeptus'
53 | HSDN_NAME = 'hsdn'
54 | INTERPRO_NAME = 'interpro'
55 | DDR_NAME = 'ddr'
56 | 
57 | # ToDo: Add databases
58 | DATABASES = [
59 |     COMPATH_NAME,
60 |     HIPPIE_NAME,
61 |     KEGG_NAME,
62 |     MIRTARBASE_NAME,
63 |     MSIG_NAME,
64 |     REACTOME_NAME,
65 |     SIDER_NAME,
66 |     WIKIPATHWAYS_NAME,
67 |     DRUGBANK_NAME,
68 |     ADEPTUS_NAME,
69 |     HSDN_NAME,
70 |     INTERPRO_NAME,
71 |     DDR_NAME,
72 | ]
73 | 
74 | ID_TO_DATABASE_MAPPING = dict(enumerate(DATABASES, start=1))
75 | 


--------------------------------------------------------------------------------
/src/biokeen/content.py:
--------------------------------------------------------------------------------
  1 | # -*- coding: utf-8 -*-
  2 | 
  3 | """Utilities for BioKEEN."""
  4 | 
  5 | import importlib
  6 | import logging
  7 | import os
  8 | import sys
  9 | from contextlib import redirect_stdout
 10 | from typing import Optional, Union
 11 | 
 12 | import numpy as np
 13 | import pkg_resources
 14 | 
 15 | from bio2bel import AbstractManager
 16 | from bio2bel.manager.bel_manager import BELManagerMixin
 17 | from pybel import from_json_path, from_web, to_json_path
 18 | from .constants import EMOJI, biokeen_config
 19 | from .convert import to_pykeen_df, to_pykeen_path, to_pykeen_summary_path
 20 | 
 21 | _SPECIAL_CASES = {
 22 |     'compath': 'compath_resources',
 23 | }
 24 | 
 25 | logger = logging.getLogger(__name__)
 26 | 
 27 | 
 28 | def install_bio2bel_module(name: str, connection: Optional[str] = None, rebuild: bool = False) -> Optional[str]:
 29 |     """Install Bio2BEL module.
 30 | 
 31 |     :param name: The name of the Bio2BEL module
 32 |     :param connection: The optional database connection
 33 |     :param rebuild: Should the cache not be used? Defaults to False.
 34 |     """
 35 |     module_name = _SPECIAL_CASES.get(name, f'bio2bel_{name}')
 36 | 
 37 |     pykeen_df_path = os.path.join(biokeen_config.data_directory, f'{name}.{biokeen_config.keen_tsv_ext}')
 38 |     pykeen_df_summary_path = os.path.join(biokeen_config.data_directory, f'{name}.keen.summary.json')
 39 |     json_path = os.path.join(biokeen_config.data_directory, f'{name}.bel.json')
 40 | 
 41 |     if os.path.exists(pykeen_df_path) and not rebuild:
 42 |         logger.info(f'{EMOJI} {module_name} has already been retrieved. See: {pykeen_df_path}')
 43 |         return pykeen_df_path
 44 | 
 45 |     if os.path.exists(json_path) and not rebuild:
 46 |         logger.info(f'{EMOJI} loaded {module_name} JSON: {json_path}')
 47 |         graph = from_json_path(json_path)
 48 |         df = to_pykeen_df(graph)
 49 |         to_pykeen_path(df, pykeen_df_path)
 50 |         to_pykeen_summary_path(df, pykeen_df_summary_path)
 51 |         return pykeen_df_path
 52 | 
 53 |     bio2bel_module = ensure_bio2bel_installation(module_name)
 54 |     logger.debug(f'{EMOJI} imported {module_name}')
 55 | 
 56 |     manager_cls = bio2bel_module.Manager
 57 | 
 58 |     if not issubclass(manager_cls, BELManagerMixin):
 59 |         version = pkg_resources.get_distribution(module_name).version
 60 |         logger.warning(f'{EMOJI} {module_name} v{version} does not produce BEL')
 61 |         sys.exit(1)
 62 | 
 63 |     manager = manager_cls(connection=connection)
 64 | 
 65 |     if issubclass(manager_cls, AbstractManager):
 66 |         if not manager.is_populated():
 67 |             logger.info(f'{EMOJI} populating {module_name}')
 68 |             manager.populate()
 69 |         else:
 70 |             logger.debug(f'{EMOJI} {module_name} has already been populated')
 71 | 
 72 |     logger.debug(f'{EMOJI} generating BEL for {module_name}')
 73 |     graph = manager.to_bel()
 74 | 
 75 |     logger.debug(f'Summary: {graph.number_of_nodes()} nodes / {graph.number_of_edges()} edges')
 76 |     to_json_path(graph, json_path, indent=2)
 77 | 
 78 |     logger.debug(f'{EMOJI} generating PyKEEN TSV for {module_name}')
 79 |     df = to_pykeen_df(graph)
 80 |     to_pykeen_summary_path(df, pykeen_df_summary_path)
 81 |     success = to_pykeen_path(df, pykeen_df_path)
 82 | 
 83 |     if success:
 84 |         logger.debug(f'{EMOJI} wrote PyKEEN TSV to {pykeen_df_path}')
 85 |         return pykeen_df_path
 86 | 
 87 |     logger.warning(f'{EMOJI} no statements generated')
 88 | 
 89 | 
 90 | def ensure_bio2bel_installation(package: str):
 91 |     """Import a package, or install it."""
 92 |     try:
 93 |         b_module = importlib.import_module(package)
 94 | 
 95 |     except ImportError:
 96 |         logger.info(f'{EMOJI} pip install {package}')
 97 |         # Install this package using pip
 98 |         # https://stackoverflow.com/questions/12332975/installing-python-module-within-code
 99 |         from pip._internal import main as pip_main
100 | 
101 |         with redirect_stdout(sys.stderr):
102 |             pip_exit_code = pip_main(['install', '-q', package])  # -q means quiet
103 | 
104 |         if 0 != pip_exit_code:  # command failed
105 |             logger.warning(f'{EMOJI} could not find {package} on PyPI. Try installing from GitHub with:')
106 |             name = package.split("_")[-1]
107 |             logger.warning(f'\n   pip install git+https://github.com/bio2bel/{name}.git\n')
108 |             sys.exit(1)
109 | 
110 |         try:
111 |             return importlib.import_module(package)
112 |         except ImportError:
113 |             logger.exception(f'{EMOJI} failed to import {package}')
114 |             sys.exit(1)
115 | 
116 |     return b_module
117 | 
118 | 
119 | BIO2BEL_PREFIX = 'bio2bel'
120 | 
121 | 
122 | def handle_bio2bel(module_name: str) -> np.ndarray:
123 |     """Load a Bio2BEL repository.
124 | 
125 |     :param module_name: The name of the bio2bel repository (with no prefix)
126 |     """
127 |     path = install_bio2bel_module(module_name)
128 |     return np.loadtxt(
129 |         fname=path,
130 |         dtype=str,
131 |         comments='@Comment@ Subject Predicate Object',
132 |         delimiter='\t',
133 |     )
134 | 
135 | 
136 | def handle_bel_commons(network_id: Union[int, str], host: Optional[str] = None) -> np.ndarray:
137 |     """Load a BEL document from BEL Commons.
138 | 
139 |     :param network_id: The network identifier in BEL Commons
140 |     :param host: The host for BEL Commons. Defaults to the Fraunhofer SCAI public instance.
141 |     """
142 |     graph = from_web(int(network_id), host=host)
143 |     df = to_pykeen_df(graph)
144 |     return df.to_numpy()
145 | 


--------------------------------------------------------------------------------
/src/biokeen/convert/__init__.py:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | 
3 | """Conversion from BEL to proper triples."""
4 | 
5 | from .io import get_pykeen_summary, get_triple, to_pykeen_df, to_pykeen_path, to_pykeen_summary_path  # noqa: F401
6 | 


--------------------------------------------------------------------------------
/src/biokeen/convert/converters.py:
--------------------------------------------------------------------------------
  1 | # -*- coding: utf-8 -*-
  2 | 
  3 | """Conversion base classes."""
  4 | 
  5 | from abc import ABC, abstractmethod
  6 | from typing import Dict, Tuple
  7 | 
  8 | from pybel.constants import (
  9 |     ACTIVITY, ASSOCIATION, CORRELATIVE_RELATIONS, DECREASES, DIRECTLY_DECREASES, DIRECTLY_INCREASES, EQUIVALENT_TO,
 10 |     HAS_COMPONENT, INCREASES, IS_A, MODIFIER, OBJECT, PART_OF, REGULATES, RELATION,
 11 | )
 12 | from pybel.dsl import (
 13 |     Abundance, BaseAbundance, BaseEntity, BiologicalProcess, ComplexAbundance, MicroRna,
 14 |     NamedComplexAbundance, Pathology, Protein, Rna,
 15 | )
 16 | 
 17 | 
 18 | class Converter(ABC):
 19 |     """A condition and converter for a BEL edge."""
 20 | 
 21 |     @staticmethod
 22 |     @abstractmethod
 23 |     def predicate(u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> bool:
 24 |         """Test a BEL edge."""
 25 | 
 26 |     @staticmethod
 27 |     @abstractmethod
 28 |     def convert(u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> Tuple[str, str, str]:
 29 |         """Convert a BEL edge."""
 30 | 
 31 | 
 32 | class SimpleConverter(Converter):
 33 |     """A class for converting the source and target that have simple names."""
 34 | 
 35 |     @classmethod
 36 |     def convert(cls, u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> Tuple[str, str, str]:
 37 |         """Convert a BEL edge."""
 38 |         return (
 39 |             f'{u.namespace}:{u.identifier or u.name}',
 40 |             edge_data[RELATION],
 41 |             f'{v.namespace}:{v.identifier or v.name}',
 42 |         )
 43 | 
 44 | 
 45 | class TypedConverter(Converter):
 46 |     """A class for converting the source and target but replaces the relation."""
 47 | 
 48 |     target_relation = None
 49 | 
 50 |     @classmethod
 51 |     def convert(cls, u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> Tuple[str, str, str]:
 52 |         """Convert a BEL edge."""
 53 |         return (
 54 |             f'{u.namespace}:{u.identifier or u.name}',
 55 |             cls.target_relation,
 56 |             f'{v.namespace}:{v.identifier or v.name}',
 57 |         )
 58 | 
 59 | 
 60 | class SimplePredicate(Converter):
 61 |     """Converts BEL statements based on a given relation."""
 62 | 
 63 |     relation = ...
 64 | 
 65 |     @classmethod
 66 |     def predicate(cls, u, v, key, edge_data) -> bool:
 67 |         """Test a BEL edge has a given relation."""
 68 |         return edge_data[RELATION] == cls.relation
 69 | 
 70 | 
 71 | class SimpleTypedPredicate(SimplePredicate):
 72 |     """Finds BEL statements like ``A(X) B C(Y)`` where relation B and types A and C are defined in the class."""
 73 | 
 74 |     subject_type = ...
 75 |     object_type = ...
 76 | 
 77 |     @classmethod
 78 |     def predicate(cls, u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> bool:
 79 |         """Test a BEL edge."""
 80 |         return super().predicate(u, v, key, edge_data) and (
 81 |             isinstance(u, cls.subject_type) and
 82 |             isinstance(v, cls.object_type)
 83 |         )
 84 | 
 85 | 
 86 | class _PartOfConverter(SimpleTypedPredicate, TypedConverter):
 87 |     relation = PART_OF
 88 |     target_relation = 'partOf'
 89 | 
 90 | 
 91 | class PartOfNamedComplexConverter(_PartOfConverter):
 92 |     """Converts BEL statements like ``p(X) partOf complex(Y)``."""
 93 | 
 94 |     subject_type = Protein
 95 |     object_type = NamedComplexAbundance
 96 | 
 97 | 
 98 | class SubprocessPartOfBiologicalProcess(_PartOfConverter):
 99 |     """Converts BEL statements like ``bp(X) partOf bp(Y)``."""
100 | 
101 |     subject_type = BiologicalProcess
102 |     object_type = BiologicalProcess
103 | 
104 | 
105 | class ProteinPartOfBiologicalProcess(_PartOfConverter):
106 |     """Converts BEL statements like ``p(X) partOf bp(Y)``."""
107 | 
108 |     subject_type = Protein
109 |     object_type = BiologicalProcess
110 | 
111 | 
112 | class NamedComplexHasComponentConverter(SimpleTypedPredicate):
113 |     """Converts BEL statements like ``complex(X) hasComponent p(Y)``."""
114 | 
115 |     subject_type = NamedComplexAbundance
116 |     relation = HAS_COMPONENT
117 |     object_type = Protein
118 |     target_relation = 'partOf'
119 | 
120 |     @classmethod
121 |     def convert(cls, u: BaseEntity, v: BaseEntity, key: str, data: Dict) -> Tuple[str, str, str]:
122 |         """Convert a BEL edge."""
123 |         return (
124 |             f'{v.namespace}:{v.identifier or v.name}',
125 |             cls.target_relation,
126 |             f'{u.namespace}:{u.identifier or u.name}',
127 |         )
128 | 
129 | 
130 | class ListComplexHasComponentConverter(SimpleTypedPredicate):
131 |     """Converts BEL statements like ``complex(p(X), p(Y), ...) hasComponent p(X)``."""
132 | 
133 |     subject_type = ComplexAbundance
134 |     relation = HAS_COMPONENT
135 |     object_type = Protein
136 |     target_relation = 'partOf'
137 | 
138 |     @classmethod
139 |     def convert(cls, u: ComplexAbundance, v: BaseAbundance, key: str, data: Dict) -> Tuple[str, str, str]:
140 |         """Convert a BEL edge."""
141 |         return (
142 |             f'{v.namespace}:{v.identifier or v.name}',
143 |             cls.target_relation,
144 |             str(u),
145 |         )
146 | 
147 | 
148 | '''
149 | class TranslationConverter(TypedConverter):
150 |     relation = TRANSLATED_TO
151 |     target_relation = 'ribosomallyTranslatesTo'
152 | 
153 | 
154 | class TranscriptionConverter(TypedConverter):
155 |     relation = TRANSCRIBED_TO
156 |     target_relation = 'ribosomallyTranslatesTo'
157 | '''
158 | 
159 | 
160 | class IsAConverter(SimplePredicate, SimpleConverter):
161 |     """Converts BEL statements like ``X isA Y``."""
162 | 
163 |     relation = IS_A
164 |     target_relation = 'isA'
165 | 
166 | 
167 | class EquivalenceConverter(SimplePredicate, SimpleConverter):
168 |     """Converts BEL statements like ``X eq Y``."""
169 | 
170 |     relation = EQUIVALENT_TO
171 |     target_relation = 'equivalentTo'
172 | 
173 | 
174 | class CorrelationConverter(SimpleConverter):
175 |     """Converts BEL statements like ``A(B) pos|neg|noCorrelation C(D)``."""
176 | 
177 |     @staticmethod
178 |     def predicate(u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> bool:
179 |         """Test a BEL edge."""
180 |         return edge_data[RELATION] in CORRELATIVE_RELATIONS
181 | 
182 | 
183 | class AssociationConverter(Converter):
184 |     """Converts BEL statements like ``a(X) -- path(Y)``."""
185 | 
186 |     @staticmethod
187 |     def predicate(u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> bool:
188 |         """Test a BEL edge."""
189 |         return edge_data[RELATION] == ASSOCIATION
190 | 
191 |     @staticmethod
192 |     def convert(u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> Tuple[str, str, str]:
193 |         """Convert a BEL edge."""
194 |         return (
195 |             f'{u.namespace}:{u.identifier or u.name}',
196 |             edge_data.get('association_type', ASSOCIATION),  # allow more specific association to be defined
197 |             f'{v.namespace}:{v.identifier or v.name}',
198 |         )
199 | 
200 | 
201 | class DrugEffectConverter(SimpleConverter, SimpleTypedPredicate):
202 |     """Converts BEL statements like ``a(X) ? path(Y)``."""
203 | 
204 |     subject_type = Abundance
205 |     relation = ...
206 |     object_type = Pathology
207 | 
208 | 
209 | class DrugIndicationConverter(DrugEffectConverter):
210 |     """Converts BEL statements like ``a(X) -| path(Y)``."""
211 | 
212 |     relation = DECREASES
213 | 
214 | 
215 | class DrugSideEffectConverter(DrugEffectConverter):
216 |     """Converts BEL statements like ``a(X) -> path(Y)``."""
217 | 
218 |     relation = INCREASES
219 | 
220 | 
221 | class RegulatesAmountConverter(TypedConverter):
222 |     """Converts BEL statements like ``A(B) reg C(D)``."""
223 | 
224 |     relation = REGULATES
225 |     target_relation = 'regulatesAmountOf'
226 | 
227 |     @classmethod
228 |     def predicate(cls, u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> bool:
229 |         """Test a BEL edge."""
230 |         object_modifier = edge_data.get(OBJECT)
231 |         return edge_data[RELATION] == cls.relation and (not object_modifier or not object_modifier.get(MODIFIER))
232 | 
233 | 
234 | class IncreasesAmountConverter(RegulatesAmountConverter):
235 |     """Converts BEL statements like ``A(B) -> C(D)``."""
236 | 
237 |     relation = INCREASES
238 |     target_relation = 'increasesAmountOf'
239 | 
240 | 
241 | class DecreasesAmountConverter(RegulatesAmountConverter):
242 |     """Converts BEL statements like ``A(B) -| C(D)``."""
243 | 
244 |     relation = DECREASES
245 |     target_relation = 'decreasesAmountOf'
246 | 
247 | 
248 | class RegulatesActivityConverter(TypedConverter):
249 |     """Converts BEL statements like ``A(B) reg act(C(D) [, ma(E)])``."""
250 | 
251 |     relation = REGULATES
252 |     target_relation = 'activityDirectlyRegulatesActivityOf'
253 | 
254 |     @classmethod
255 |     def predicate(cls, u: BaseEntity, v: BaseEntity, key: str, edge_data: Dict) -> bool:
256 |         """Test a BEL edge."""
257 |         object_modifier = edge_data.get(OBJECT)
258 |         return edge_data[RELATION] == cls.relation and object_modifier and object_modifier.get(MODIFIER) == ACTIVITY
259 | 
260 | 
261 | class IncreasesActivityConverter(RegulatesActivityConverter):
262 |     """Converts BEL statements like ``A(B) -> act(C(D) [, ma(E)])``."""
263 | 
264 |     relation = INCREASES
265 |     target_relation = 'activityDirectlyPositivelyRegulatesActivityOf'
266 | 
267 | 
268 | class DecreasesActivityConverter(RegulatesActivityConverter):
269 |     """Converts BEL statements like ``A(B) -| act(C(D) [, ma(E)])``."""
270 | 
271 |     relation = DECREASES
272 |     target_relation = 'activityDirectlyNegativelyRegulatesActivityOf'
273 | 
274 | 
275 | class MiRNARegulatesExpressionConverter(TypedConverter, SimpleTypedPredicate):
276 |     """Converts BEL statements like ``m(X) reg r(Y)``."""
277 | 
278 |     subject_type = MicroRna
279 |     relation = REGULATES
280 |     object_type = Rna
281 |     target_relation = 'regulatesExpressionOf'
282 | 
283 | 
284 | class MiRNAIncreasesExpressionConverter(MiRNARegulatesExpressionConverter):
285 |     """Converts BEL statements like ``m(X) -> r(Y)``."""
286 | 
287 |     relation = INCREASES
288 |     target_relation = 'increasesExpressionOf'
289 | 
290 | 
291 | class MiRNADirectlyIncreasesExpressionConverter(MiRNARegulatesExpressionConverter):
292 |     """Converts BEL statements like ``m(X) => r(Y)``."""
293 | 
294 |     relation = DIRECTLY_INCREASES
295 |     target_relation = 'increasesExpressionOf'
296 | 
297 | 
298 | class MiRNADecreasesExpressionConverter(MiRNARegulatesExpressionConverter):
299 |     """Converts BEL statements like ``m(X) -| r(Y)``."""
300 | 
301 |     relation = DECREASES
302 |     target_relation = 'repressesExpressionOf'
303 | 
304 | 
305 | class MiRNADirectlyDecreasesExpressionConverter(MiRNARegulatesExpressionConverter):
306 |     """Converts BEL statements like ``m(X) =| r(Y)``."""
307 | 
308 |     relation = DIRECTLY_DECREASES
309 |     target_relation = 'repressesExpressionOf'
310 | 


--------------------------------------------------------------------------------
/src/biokeen/convert/io.py:
--------------------------------------------------------------------------------
  1 | # -*- coding: utf-8 -*-
  2 | 
  3 | """Input and output for BEL conversion."""
  4 | 
  5 | import itertools as itt
  6 | import json
  7 | import logging
  8 | from collections import Counter
  9 | from typing import Dict, Optional, Tuple
 10 | 
 11 | import pandas as pd
 12 | from tqdm import tqdm
 13 | 
 14 | from pybel import BELGraph
 15 | from pybel.dsl import BaseEntity
 16 | from .converters import (
 17 |     AssociationConverter, CorrelationConverter, DecreasesAmountConverter, DrugIndicationConverter,
 18 |     DrugSideEffectConverter, EquivalenceConverter, IncreasesAmountConverter, IsAConverter,
 19 |     ListComplexHasComponentConverter, MiRNADecreasesExpressionConverter, MiRNADirectlyDecreasesExpressionConverter,
 20 |     NamedComplexHasComponentConverter, PartOfNamedComplexConverter, ProteinPartOfBiologicalProcess,
 21 |     RegulatesActivityConverter, RegulatesAmountConverter, SubprocessPartOfBiologicalProcess,
 22 | )
 23 | from ..constants import EMOJI
 24 | 
 25 | __all__ = [
 26 |     'to_pykeen_path',
 27 |     'to_pykeen_df',
 28 |     'get_pykeen_summary',
 29 |     'to_pykeen_summary_path',
 30 |     'get_triple',
 31 | ]
 32 | 
 33 | logger = logging.getLogger(__name__)
 34 | 
 35 | 
 36 | def to_pykeen_path(df: pd.DataFrame, path: str) -> bool:
 37 |     """Write the relationships in the BEL graph to a KEEN TSV file.
 38 | 
 39 |     If you have a BEL graph, first do:
 40 | 
 41 |     >>> from biokeen.convert import to_pykeen_df, to_pykeen_path
 42 |     >>> graph = ...  # Something from PyBEL
 43 |     >>> df = to_pykeen_df(graph)
 44 |     >>> to_pykeen_path(df, 'graph.keen.tsv')
 45 |     """
 46 |     if len(df.index) == 0:
 47 |         return False
 48 |     df.to_csv(path, sep='\t', index=None, header=None)
 49 |     return True
 50 | 
 51 | 
 52 | def get_pykeen_summary(df: pd.DataFrame) -> Dict:
 53 |     """Summarize a KEEN dataframe."""
 54 |     entity_count = Counter(itt.chain(df[df.columns[0]], df[df.columns[2]]))
 55 |     return {
 56 |         'namespaces': Counter(
 57 |             element.split(':')[0]
 58 |             for element in itt.chain(df[df.columns[0]], df[df.columns[2]])
 59 |         ),
 60 |         'entities': len(entity_count),
 61 |         'relations': len(df.index),
 62 |     }
 63 | 
 64 | 
 65 | def to_pykeen_summary_path(df: pd.DataFrame, path: str, indent=2, **kwargs):
 66 |     """Write the summary of a KEEN dataframe to a file."""
 67 |     with open(path, 'w') as file:
 68 |         json.dump(get_pykeen_summary(df), file, indent=indent, **kwargs)
 69 | 
 70 | 
 71 | def to_pykeen_df(graph: BELGraph, use_tqdm: bool = True) -> pd.DataFrame:
 72 |     """Get a DataFrame representing the triples."""
 73 |     it = graph.edges(keys=True)
 74 | 
 75 |     if use_tqdm:
 76 |         it = tqdm(it, total=graph.number_of_edges(), desc=f'{EMOJI} preparing TSV')
 77 | 
 78 |     triples = (
 79 |         get_triple(graph, u, v, key)
 80 |         for u, v, key in it
 81 |     )
 82 | 
 83 |     # clean duplicates and Nones
 84 |     triples = list(sorted({triple for triple in triples if triple is not None}))
 85 | 
 86 |     return pd.DataFrame(triples, columns=['subject', 'predicate', 'object'])
 87 | 
 88 | 
 89 | def get_triple(graph: BELGraph, u: BaseEntity, v: BaseEntity, key: str) -> Optional[Tuple[str, str, str]]:  # noqa: C901
 90 |     """Get the triples' strings that should be written to the file."""
 91 |     data = graph[u][v][key]
 92 | 
 93 |     # order is important
 94 |     converters = [
 95 |         NamedComplexHasComponentConverter,
 96 |         ListComplexHasComponentConverter,
 97 |         PartOfNamedComplexConverter,
 98 |         SubprocessPartOfBiologicalProcess,
 99 |         ProteinPartOfBiologicalProcess,
100 |         RegulatesActivityConverter,
101 |         MiRNADecreasesExpressionConverter,
102 |         MiRNADirectlyDecreasesExpressionConverter,
103 |         IsAConverter,
104 |         EquivalenceConverter,
105 |         CorrelationConverter,
106 |         AssociationConverter,
107 |         DrugIndicationConverter,
108 |         DrugSideEffectConverter,
109 |         RegulatesAmountConverter,
110 |         IncreasesAmountConverter,
111 |         DecreasesAmountConverter,
112 |     ]
113 | 
114 |     for converter in converters:
115 |         if converter.predicate(u, v, key, data):
116 |             return converter.convert(u, v, key, data)
117 | 
118 |     logger.warning(f'{EMOJI} unhandled: {graph.edge_to_bel(u, v, data)}')
119 |     return None
120 | 


--------------------------------------------------------------------------------
/tests/__init__.py:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | 
3 | """Tests for BioKEEN."""
4 | 


--------------------------------------------------------------------------------
/tests/test_convert.py:
--------------------------------------------------------------------------------
  1 | # -*- coding: utf-8 -*-
  2 | 
  3 | """Tests for the conversion procedure."""
  4 | 
  5 | import unittest
  6 | from typing import Tuple, Type
  7 | 
  8 | from biokeen.convert import get_triple
  9 | from biokeen.convert.converters import (
 10 |     AssociationConverter, Converter, CorrelationConverter, DecreasesAmountConverter, DrugIndicationConverter,
 11 |     DrugSideEffectConverter, EquivalenceConverter, IncreasesAmountConverter, IsAConverter,
 12 |     MiRNADecreasesExpressionConverter, NamedComplexHasComponentConverter,
 13 |     PartOfNamedComplexConverter, RegulatesActivityConverter, RegulatesAmountConverter, SubprocessPartOfBiologicalProcess
 14 | )
 15 | from pybel import BELGraph
 16 | from pybel.constants import (
 17 |     ASSOCIATION, DECREASES, EQUIVALENT_TO, HAS_COMPONENT, INCREASES, IS_A, NEGATIVE_CORRELATION, OBJECT, PART_OF,
 18 |     POSITIVE_CORRELATION, REGULATES, RELATION,
 19 | )
 20 | from pybel.dsl import (
 21 |     Abundance, BaseEntity, BiologicalProcess, MicroRna, NamedComplexAbundance, Pathology, Protein,
 22 |     Rna, activity,
 23 | )
 24 | from pybel.testing.utils import n
 25 | from pybel.typing import EdgeData
 26 | 
 27 | 
 28 | def _rel(x):
 29 |     return {RELATION: x}
 30 | 
 31 | 
 32 | def _rela(x, y=None):
 33 |     return {RELATION: x, OBJECT: activity(y)}
 34 | 
 35 | 
 36 | def _assoc(y):
 37 |     return {RELATION: ASSOCIATION, 'association_type': y}
 38 | 
 39 | 
 40 | a1 = Abundance('CHEBI', '1')
 41 | p1 = Protein('HGNC', '1')
 42 | pf1 = Protein('INTERPRO', '1')
 43 | d1 = Pathology('MESH', '1')
 44 | b1 = BiologicalProcess('GO', '1')
 45 | b2 = BiologicalProcess('GO', '2')
 46 | m1 = MicroRna('MIRBASE', '1')
 47 | r1 = Rna('HGNC', '1')
 48 | r2 = Rna('HGNC', '2')
 49 | nca1 = NamedComplexAbundance('FPLX', '1')
 50 | 
 51 | converters_true_list = [
 52 |     (NamedComplexHasComponentConverter, nca1, p1, _rel(HAS_COMPONENT), ('HGNC:1', 'partOf', 'FPLX:1')),
 53 |     (PartOfNamedComplexConverter, p1, nca1, _rel(PART_OF), ('HGNC:1', 'partOf', 'FPLX:1')),
 54 |     (SubprocessPartOfBiologicalProcess, b1, b2, _rel(PART_OF), ('GO:1', 'partOf', 'GO:2')),
 55 |     (AssociationConverter, r1, r2, _rel(ASSOCIATION), ('HGNC:1', 'association', 'HGNC:2')),
 56 |     (AssociationConverter, r1, r2, _assoc('similarity'), ('HGNC:1', 'similarity', 'HGNC:2')),
 57 |     (CorrelationConverter, r1, r2, _rel(POSITIVE_CORRELATION), ('HGNC:1', 'positiveCorrelation', 'HGNC:2')),
 58 |     (IsAConverter, p1, pf1, _rel(IS_A), ('HGNC:1', 'isA', 'INTERPRO:1')),
 59 |     # Found in ADEPTUS
 60 |     (CorrelationConverter, d1, r1, _rel(POSITIVE_CORRELATION), ('MESH:1', 'positiveCorrelation', 'HGNC:1')),
 61 |     (CorrelationConverter, d1, r1, _rel(NEGATIVE_CORRELATION), ('MESH:1', 'negativeCorrelation', 'HGNC:1')),
 62 |     # Found in LINCS (not integrated yet)
 63 |     (RegulatesAmountConverter, a1, r1, _rel(REGULATES), ('CHEBI:1', 'regulatesAmountOf', 'HGNC:1')),
 64 |     (IncreasesAmountConverter, a1, r1, _rel(INCREASES), ('CHEBI:1', 'increasesAmountOf', 'HGNC:1')),
 65 |     (DecreasesAmountConverter, a1, r1, _rel(DECREASES), ('CHEBI:1', 'decreasesAmountOf', 'HGNC:1')),
 66 |     # Found in SIDER
 67 |     (DrugSideEffectConverter, a1, d1, _rel(INCREASES), ('CHEBI:1', 'increases', 'MESH:1')),
 68 |     (DrugIndicationConverter, a1, d1, _rel(DECREASES), ('CHEBI:1', 'decreases', 'MESH:1')),
 69 |     # Found in miRTarBase
 70 |     (MiRNADecreasesExpressionConverter, m1, r1, _rel(DECREASES), ('MIRBASE:1', 'repressesExpressionOf', 'HGNC:1')),
 71 |     # Found in DrugBank
 72 |     (RegulatesActivityConverter, a1, p1, _rela(REGULATES), ('CHEBI:1', 'activityDirectlyRegulatesActivityOf',
 73 |                                                             'HGNC:1')),
 74 |     # Found in ComPath
 75 |     (EquivalenceConverter, b1, b2, _rel(EQUIVALENT_TO), ('GO:1', 'equivalentTo', 'GO:2')),
 76 |     (SubprocessPartOfBiologicalProcess, b1, b2, _rel(PART_OF), ('GO:1', 'partOf', 'GO:2')),
 77 |     # Found in HSDN
 78 | ]
 79 | 
 80 | converters_false_list = [
 81 |     (NamedComplexHasComponentConverter, nca1, p1, _rel(PART_OF)),
 82 |     (PartOfNamedComplexConverter, nca1, p1, _rel(HAS_COMPONENT)),
 83 | ]
 84 | 
 85 | 
 86 | class TestConverters(unittest.TestCase):
 87 |     """Tests for the converter classes."""
 88 | 
 89 |     def help_test_convert(self,
 90 |                           converter: Type[Converter],
 91 |                           u: BaseEntity,
 92 |                           v: BaseEntity,
 93 |                           edge_data: EdgeData,
 94 |                           triple: Tuple[str, str, str],
 95 |                           ) -> None:
 96 |         """Test a converter class."""
 97 |         self.assertTrue(issubclass(converter, Converter), msg=f'Not a Converter: {converter.__name__}')
 98 |         key = n()
 99 |         self.assertTrue(
100 |             converter.predicate(u, v, key, edge_data),
101 |             msg=f'Predicate failed: {converter.__name__}',
102 |         )
103 |         self.assertEqual(
104 |             triple,
105 |             converter.convert(u, v, key, edge_data),
106 |             msg=f'Conversion failed: {converter.__name__}',
107 |         )
108 |         graph = BELGraph()
109 |         graph.add_edge(u, v, key=key, **edge_data)
110 |         self.assertEqual(
111 |             triple,
112 |             get_triple(graph, u, v, key),
113 |             msg=f'get_triple failed: {converter.__name__}',
114 |         )
115 | 
116 |     def test_converters_true(self):
117 |         """Test passing converters."""
118 |         for converter, u, v, edge_data, triple in converters_true_list:
119 |             with self.subTest(msg=f'Converter: {converter.__qualname__}'):
120 |                 self.help_test_convert(converter, u, v, edge_data, triple)
121 | 
122 |     def test_converters_false(self):
123 |         """Test falsification of converters."""
124 |         for converter, u, v, edge_data in converters_false_list:
125 |             with self.subTest():
126 |                 self.assertFalse(converter.predicate(u, v, n(), edge_data))
127 | 


--------------------------------------------------------------------------------
/tests/test_simple.py:
--------------------------------------------------------------------------------
 1 | # -*- coding: utf-8 -*-
 2 | 
 3 | """Simple tests for BioKEEN."""
 4 | 
 5 | import unittest
 6 | 
 7 | from biokeen.constants import VERSION
 8 | 
 9 | 
10 | class TestImport(unittest.TestCase):
11 |     """Simple tests for importing BioKEEN."""
12 | 
13 |     def test_version_type(self):
14 |         """Test the type of the version string."""
15 |         self.assertIsInstance(VERSION, str)
16 | 


--------------------------------------------------------------------------------
/tox.ini:
--------------------------------------------------------------------------------
  1 | [tox]
  2 | envlist =
  3 |     # always keep coverage-clean first
  4 |     coverage-clean
  5 |     # code linters/stylers
  6 |     manifest
  7 |     flake8
  8 |     pyroma
  9 |     xenon
 10 |     # documentation linters/checkers
 11 |     doc8
 12 |     readme
 13 |     docs
 14 |     # the actual tests
 15 |     py
 16 |     # always keep coverage-report last
 17 |     coverage-report
 18 | 
 19 | [testenv]
 20 | commands = coverage run -p -m pytest --durations=20 {posargs:tests}
 21 | passenv = KEEN_TEST_CONNECTOR KEEN_TEST_CONNECTION TRAVIS CI DB
 22 | deps =
 23 |     coverage
 24 |     pytest
 25 |     mock
 26 |     pathlib
 27 |     {env:KEEN_TEST_CONNECTOR:}
 28 | whitelist_externals =
 29 |     /bin/cat
 30 |     /bin/cp
 31 |     /bin/mkdir
 32 |     /usr/bin/git
 33 |     /usr/local/bin/git
 34 | 
 35 | [testenv:coverage-clean]
 36 | deps = coverage
 37 | skip_install = true
 38 | commands = coverage erase
 39 | 
 40 | [testenv:manifest]
 41 | deps = check-manifest
 42 | skip_install = true
 43 | commands = check-manifest
 44 | 
 45 | [testenv:flake8]
 46 | skip_install = true
 47 | deps =
 48 |     flake8
 49 |     flake8-bandit
 50 |     flake8-colors
 51 |     flake8-docstrings
 52 |     flake8-import-order
 53 |     pep8-naming
 54 | commands =
 55 |     flake8 src/biokeen/ tests/ setup.py
 56 | description = Run the flake8 tool with several plugins (bandit, docstrings, import order, pep8 naming).
 57 | 
 58 | [testenv:xenon]
 59 | deps = xenon
 60 | skip_install = true
 61 | commands = xenon --max-average A --max-modules A --max-absolute B .
 62 | description = Run the xenon tool to monitor code complexity.
 63 | 
 64 | [testenv:pyroma]
 65 | deps =
 66 |     pygments
 67 |     pyroma
 68 | skip_install = true
 69 | commands = pyroma --min=10 .
 70 | description = Run the pyroma tool to check the project's package friendliness.
 71 | 
 72 | [testenv:doc8]
 73 | skip_install = true
 74 | deps =
 75 |     sphinx
 76 |     doc8
 77 | commands =
 78 |     doc8 docs/source/ README.rst
 79 | description = Run the doc8 tool to check the style of the RST files in the project docs.
 80 | 
 81 | [testenv:readme]
 82 | commands = rst-lint README.rst
 83 | skip_install = true
 84 | deps =
 85 |     restructuredtext_lint
 86 |     pygments
 87 | 
 88 | [testenv:docs]
 89 | changedir = docs
 90 | extras =
 91 |     docs
 92 | commands =
 93 |     mkdir -p {envtmpdir}
 94 |     cp -r source {envtmpdir}/source
 95 |     sphinx-build -W -b html -d {envtmpdir}/build/doctrees {envtmpdir}/source {envtmpdir}/build/html
 96 |     sphinx-build -W -b coverage -d {envtmpdir}/build/doctrees {envtmpdir}/source {envtmpdir}/build/coverage
 97 |     cat {envtmpdir}/build/coverage/c.txt
 98 |     cat {envtmpdir}/build/coverage/python.txt
 99 | 
100 | [testenv:coverage-report]
101 | deps = coverage
102 | skip_install = true
103 | commands =
104 |     coverage combine
105 |     coverage report
106 | 
107 | ####################
108 | # Deployment tools #
109 | ####################
110 | 
111 | [testenv:bumpversion]
112 | commands = bumpversion {posargs}
113 | skip_install = true
114 | passenv = HOME
115 | deps =
116 |     bumpversion
117 | 
118 | [testenv:build]
119 | skip_install = true
120 | deps =
121 |     wheel
122 |     setuptools
123 | commands =
124 |     python setup.py -q sdist bdist_wheel
125 | 
126 | [testenv:release]
127 | skip_install = true
128 | deps =
129 |     {[testenv:build]deps}
130 |     twine >= 1.5.0
131 | commands =
132 |     {[testenv:build]commands}
133 |     twine upload --skip-existing dist/*
134 | 
135 | [testenv:finish]
136 | skip_install = true
137 | passenv = HOME
138 | deps =
139 |     {[testenv:release]deps}
140 |     bumpversion
141 | commands =
142 |     bumpversion release
143 |     {[testenv:release]commands}
144 |     git push
145 |     bumpversion patch
146 |     git push
147 | 


--------------------------------------------------------------------------------