9 | FIVEx is a tool for browsing expression and splice QTL (cis-eQTL and cis-sQTL) datasets. It provides a number of tools aimed at functional interpretation. 10 | See our preprint for details. 11 |
12 | 13 |14 | Created by Alan Kwong, Mukai Wang, Peter VandeHaar, Andy Boughton, and Hyun Min Kang. 15 | Source code can be found on GitHub. 16 |
17 |24 | We present RNA-seq data aggregated, harmonized, and analyzed by the EBI eQTL Catalogue. See their site for more details of contributing studies or to download raw data. 25 |
26 |27 | Genome positions use coordinates from human genome build GRCh38. 28 |
29 |35 | Study Name 36 | | 37 |38 | Cell/Tissue Type 39 | | 40 |41 | Samples 42 | | 43 |44 | Donors 45 | | 46 |47 | Reference 48 | | 49 |
---|---|---|---|---|
55 | Alasoo_2018 56 | | 57 |58 | macrophages 59 | | 60 |61 | 336 62 | | 63 |64 | 84 65 | | 66 |67 | Alasoo et al. 2018 68 | | 69 |
72 | BLUEPRINT 73 | | 74 |75 | monocytes, neutrophils, CD4+ T cells 76 | | 77 |78 | 554 79 | | 80 |81 | 197 82 | | 83 |84 | Chen et al. 2016 85 | | 86 |
89 | BrainSeq 90 | | 91 |92 | brain (dorsolateral prefrontal cortex) 93 | | 94 |95 | 484 96 | | 97 |98 | 484 99 | | 100 |101 | Jaffe et al. 2018 102 | | 103 |
106 | FUSION 107 | | 108 |109 | adipose, muscle 110 | | 111 |112 | 559 113 | | 114 |115 | 302 116 | | 117 |118 | Taylor et al. 2019 119 | | 120 |
123 | GENCORD 124 | | 125 |126 | LCLs, fibroblasts, T cells 127 | | 128 |129 | 560 130 | | 131 |132 | 195 133 | | 134 |135 | Gutierrez-Arcelus et al. 2013 136 | | 137 |
140 | GEUVADIS 141 | | 142 |143 | LCLs 144 | | 145 |146 | 445 147 | | 148 |149 | 445 150 | | 151 |152 | Lappalainen et al. 2013 153 | | 154 |
157 | GTEx v7 158 | | 159 |160 | 49 tissues 161 | | 162 |163 | 8356 164 | | 165 |166 | 507 167 | | 168 |169 | The GTEx Consortium 2020 170 | | 171 |
174 | HipSci 175 | | 176 |177 | iPSCs 178 | | 179 |180 | 322 181 | | 182 |183 | 322 184 | | 185 |186 | Kilpinen et al. 2017 187 | | 188 |
191 | Lepik_2017 192 | | 193 |194 | blood 195 | | 196 |197 | 491 198 | | 199 |200 | 491 201 | | 202 |203 | Lepik et al. 2017 204 | | 205 |
208 | Nedelec_2016 209 | | 210 |211 | macrophages 212 | | 213 |214 | 493 215 | | 216 |217 | 168 218 | | 219 |220 | Nédélec et al. 2016 221 | | 222 |
225 | Quach_2016 226 | | 227 |228 | monocytes 229 | | 230 |231 | 969 232 | | 233 |234 | 200 235 | | 236 |237 | Quach et al. 2016 238 | | 239 |
242 | ROSMAP 243 | | 244 |245 | brain (dorsolateral prefrontal cortex) 246 | | 247 |248 | 576 249 | | 250 |251 | 576 252 | | 253 |254 | Ng et al. 2017 255 | | 256 |
259 | Schmiedel_2018 260 | | 261 |262 | immune cells 263 | | 264 |265 | 1331 266 | | 267 |268 | 91 269 | | 270 |271 | Schmiedel et al. 2018 272 | | 273 |
276 | Schwartzentruber_2018 277 | | 278 |279 | sensory neurons 280 | | 281 |282 | 98 283 | | 284 |285 | 98 286 | | 287 |288 | Schwartzentruber et al. 2018 289 | | 290 |
293 | TwinsUK 294 | | 295 |296 | adipose, LCLs, skin, blood 297 | | 298 |299 | 1364 300 | | 301 |302 | 433 303 | | 304 |305 | Buil et al. 2015 306 | | 307 |
310 | van_de_Bunt_2015 311 | | 312 |313 | pancreatic islets 314 | | 315 |316 | 117 317 | | 318 |319 | 117 320 | | 321 |322 | van de Bunt et al. 2015 323 | | 324 |
9 | A longstanding issue in using eQTL association results to elucidate biological mechanisms has been the difficulty of identifying true causal variants for gene expression in the presence of linkage disequilibrium (LD). This is further complicated by the use of P-values in evaluating eQTL significance, which may fail to convey uncertainty about the association results. 10 |
11 | 12 |13 | To address these issues, FIVEx incorporates results from SuSiE, a Bayesian variable selection method designed to highlight significantly associated variants in the presence of high correlation (in the form of LD) and quantify the uncertainty of those associations: 14 |
15 | 16 |22 | In the region view, variants which appear to be associated with changes in gene expression are grouped into credible sets. Each credible set represents a set of variants within which an association signal exists. The PIP value of a variant indicates the strength of evidence that the variant is the effect variable in SuSiE's model. 23 |
24 | 25 |26 | In the single-variant view, associations are summarized across many different studies, tissues, and genes. The abliity to group data according to various categories makes it easy to see if the variant has significant association signals in specific tissues or genes, along with the size and direction of their effects. 27 |
28 | 29 |31 | Though eQTLs provide an extensive overview of the relationships between variants and gene expression, it does not contain more detailed information about different transcripts and their related splicing events. Thus, we examine splice QTLs (sQTLs) for a more fine-grained analysis of these splicing events using txrevise (Alasoo et al. 2019). 32 |
33 |34 | Briefly, txrevise identifies two different groupings of exons for any given gene (see Glossary) for downstream analysis. A txrevise event contains 3 pieces of information, gene ID, transcript ID, and grouping, in the following format: 35 |
36 |37 | [gene_ID].grp_[group].contained.[transcript_ID] 38 |
39 |40 | Splice QTL associations are analogous to corresponding eQTL associations, using txrevise events in the place of gene expressions as covariates. 41 |
42 | 43 |8 | 15 |
16 |18 | 25 |
26 |28 | 35 |
36 |38 | 45 |
46 |48 | 55 |
56 |58 | 65 |
66 |