Downloads

A gene table for every genome on the site, built from the database when you ask for it, so it matches what a search returns today.

Gene tables

Tab separated, one row per gene: symbol, description, biotype, Ensembl, UniProt and Entrez identifiers, and the chromosome and coordinates on the assembly named below. Rows are ordered by symbol. Outside the model organisms most genes have no curated symbol and carry their Ensembl identifier in that column instead — 88% of the wheat rows, two thirds of the rice — so they sort in among the named genes rather than apart from them.

The description column is empty for just under half the rows, because most sources publish a description only for the genes they have curated: nearly every wheat and rice row has none, while human, mouse and the other model organisms are close to complete. Where a genome's own source publishes no description for any of its genes, the protein name from UniProt is used instead — that is how Chlamydomonas, which had none at all, now has one on 96% of its rows.

Gene Ontology

The GMT column above gives each genome's GO annotations in the standard GMT layout: one row per term, the term identifier twice, then every gene annotated with it. These are the annotations behind a GO:0006281 search here, so a term returns the same genes in the file as it does on the site.

The annotations were originally downloaded from g:Profiler, one combined GMT per organism, and loaded against each genome's gene table. That is where most of the genomes get theirs. The rest come from elsewhere, either because g:Profiler does not cover the organism or because its file is built on a different assembly, which resolves no genes at all: from Ensembl BioMart's own GO export, from UniProt's GO annotations — the viruses, the bacteria, cotton and the genomes imported from NCBI — and E. coli's from the GO Consortium's EcoCyc release. 168 of the 170 have annotations loaded; the exceptions are named below. The files here are rebuilt from those imports rather than being copies of g:Profiler's own, which has two consequences worth knowing. Only annotations whose genes matched the gene table are included, so a file is a little smaller than the original. And the second column of a GMT is normally the term name, which GeneList does not store, so the identifier is repeated in its place.

For the original files, with the term names and every ontology branch, go to the g:Profiler data page, which is kept current by the people who build them. Two genomes have no GO annotations loaded and so no GMT to download: Red-eared slider and Iberian ribbed newt. They are shown with a dash in the table above.

Where the data comes from

Most gene annotations are retrieved from Ensembl BioMart and are subject to the terms under which Ensembl makes them available. Fifty-five genomes Ensembl either does not carry or carries only on an assembly too old to be worth using — every bacterium, archaeon and virus, but also cotton, the axolotl, the alligator and the bottlenose dolphin among others — come instead from Ensembl Bacteria and NCBI Datasets, whose annotations carry the locus tags those communities use. All of it is a snapshot rather than a live mirror, so for anything that has to be current, take it from the source directly.

A gene list you have searched can also be downloaded from the results panel, which gives you just the genes you asked for.