1/
GO enrichment is a tool. As with most tools, it is often used inappropriately.  Used correctly  GO enrichment can help to provide functional clues, form hypotheses, confirm a hypothesis or to generate a new hypothesis. https://twitter.com/jgschraiber/status/1313254828932427776
2/
GO enrichments should be used judiciously in manuscripts. They should add something to the narrative (why you did something, what it demonstrated, why you did what you did next). If it does not belong in your manuscript DO NOT INCLUDE IT, Durr!
3/ Users @PomBase provide  many great examples of GO enrichment supporting their findings, or leading them to additional genes of interest. Will add some of these later.
4/ Don't let reviewers make you include unnecessary GO enrichments.  Journal editors should override such bonkers reviewer requests, (or you can explain why these are unnecessary or inappropriate). A good reviewer would not ask for unnecessary figures/experiments.
5/
GO enrichments reflect underlying GO annotation. Biocuration is massively underfunded. There are 0.5 million reviewed proteins in Swiss-Prot (the manually curated portion of @uniprot). There are >300 million proteins in the NCBI non-redundant protein database!
6/ 30M publications in PubMed.
Only a few hundred @biocurator 's world-wide (curating all aspects of biology GO, phenotypes, sequence, structures..all of your favourite MODS etc. @FlyBaseDotOrg @wormbase @ratgenome @mgi_mouse @yeastgenome). Just think about that for a minute….
7/ Yes of course GO annotation changes over time, knowledge is evolving- there are massive gaps in our biological  knowledge. We try our best to incorporate new findings in a timely manner.
8/ All members of the @news4go consortium are redoubling efforts to weed out incorrect annotation, outdated science, and improve coverage.
*Please* let us know if you find mistakes or missing information in GO annotation.
9/ You can look up your own publications in the GO database using this URL http://amigo.geneontology.org/amigo/reference/PMID:32735772 and your PubMed ID... Take ownership of your own data/annotation.
10/ Research is biased,(36,266 papers with P53 in the title) and therefore there is an implicit bias in annotation density- ensure that whichever tools you use also recognise the *unstudied* proteins, not only the studied ones- herein lie many Nature papers.
11/
A recent example https://www.pombase.org/reference/PMID:32848252@pombase...saved me countless hours of fruitless experiments and helped open up many new, unexpected avenues of investigation.” @Dey_Gautam (PomBase 2019 survey response)
qualifying 11/ this was a result of targeting 'unstudied' proteins. Do not forget that GO can only tell you about what we already know. Look at the other stuff ffs!
12/ Finally, I challenge anyone to point out incorrect GO annotation @PomBase. We will fix any reported errors on our website https://www.pombase.org/  by the end of this week. Game on.
13/ One more. Always do a 'slim' with your dataset before
an enrichment to assess the unannotated portion, and get a feel for what is there compared to the entire set. Understand your data. Protocol here: https://royalsocietypublishing.org/doi/10.1098/rsob.180241
You can follow @ValWood5.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: