Create an export of all GO Central annotations in a standard protein interaction format

[cmungall]

Following from discussion on http://wiki.geneontology.org/index.php/Annotation_Conf._Call_2016-03-22

TBD:

• [ ] decide on a format or formats -- @RLovering @tonysawfordebi @ValWood - what do you currently use? PSI?
• [ ] decide on algorithm
  • [ ] WITH from any protein binding [exclude different species?]
  • [ ] subset of annotation extensions - which relationship types
• [ ] implement. There are potentially >1 @tonysawfordebi has one for GOA already, but only does protein bindings
• [ ] create jenkins job. See also #169 (which covers the lego subset)

[tonysawfordebi]

https://github.com/MICommunity/psicquic/blob/wiki/MITAB27Format.md

[ValWood]

I didn't answer this did I. We only have interactions in GAF format or in BioGRID.

In fact we curate interactiosn into BIOGRID. They use a 'spoke model' for bait interactir so the interactors are not always direct. We only curate GO protein binding when we are convinced it is a high quality direct interaction (either within complex, or transient).

Its a bit redundant, and sub-optimal but it works for us at present. Other methods of curation are a bit time consuming for us in terms of metadata and detail, and our users seem happy so far.

[ValWood]

https://github.com/MICommunity/psicquic/blob/wiki/MITAB27Format.md

Confidence score, term from a vocabulary, "author based confidence" Useful REALLY???? Checksum ??? First author surname, when you have Pubmed ID???

wow, yuck, that is why we don't use this, life is too short. I hope its all cardinality zero! (not recorded in the format)

[RLovering]

please see https://www.ucl.ac.uk/functional-gene-annotation/psicquic and links from there

All PPIs from all species are exported based on IPI evidence and protein binding (and Child) annotations

best

Ruth