Annotation Conf. Call, May 13, 2014

From GO Wiki
Jump to: navigation, search

Agenda

SGD: Rama, Paul, Diane
EBI: Jane, Paola, ALeks, Prudence, Rachael
MGI: Karen, Judy, Li
WB: Kimberly
Rat: Stan
Berkeley: Suzi, Chris
USC: Paul T
UCL: Ruth, ?

E2/E3 changes

See SF item.

Extracellular branch of GO CC (David OS)

I've done some work on fixing a number of issues with the extracellular branch of GO CC. I'd like to get some feedback from annotators on my proposed changes before committing them.

slides here

With/from cardinality (Rachael)

Examples of multiple with strings for IPI.

This follows on from the discussion of whether to allow more than one ID in the with/from field for certain evidence codes (see http://wiki.geneontology.org/index.php/Manager_16April2014).

Examples for possible representation of multiple with/from entries in protein binding (IPI) annotations are shown below. Reciprocal annotations are not shown, but they would follow the same format as that shown for each example. The evidence would need to be from the same paper.

1. There is evidence to say that Protein A binds proteins B and C at the same time, e.g. by cross-linking study (comma separated);

DB Object GO ID Reference With/from
Q03388 VPS72 GO:0005515 PMID:24034246 UniProtKB:Q03940,UniProtKB:Q12464,UniProtKB:P35817,UniProtKB:P60010

(see Complex Portal http://wwwdev.ebi.ac.uk/intact/complex/details/EBI-8790722)

2. There is evidence to say that Protein A binds proteins B and C independently, e.g. by co-sedimentation assay (pipe separated);

DB Object GO ID Reference With/from
Q03940 RVB1 GO:0005515 PMID:24034246 UniProtKB:Q03388|UniProtKB:P80428

3. There is no evidence to say whether Protein A binds proteins B and C at the same time or independently, e.g. in vitro binding assay (separate annotations);

DB Object GO ID Reference With/from
P19454 CKA2 GO:0005515 PMID:24034246 UniProtKB:P43639
P19454 CKA2 GO:0005515 PMID:24034246 UniProtKB:P38930

If we do this, we should consider also the possibility of deepening protein binding annotations to child terms of protein binding.

Multiple with/from fields could only be applied to new annotations, we couldn't retrofit without going back to the papers, whereas we could deepen the protein binding term for old annotations.

If we agree to multiple with/from entries for IPI, it might be worth thinking how we could capture the details of binding partners in a complex though. e.g. If a complex of three proteins was described, could we annotate to a protein complex term and somehow indicate (via part_of or has_part or some other relation) what each component binds? The idea would be to have a way to say, for example, that a gene product is part of a complex and in the context of that complex it has protein kinase binding and calmodulin binding activity while another member of the complex has only calmodulin binding in the context of that complex.

Alzheimer's Disease Annotation Project (Jane/Paola/Sejal)

http://wiki.geneontology.org/index.php/Alzheimer%27s_Disease_Annotation_Project

GOC call on June 18th

We will have a consortium wide conference call on June 18th for 1.5 hours. Please mark your calendars. http://gocwiki.geneontology.org/index.php/Manager_7May2014#Minutes

Discussion

E2/E3 terms

  • Why are we not using E2 and E3 in the term name itself? E2 and E3 don't have a clear boundary and are sometimes carried out by one complex or by two separate gene products. Ontology editors will address this question further. These terms have been added to the ontology

With column

  • There should be some consistency in the way multiple IDs are entered in the With column and Col-16
  • We can't retrofit existing annotations, we are looking at implementing these for new annotations
  • What binds to what? we should coordinate this with IntAct and put IntAct ID in col-16
  • An important issue to consider is how they will be parses and displayed on AmiGO, QuickGO, what they mean, how we make it clear to users the meaning of pipe vs comma
  • Rachael will check with Tony to figure out how long it will take to make protein2GO handle this
  • this should be done minimally for IGI sooner than later