Ontology meeting 2013-02-7
ATTENDEES: Jane, Paola (having trouble w/ computer), Judy, Harold, David, Tanya, Chris, Heiko
What next? What journal, what format... keep pre-submission enquiry?
Submit to another journal. Need to finish last bit: finish last figure and add text referring to it to the paper. Rework text to fit the new target journal? Chris can work on this on Monday. Tanya to reformat references (DONE).
Sandra Orchard is launching a new protein complex portal at EBI as part of InTact, the idea being that it will be the reference db for protein complexes. DB entities will be species-specific so they will still need new GO cc terms.
Meant to be complementary resource to PRO
They have a curator, and we've agreed that once she's up to speed (she started a couple of months ago) she will add cc protein complexes directly to GO using TG FF. She'll be annotating her own complex sets (human, E. coli and yeast in the first instance) but I think we should assign the GO tickets for protein complexes to her too.
They're also going to suggest some higher-structure for the GO protein complex node which currently totally flat. This will be functionally-based which I think makes sense.
Assign new complex requests to new EBI curator when she is up to speed. Could create new complex terms in GO directly using TG FF or create a new template (composition based) for this purpose. Harold objects to this - for GO - define a complex based on what it does or the process it performs. Jane: ideally define complexes functionally, not compositionally. They are thinking of functionally-based higher order categories.
As a related issue, I notice we haven't always been adding a part_of for protein complex location. This is more important than I realised and I'm guilty of not adding it too because it can be tricky. Let's discuss
Jane to check part_of relationships for complexes. Add those that are obvious. In future, add part_of for new terms as part of ontology editing session.
add capable _of xps for complex terms to the xp file, review in obo format, divide up depending on size of file, x-complex.obo in the extension directory
Harold also wanted to revisit the discussion on component based definitions for complexes vs function/process based. He is concerned that often the exact components are either not known completely or can vary over conditions of analysis.
We need to talk with Sandra too about the intersection between IntAct and PRO relative to the representation of protein complexes. PRO is actively curating these also. Is Sandra in touch with them. Harold is one of the curators for PRO complexes.
Harold will touch base with Darren wrt PRO/IntACT interactions.
Interactions with CL
We need to keep this going or it will just drift further.
Add issue into JIRA so we can track it and assign a priority.
Interactions with CHEBI
Jane: I spoke to Janna about the lag issue we have with CHEBI releases. She's very happy to start producing a pre-release OWL version of CHEBI on a regular basis, perhaps nightly. Their OWL representation is non-standard (it has bits missing to reduce its size) but it would be straightforward to translate it into regular OWL if we need it. This can happen fairly quickly - Janna will get back to us.
They're also planning to add a lot more axioms in the fairly near future - they have some disjoints already.
Chris:Need to align on synonyms so that we can use them. Using OBO2OWL converter should address this. CHEBI synonyms are not currently used for term generation in TG because of inappropriate kinds of synonyms - SMILES, etc.
protein domain specific binding
Do we really want this as a template? Wouldn't this be better post-composed as an annotation extension? An InterPro id could go in column 16
What would we use as the external 'ontology'? Interpro domains? Is this available in OBO?
Current ontology has very flat structure under protein domain specific binding. Jane will talk to Sarah. If we can use this file, or a newer version, we can create xps for these and create a new template. Jane will talk to Tony about master's project of guy @ EBI last year that might be useful.
Protein/glycoprotein/lipoprotein terms in ChEBI
Chris to report if there's any progress - Background here: http://wiki.geneontology.org/index.php/Ontology_meeting_2013-01-24#Protein.2Fglycoprotein.2Flipoprotein_terms_in_ChEBI
Chris:Nothing new to report. Problem: Will PRO and ChEBI converge on one protein term and, if so, where will it live? Will this definition cover glycoproteins? Harold and Judy will take this on.
Collaboration with group at NJIT
Jane: They've produced this visualization of GO bp + xps that potentially allows us to spot problems. Wondering whether we should agree to work with them further. I'll talk everyone through this figure.
Very keen group. Figure linked is a shot from the tool which is meant to reveal 'errors' in the xps. Jane will examine outliers as pilot project (some time only, not a lot) and see if this is worth pursuing. May reveal something that our internal checks don't. Could reveal the number of terms in each template in a graphical manner. Could see different relationship patterns for terms within the same ontology.
Summary of OBO2OWL meeting
Write up report with number of crossproducts added (look at svn, look at diff, count), number of CL terms added, changes to the x-cell.owl file. As a way to quantitate productivity of the meeting. David will do this, starting from the meeting minutes. Put on the wiki.
Monthly update: Project Management in JIRA
review issues next week
Next priorities: Missing xps, then xps for CL.