Ontology meeting 2014-02-27
Attendees: Chris, DavidH, DavidOS, Harold, Heiko, Paola, Tanya
Minutes: Tanya
GOC and SAB meetings
What do we need to prepare? What do we want to present? Discuss and coordinate. Paola will then add to GOC meeting agenda. Some possibilities:
Jane: New viral terms with mapping to ViralZone/UniProtKW; general term stats (for SAB).
Yes. Chris/Heiko can assist with graphs if Jane can get numbers.
David OS: Multi-organism BP/CC annotation model.
Yes.
Chris: progress report on our transition to OWL - the megafile etc.
Yes.
Paola: overview of the cilial and Giardia work; probably a quick update on apoptosis (necroptosis annotation, QC of execution phase, addition of annotation guidance and examples in the ontology itself; where we are with the paper); GO enrichment analysis.
Yes.
DavidH and Tanya: present some of the modelling they've been doing?
Yes. Glycolysis, molybdenum cofactor biosynthesis, DNA replication
DavidH: ubiquitin story -
Is that ok with everyone? What else?
Highlight things like interactions with other groups, annotation groups, Giardia. New documentation pages in Drupal.
Follow-up: Next steps in the post-megafile world
Any update on this? Notes and discussion from last week:
[Chris] https://www.ebi.ac.uk/panda/jira/browse/GO-222
I'd like to push on with subtasks 5 and 6 since this should simplify the file dependencies
LAST TIME: Tentatively agreed that disjoints should and taxon constraints should go into main file, with the exception of GCI disjoints - pending testing in OE TODO - David to test OE with editors GO file + disjoints, with a global filter in place to remove them from view. Note that reasoning with taxon constraints won't work without expansion. TODO - Chris to add make option for GCI expansion (EL version) of taxon constraint assertions.
Chris will try to fix never_in_taxon today and sort that out. Want to get taxon constraints into the main file. DavidOS - tested w disjointness axioms and global filters, minor performance hit, looks like ready to go in, GCIs need to be split out
Pairwise disjoints will go in today.
Dictionary errors in OE
These are now up to 825... following email discussion from last week, could we please fix these so the verification check becomes useful again?
Errors mostly on names of chemical compounds. Solution would be to copy the missing names into the dictionary file but there are a ton of them. Manual propagation is a pain. These are not typos. Can we strip these types of terms from the spellchecker? Technically difficult to spell check only GO terms. Spell checking is ubiquitous. Is the manual solution the only one? How about loading a 'reduced' version of GO which doesn't have all the ChEBI terms in it so that you only look at the set of terms that is relevant. How about just loading all the ChEBI names into the dictionary file? Too big? Hard to keep up to date with ChEBI changing?
DavidOS will write a script to add the ChEBI names into the dictionary that can be run regularly.
SOP for adding relations
If I've added a new relation to the relations ontology, is it sufficient to have this is in gene_ontology_write.obo:
[Typedef] id: some_obo_only_readable_id xref: RO:nnnnnnn
for RO-imports to be automagically updated? (presumably adding more could cause problems with duplication of APs)
This should work. Let's see if that happens. Two day process. Will report back.
Or should I be using the new term imports file (extensions/imports_requests.obo) but with a typedef filled in?
Not this way.
OE warnings on save
From Paola's email 2/26:
15 non-critical warnings:
cell (CL:0000000) generated 1 warning: The terms cell (CL:0000000) and cell (GO:0005623) share the name or synonym "cell".
cell (GO:0005623) generated 1 warning: The terms cell (CL:0000000) and cell (GO:0005623) share the name or synonym "cell".
[Chris] known issue, needs to be resolved, will be collapsed eventually, may obsolete CL:cell and replace with GO:cell [DavidOS]: But may be a distinction between GO:cell including encapsulating structures and CL:cell not. Also - there is a potentially serious issue for searching, querying and editing in Protege when there are multiple terms with the same label. Not possible to know which term has been chosen when writing axioms; search will randomly go to one of two terms and not show other. The obvious solution to this is to use OBO foundry Unique names to disambiguate - Protege can be set to display these instead of label if they are present , but this will only be practical if we can do this while making sure new term generation defaults to rdfs:label. I have tested this - and it works fine. TODO: work out how to include obo foundry unique label (http://purl.obolibrary.org/obo/IAO_0000589) in standard set of annotation properties in GO file.
cyclin-dependent protein serine/threonine kinase activity (GO:0004693) generated 1 warning: Term cyclin-dependent protein serine/threonine kinase activity (GO:0004693) contains errors
kinase activity (GO:0016301) generated 1 warning: Term kinase activity (GO:0016301) contains errors
[Paola] These two have been around for a while and I can't see where the problem is...?
epidermal cell (PO:0004013) generated 1 warning: The terms epidermal cell (PO:0004013) and epidermal cell (CL:0000362) share the name or synonym "epidermal cell".
epidermal cell (CL:0000362) generated 1 warning: The terms epidermal cell (PO:0004013) and epidermal cell (CL:0000362) share the name or synonym "epidermal cell".
[Chris] 'epidermal' is pretty vague. You can pick and choose from a bunch of concepts, and get either a very broad concept (covering any outer layer of cells) or specific (the specific layer in vertebrates that is adjacent to the dermis). Uberon has 'outer epithelial layer' and the more specific 'skin epidermis' with the CL terms associated with the former (which is necessary as 'epidermal cell (sensu Arthropoda)' is a subclass. There is a taxon constraint to Eumetazoa (some sponge lineages may have epithelial cells but I don't believe they form an outer layer in the same way). I favor keeping the plants and eumetazoan concepts distinct, and PO coming up with another name (possibly encoded as 'obo foundry unique label'). Of course, GO conflates the two concepts: http://amigo2.berkeleybop.org/cgi-bin/amigo2/amigo/term/GO:0009913
gamete (PO:0025006) generated 1 warning: The terms gamete (PO:0025006) and gamete (CL:0000300) share the name or synonym "gamete".
gamete (CL:0000300) generated 1 warning: The terms gamete (PO:0025006) and gamete (CL:0000300) share the name or synonym "gamete".
[Chris] Tanya, which is correct? [Tanya] Defs are similar enough that they could be merged. Punt to PO folks.
micropyle (GO:0070825) generated 1 warning: The terms micropyle (GO:0070825) and micropyle (PO:0020025) share the name or synonym "micropyle".
micropyle (PO:0020025) generated 1 warning: The terms micropyle (GO:0070825) and micropyle (PO:0020025) share the name or synonym "micropyle".
[Chris] Tanya, which is correct? [Tanya] V. interesting. Hadn't heard of the 'other' micropyle before. Animal micropyle = opening in ovum that allows sperm to enter. Plant micropyle = opening in ovule that allows pollen/pollen tube to enter.
molybdopterin synthase activity (GO:0030366) generated 1 warning: The term molybdopterin synthase activity (GO:0030366) has a synonym with the identical name.
[Paola] Had a look at the log but it doesn't say who did this edit (and no initials are attached to the term or file). Maybe David H? Can the exact synonym be deleted, or was it meant to be something different?
Delete synonym.
root (PO:0009005) generated 1 warning: The terms root (PO:0009005) and root (NCBITaxon:1) share the name or synonym "root".
root (NCBITaxon:1) generated 1 warning: The terms root (PO:0009005) and root (NCBITaxon:1) share the name or synonym "root".
[Chris] he he, yes I think we should just trim NCBITaxon in our import here, or bring in the CARO awkwardly named grouping class [DavidOS] CARO gropuing class = 'organism, virus or viroid'? I'd be very happy with that solution as it will fit with work we need to integrate from PCO.
xanthophore (GO:0031633) generated 1 warning: The terms xanthophore (GO:0031633) and xanthophore (CL:0000430) share the name or synonym "xanthophore".
xanthophore (CL:0000430) generated 1 warning: The terms xanthophore (GO:0031633) and xanthophore (CL:0000430) share the name or synonym "xanthophore".
[Paola] The GO term has a def. comment specifying the difference.
[Chris] One of us needs to yield here…
Progress on or plans for Protege plugin ?
No progress so far, may fall on us to work on it.