Ontology meeting 2012-05-23
Chris will expand on the following:
if there are any conflicts, choose "(p)" postpone. Then edit the gene_ontology_write.obo file and resolve the conflicts.
This should be exactly the same as the old cvs workflow.
After you have resolved these, type:
svn resolved gene_ontology_write.obo
PS: What is the svn command for checking out a new folde if we want to grab a clean copy? (svn co [path] isn't sufficient it seems).
Jane and TOny have solved this... just remove the file (rm) and update (up) to get a new version (as long as folder is in place).
FOLLOW-UP ON CHEBI ALIGNMENT
- Follow-up on all action items from last week:
- Reviewing links to be removed
This is the build artefact of most interest - it contains all of the links in the GO process graph between defined classes that *cannot* be re-inferred using CHEBI. We only try and re-infer links between two classes where both classes have a logical definition referencing CHEBI.
(T/D) Still something off:
SubClassOf(GO:0015746 'citrate transport' GO:0015711 'organic anion transport')
[Term] id: GO:0015746 ! citrate transport intersection_of: GO:0006810 ! transport intersection_of: results_in_transport_of CHEBI:16947 ! citrate(3-)
- Open issues with adding chemical xps are listed in this wiki (to be edited as we go along):
In particular: shall we go ahead and delete the xps for "secretion" terms for now? And/or resolve on a more complex relation such as e.g.
- Also: how are we going to deal with metabolic terms in GO where the CHEBI entry is only a role in ChEBI?
secondary metabolite (CHEBI:26619) is_a role only in ChEBI.
- Related: what parentage for GO terms such as GO:0042417 ! dopamine metabolic process?
See email thread on the go-ontology mailing list; Chris suggests the following as an alternate plan:
"we treat CHEBI roles **as if they were structural classifications**, and if we want to indicate the context, **we do this ourselves using GO, rather than the CHEBI role hierarchy**".
- Related: should we replace 'secondary metabolic process ; GO:0019748' with 'secondary metabolite metabolism ; GO:NEW'? as suggested by Diane:
- If so, is the 'has_role' relationship in ChEBI replaced by an 'is_a' relationship in GO? E.g.
secondary metabolite metabolism ; GO:NEW --[isa]endocrocin metabolic process ; GO:1900600
- DH: chemical HAS_ROLE in CHEBI means that the chemical can have the role. The classic example is dopamine, which is NOT ALWAYS a neurotransmitter (e.g. outside of the nervous system and in plants). Therefore, the problem with putting all these chemicals under 'secondary metabolite metabolism' is that they might not always be secondary metabolites in all organisms.
At least 3 options:
- Add the 'x metabolism' terms under 'secondary metabolite metabolism ; GO:NEW'. Remove this parentage IF a paper shows that the chemical is NOT a secondary metabolite in a given organism/context.
- Create child terms under 'secondary metabolite metabolism ; GO:NEW'. E.g. endocrosin secondary metabolite metabolism ; GO:NEW'. Would create alot of work, and alot of redundant terms.
- Capture this at the annotation stage by linking the two terms. Annotate to 'endocrocin metabolic process ; GO:1900600' with 'is_a:secondary metabolite metabolism ; GO:NEW', or 'HAS_ROLE: secondary metabolite' in column 16.
Progress on involved_in logical definitions
There are four categories of terms:
- straightforward (GO:0043689),
- terms where a genus needs to be added to make the cross products (GO:0021812)
- terms where additional relationships are appropriate that are not necessarily specified by the logical definitions (GO:0086071)
- terms that have more than one differentium (GO:0060136).
Current relationships in this document:
results_in_transport_of results_in_transport_from results_in_transport_to results_in_transport_towards results_in_transport_involving results_in_transport_along results_in_transport_across results_in_transport_to_from_or_in results_in_assembly_of results_in_disassembly_of results_in_organization_of results_in_fusion_of results_in_fission_of results_in_localization_of results_in_morphogenesis_of results_in_connection_of results_in_distribution_of results_in_remodeling_of results_in_maturation_of mediated_by occurs_in has_output dependent_on has_level part_of
*AI: Cam-Eds to take care of these. Either need to arrange another call with the annotators (e.g. Val, Midori, Emily) and editors (including David, Chris) to get the relationships sorted. Or just go ahead and put the transport XPs straight in the file.
- DH: 'has_target' is more specific than 'has_input'. E.g. for protein phosphorylation, both ATP and the protein substrate are inputs, but only the protein substrate is the target.
- This is subjective, depending on what you're interested in. E.g. the MENGO group are interested in by-products and consider these to be the important output for a reaction.
- This may not matter because at the annotation stage, for the protein phosphorylation example, you would only put the protein substrate ID in the annotation extension column.
- For 'regulation of transcription', what is the target? The gene or the protein that comes out.