Issues with Annotation Extension relations

From GO Wiki
Jump to: navigation, search

1. exists_during;

domain: GO:cellular component

range: GO:MF, GO:BP or organism life stages

I find this relation a bit odd, if the relation is supposed to extend the GO term, then we are saying that the cellular component only exists during some MF, BP or life stage - does this mean the CC does not exist at other times? Are we really trying to say that the gene product is located in the CC during the MF, BP or life stage? I guess some CCs are only present at certain times. It's usage (below) indicates it's being used inconsistently.

Examples:

PomBase O13762 GO:nucleus exists_during(GO:single-celled organism vegetative growth phase)|exists_during(GO:cellular response to methyl methanesulfonate)

UniProt P55210 GO:nucleus exists_during(GO:erythrocyte differentiation)

These following two examples seem to comply to the definition:

DDB Q54Y26 GO:mitotic spindle exists_during(GO:mitotic prometaphase)

UniProt Q8NFG4 colocalizes_with GO:midbody exists_during(GO:cytokinesis)

2. Two relations that seem to be saying the same thing;

requires_regulator (child of dependent_on)

GO Annotation Domain: GO:0003674 (molecular_function) or GO:0008150 (biological_process)

GO Annotation Range: GO:0032991 (macromolecular complex) or MI:0315 (protein complex) or PR:000000001 (protein) or SO:0000673 (transcript) or SO:0000704 (gene)

Usage: annotated gene product requires regulation by the gene product or complex for its participation in the Molecular Function or Biological Process

versus:

requires_regulation_by (not a child of dependent_on)

GO Annotation Domain: GO:0003674 (molecular_function) or GO:0008150 (biological_process)

GO Annotation Range: GO:0032991 (macromolecular complex) or MI:0315 (protein complex) or PR:000000001 (protein) or SO:0000673 (transcript) or SO:0000704 (gene)

Usage: annotated gene product participates in Biological Process or executes Molecular Function activity only if regulated by action of gene product; regulation is indirect (or unknown whether direct or indirect)

Can we merge these? There is also a child of requires_regulation_by - requires_direct_regulator, which we could make a child of requires_regulation_by

Looking in the UniProt database, requires_regulator has only been used 4 times and requires_regulation_by only used twice (Sept 2013).

3. in_presence_of/in_absence_of;

Def: Identifies a chemical, gene product or complex in the presence/absence of which an ontology term is observed to apply to the annotated gene product

Does this mean that the entity is required to be present or absent in order for the GO term to apply? Currently, these relations are not children of dependent_on. If they are not required, then this should be clarified in the definition.

4. Relations without domain information;

axis_of

exports

has_output_o_axis_of

has_part

imports

regulates_level_of

regulates_o_has_output

regulates_o_occurs_in

5. requires_substance

There's a SourceForge ticket asking whether requires_substance should be made obsolete, because it's under-specified, and can be at least partly if not wholly superseded by activated_by and inhibited_by. See https://sourceforge.net/p/geneontology/go-relations/12/