Ontology meeting 2013-01-31

From GO Wiki
Jump to navigation Jump to search

MINUTES: David

ATTENDEES: Suzi, Paul T., Huaiyu, Donghui, Rama, David, Tanya, Chris, Jane, Heiko


Annotation group

Discuss function-process links and annotation to processes connected by MF-BP links.

Single Step Process terms (Paul, Rama et al)

The PAINT annotation group is discussing the usefulness of annotating to single-step process terms like Phosphorylation for gps that are kinases, arginine catabolic process for say Nitric oxide synthase and so on. The issues are:

  • the function term annotation already expresses the role of the gp in these cases. Do we need to annotate to the equivalent BP term? Should anybody be annotating to these single step process terms (when function is known).

There is no reason to manually make the annotation to the process term. The GOC inference file makes the transitive annotations to the corresponding processes and those annotations (hopefully) are picked up by the MODS who cannot use the MF-BP relationships. However, we should point out that processes by nature are not necessarily single-step. For example PMID:12176995 describes the activity of a gene product that recruits a kinase and is necessary for the kinase to phosphorylate a set of substrates. In this case the gene product is part of the phosphorylation process, but it is not the kinase. Having a process called 'phosphorylation' allows us the flexibility to represent a biological objective that may include the functions of the gene products other than the obvious. After examination of the definitions of some of these process terms, we now realize that they can be better defined.

  • Even if PAINT curators don't make the annotations, the MF-BP inferencing script uses the interontology links between these terms and automatically asserts these annotations.

Yes, the interontology links are no different than any other links in the ontology.

  • the process term should represent a more detailed biological role than be redundant with function

The process terms are not really redundant. Processes by nature MAY contain more than one molecular function even if it is not obvious or known. Processes are a step in granularity above the single function level. Transport is a great example of this. In the past if we found processes that are the exact same as functions, we have merged the terms.

Terms for yeast CDC28 (rama)

CDC28 is a kinase in yeast and it regulates a whole bunch of processes in yeast by kinasing various substrates. I would like to request terms like:
kinase activity involved in RNA polII transcription
Kinase activity involved in cell cycle arrest
and so on. Should we be encouraging curators to think along these lines when they request terms?

At this point in time, yes. It is appropriate to request 'molecular function involved in biological process' terms. When these are requested the part_of interontology links will be made.

PLAN:

Step 1: Paul T. to assemble list of single step processes that need to be reviewed that have come up during PAINT curation and will start JIRA project

Step 2: plan to merge kinase activity and phosphorylation - need to review annotations, computationally identify those that are annotated to p'rylation and not to kinase and see what to do with those

Step 3: global review of BP to identify 'molecular' processes and see if these can be merged into an appropriate MF - go-top to prioritize this task with respect to other ontology dev work.