Ontology meeting 2019-04-01

From GO Wiki
Jump to navigation Jump to search

Conference Line

See calendar for details



GOC Meeting Cambridge

  • GOC meeting - Cambridge, UK April 11th - 12th (right after ISB meeting)
    • has_primary_input/output report---David
    • reactome import into GO-CAM--- David, Peter
    • Rhea alignment--- Alan, Chris
    • MF refactor Pascale

Discussion topics

  • Jim will give an overview of how GO-REL is treated with respect to RO and RO-pending in the ontology loads.
  • Getting rid of regulates_o_part_of -> regulates
    • The plan going forward:
      • Jim has devised a way to make the inferences in the ontology and then strip the property chain from the relations upon release.Jim will give an overview of how these are currently being processed in the pipeline.
      • David has finished reviewing the file that Jim has created.
      • The file is called 'Subclass relationships inferred via regulates o part_of' and has two additional tabs as a sanity check for Barbara and Tanya.
        • I suspect that this will lead to a lot of changes in annotation with respect to slimming. Is there any way we can bounce this off of Val before we release it? In many cases there won't be much we can do because at a class level, removal of the chain will be valid, even though for some of the instances of gene products annotated to the regulation terms the chain will be true. For example, we will remove the positive regulation of x-receptor activity --> positive regulation of x receptor signaling because at the class level, the ligand does not positively regulate the pathway even though it does positively regulate the receptor. However, there may be other gene products that sensitize the receptor to ligand and therefore do regulate the pathway. As I go through these, I am realizing that there are a lot of problems associated with general regulation terms. To alleviate logical problems that exist now,
      • Tanya and Barbara, a good rule of thumb to use when investigating these is to consider the regulated process in the first term and determine if it is a step in the second process. If it is, decide whether or not that step would be activated or inhibited by another step in the process. If it is, then not every subclass of regulation of the first term would be a regulation of the second term. Example: regulation of 1 is inferred to be a regulation of 2. Is 1 a step in process 2 that could be positively or negatively regulated by 3, where 3 is also a part of 2? If this is true, then all cases of regulation of 1 are not regulation of 2. In this case, 1 is regulated by 3, that regulation is part of 2. In all cases that I could think of, if 1 was the start of 2, then the inference that regulation of 1 is a regulation of 2 is true and can stay because anything that regulates 1ies outside of process 2. Hope this makes sense.

  • Ensure slim documentation is generated from yaml in new GO site: what do we want to show?


Other business

On Call