Ontology meeting 2016-01-07

Attendees:Harold, DavidH, Tanya, Melanie, Paloa, DavidOS, Heiko, Chris.

Minutes:DavidH

Build fail

POTENTIAL REDUNDANT AXIOMS: GO:0098910 'regulation of atrial cardiac muscle cell action potential' part_of 'part_of' GO:0086004 'regulation of cardiac muscle cell contraction' is also a simple SubClassOf.

POTENTIAL REDUNDANT AXIOMS: GO:0098911 'regulation of ventricular cardiac muscle cell action potential' part_of 'part_of' GO:0086004 'regulation of cardiac muscle cell contraction' is also a simple SubClassOf.

https://github.com/geneontology/go-ontology/issues/12226

Chris thinks this is largely resolved. DavidOS thinks it is still throwing an error. There is a double-labeled edge. Do we want this? Are we going to run into this in LEGO where molecular functions directly activate the next molecular function but both are part of the pathway. We wouldn't want the first molecular function to regulate the pathway. We need to rethink this. We will draw up some examples of pathways.

Proposed obsoletion: single-organism process

https://github.com/geneontology/go-ontology/issues/12212

We will send out an obsoletion notice. What about the heirarchy under it? That will come later.

Double labelled edges

Can we revisit allowing these to be asserted?

e.g. This allows for any process to be classified as regulatory based on the presence of a regulates relationships, without an additional genus assertion: 'biological process regulation' EquivalentTo 'biological process' that regulates some 'biological process'

Defining regional parts: Many parts share most of the properties of a whole: membrane region, region of cytosol, region of cytoskeleton...

These can be defined using the pattern 'X that part of some X' e.g.: membrane region EquivalentTo: membrane that 'part of' some 'membrane'

This allows many classes which all biologists would agree are membranes to be classified as such, when previously we just had them as membrane parts and to autoclassify them by what membrane they are a region of e.g.

'presynaptic active zone membrane' 
SubClassOf:
  membrane
  part_of some 'presynaptic membrane'
  'plasma membrane region' { inferred }

'presynaptic membrane'
SubClassOf:
  membrane
  part_of some 'plasma membrane'
  'plasma membrane region' { inferred }

  'plasma membrane region' EquivalentTo: membrane that part_of some 'plasma membrane'

The only reason these are not allowed is that some resources can't load ontologies that have these. We have GO-basic to deal with basic issues like this (e.g. for resources that can only load DAGs). Can we just have a rule that deletes one of the edges from GO basic when two are present.

Are we going to strip these from GO basic and allow them otherwise? We need to check with the groups.We will remove the check for doubly-labeled edges in the complete GO. For go-basic we will have some type of rule that will remove doubly-labeled edges.

Additional MF terms

We have a group that is working on metabolic processes and would like to add around 1400 molecular functions to the ontology whose reactions are in EC, but are not covered by GO. Is there a way that they could supply a list of terms (special ID range?) in obo format that we could merge into the ontology? Alternatively, is there another format with which we could work?

There is already a Github issue for this. The automatic addition is problematic with respect to the hierarchy. Perhaps have them send them in 'chunks' and see how that goes without putting in a ton of resources up front.