Ontology meeting 2013-09-19: Difference between revisions
Jump to navigation
Jump to search
No edit summary |
No edit summary |
||
Line 8: | Line 8: | ||
Stems from an email thread that Chris started. Do we need new terms? | Stems from an email thread that Chris started. Do we need new terms? | ||
===High-level disjoints=== | |||
Stemming from Chris's email: | |||
We use a lot of very specific relations, especially in the development logical defs. These are necessary - if we try and use a more generic relation then the reasoner gets everything tangled. | |||
The specific relations are causing some angst when curators use them in class expressions - they'd prefer a smaller set. I would do, but we need some disjointness axioms first. | |||
Let's say we want to get rid of the ugly transports_or_maintains_localization_of and use a more generic has_input. Seems to make sense - we define things such that the input of a transport process is the cargo. Simple? | |||
consider GO:0015986 ! ATP synthesis coupled proton transport | |||
What's it's input? | |||
It's a subtype of transport, so if the rule is that input for transport is always the cargo, then the input is the proton | |||
It's a subtype of metabolic process, so the input is the thing that is modified - the ATP | |||
The fix here is to get rid of the isa-overloading in the ontology, use a different relation that isa for one of the above paths, make transport and metabolism disjoint. If they are disjoint then we can safely use a broader relation without ambiguity. | |||
There's only a handful of things we'd have to deal with if we were to make transport and metabolism disjoint (receptor internalization too). Remember, you can now do DL queries with Elk so you can just query "transport and metabolic process". | |||
David H I know I've pestered you about this isa overloading in development before, I think this is a higher priority now with more curators ramping up to use these relations. | |||
===DOCUMENTATION FOR NEW WEBSITE=== | ===DOCUMENTATION FOR NEW WEBSITE=== |
Revision as of 06:28, 18 September 2013
Minutes: Paola
Attendees:
Glycosylation
Stems from an email thread that Chris started. Do we need new terms?
High-level disjoints
Stemming from Chris's email:
We use a lot of very specific relations, especially in the development logical defs. These are necessary - if we try and use a more generic relation then the reasoner gets everything tangled. The specific relations are causing some angst when curators use them in class expressions - they'd prefer a smaller set. I would do, but we need some disjointness axioms first. Let's say we want to get rid of the ugly transports_or_maintains_localization_of and use a more generic has_input. Seems to make sense - we define things such that the input of a transport process is the cargo. Simple? consider GO:0015986 ! ATP synthesis coupled proton transport What's it's input? It's a subtype of transport, so if the rule is that input for transport is always the cargo, then the input is the proton It's a subtype of metabolic process, so the input is the thing that is modified - the ATP The fix here is to get rid of the isa-overloading in the ontology, use a different relation that isa for one of the above paths, make transport and metabolism disjoint. If they are disjoint then we can safely use a broader relation without ambiguity. There's only a handful of things we'd have to deal with if we were to make transport and metabolism disjoint (receptor internalization too). Remember, you can now do DL queries with Elk so you can just query "transport and metabolic process". David H I know I've pestered you about this isa overloading in development before, I think this is a higher priority now with more curators ramping up to use these relations.
DOCUMENTATION FOR NEW WEBSITE
- Who is dividing up the old pages to distribute to people to rewrite?
- Who is rewriting what?
Follow-up: Changes to external ontologies
Background here http://wiki.geneontology.org/index.php/Ontology_meeting_2013-09-12#Changes_to_external_ontologies
Follow-up: Equivalent classes for Acid/Base transport and transport activity
Follow-up: protein localization to xxx
Have we fixed this? http://wiki.geneontology.org/index.php/Ontology_meeting_2013-09-12#Template_for_.27protein_localization_to_xxx.27_.3F