Ontology meeting 2012-04-11
From GO Public
- 45 min call
- 8:45am Pacific/11:45 East Coast/16:45 UK
Attending: Chris, Heiko, Paola, Harold, David, Karen, Tanya
DISCUSSION 1: Chemical XPS
D & T - what to report from the list of chemical-x inferences? Where next? PAPER!
- report and discussion
uchebi.owl - the union classes are created, not all links are present in ChEBI, example: 'dicarboxylic acid' and dicarboxylic acid anion exist in ChEBI but are not linked
Do we need new terms that mirror chebi but include the 'I-don't-cared-ness' of whether the chemical is the base or the acid?
Aggressive strategy: create uchebi terms all the way up the graph - too much separation from ChEBI
ACTION ITEM: focus on oxoacids, look for which ones are used in xps in GO terms have missing anions or missing conjugate base/acid relationships. This will be test set. Look at compound terms where ladder is complete and see if they are good in Protege. Request new terms or new relationships from ChEBI.
Suggestion: in logical definitions, refer to actual chemical, in label - keep the ambiguity? (We are back to where we started.) - NO! We do need the union classes to cover different physiological pHs.
Other issue: cases where there are triples of conj base/acids - serine/serinium/serinate - shall we just focus on the ones that are physiologically relevant? - new meta-data about what chemicals don't exist at physiological conditions - for example, 'physiological serine and conjugate'. Allows us to eliminate some classes that are not relevant.
- generating inferred links report
new OORT release (Heiko), addresses java memory issues for running reasoner-diff script
HOW TO GET x-chebi-diffs.txt
make sure you are in ontology/extensions directory %export PATH="$PATH:/Applications/OBOReleaseManager/scripts" %reasoner-diff --reasoner elk --catalog-xml `pwd`/catalog-v001.xml FULLPATH/ontology/go.owl `pwd`/uchebi.owl FULLPATH/ontology/external/obo-relations/ro.owl --change `pwd`/x-chemical.owl --report `pwd`/x-chebi-diffs.txt
FULLPATH = set to the path in your computer to the file, varies from my machine to yours
32 inferred axioms - look good, to be reviewed this pm
DISCUSSION 2: Protein/Peptide
Discussion ongoing w/ ChEBI and PRO. ChEBI to "remove the protein branch from its ontology by June of this year."
'length does not matter' = we may just map oligopeptide, polypeptide and peptide to the same term = 'amino acid chain' (PRO)
DISCUSSION 3: TermGenie paper
Link to Google Doc sent. Please add comments.