OBO-Edit Working Group, Tuesday November 17th, 2009 8.30 a.m. PDT
Conference call details:
Outside US: 1-719-867-3417
Action items carried over
I (Terry Meehan, MGI) am generating cross-product definitions in the Cell Ontology. To do this, I load the Cell ontology (CL), the three GO ontologies, PR0 and the Uberon ontology. I accidentally generated a cyclical relationship between the two terms after reasoning (see pictures). I *think* this resulted from using two related GO processes in the cross-product definition but am not sure due to some buggy issues.
- The "Explanation" feature both in the graph editor and in the Assert Implied links panel fail to explain reasoning. This is an issue Amina and Chris are aware of.
- Perhaps related... the Parent editor fails to show implied links, even when the checkbox is checked.
- Tree editor fails to show any relationship between the two terms.
- Should have the verification manager prevented this cyclical relationship from occurring? The reasoner stopped running checks due to the number of errors (>200). Most of these errors are miss-spellings in Uberon.
[Term] id: CL:0000782 name: myeloid dendritic cell namespace: cell def: "A dendritic cell of the myeloid lineage." [GOC:add, PMID:10449155, PMID:9521319] synonym: "DC1" EXACT  synonym: "interdigitating cell" BROAD  synonym: "veiled cell" BROAD  is_a: CL:0000990 ! conventional dendritic cell relationship: develops_from CL:0000576 ! monocyte [Term] id: CL:0000840 name: immature myeloid dendritic cell namespace: cell def: "An immature cell of the myeloid dendritic cell lineage, characterized by high levels of antigen uptake via endocytosis, macropinocytosis, and phagocytosis, and typically found resident in the tissues." [GOC:add, PMID:9521319] is_a: CL:0000782 ! myeloid dendritic cell intersection_of: CL:0000782 ! myeloid dendritic cell intersection_of: capable_of GO:0006897 ! endocytosis relationship: develops_from CL:0000576 ! monocyte [Term] id: CL:0000841 name: mature myeloid dendritic cell namespace: cell def: "A mature cell of the myeloid dendritic cell lineage, characterized by a high capacity for antigen presentation and typically found in a lymph node." [GOC:add, GOC:tfm, PMID:9521319] is_a: CL:0000782 ! myeloid dendritic cell intersection_of: CL:0000782 ! myeloid dendritic cell intersection_of: capable_of GO:0019882 ! antigen processing and presentation relationship: develops_from CL:0000840 ! immature myeloid dendritic cell
Discussion with Amina, Chris and David.
Amina: Explanation feature to be worked on.
Terry: Graph to be simplified.
Terry: Perhaps load into Protege 4 and see if different results show.
Terry: Check file in? Yes, will do later.
David figured out the issue after looking at my file. The following is from his email:
Here's the problem: name: dendritic cell intersection_of: leukocyte intersection_of: capable_of antigen processing and presentation name: myeloid dendritic cell is_a: dendritic cell name: mature myeloid dendritic intersection_of: myeloid dendritic cell intersection_of: capable_of antigen processing and presentation In OWL Manchester: mature myeloid dendritic cell EquivalentTo: myeloid dendritic cell and capable_of some 'antigen processing and presentation) myeloid dendritic cell SubClassOf dendritic cell dendritic cell EquivalantTo: (leukocyte and capable_of some 'antigen processing and presentation') therefore: mature myeloid dendritic cell 'dendritic cell' EquivalentTo 'myeloid dendritic cell' Conclusions: 1. OBO needs a better way of representing EquivalentTo than making it look like there's a cycle. 2. Making OBO files this complicated is dangerous if you're not running a reasoner regularly to test them! To do this, you don't need all of GO. Instead, you can just import the subset of terms you need. For a pre-reasoned ontology (as GO is) this = all asserted is_a paths to root.
Verification checker showing more than 200 non-fatal errors. Does it still look for fatal errors or does it just stop looking after 200 non-fatal errors?
David: I think it still finds fatal errors.
Amina: Working on a release. Hopefully this Thursday.
Amina is going to make the release and then we can do a lot of testing. Chris suggests that we should make it a beta release, since there are a number of fixes that haven't yet been tested.
Jen: Reports Explanation component problem. (Bug report submitted)
David: Using OBOMerge and having a subtle issue.
If def changed in both files only the first def is kept. The other is overwritten.
This is in the tracker with full details.
Would be better if the merged term kept both defs and flagged to the user to go back and fix.
Could also just treat as a clash and cause merge failure.
Adding relationship bug in source copy
Amina has managed to reproduce the bug that Jen sees in the source version of OBO-Edit where no relationships can be added.
David is having trouble with memory preference being stuck at 1024M on installation. No matter how he changes it is always reverts. He has not tried directly editing the vmoptions file. It is the configuration manager that has the problem. (vmoptions in the installation directory)
Harold asks about Mac log file.
Amina has not had a chance to test except on her own machine but will look at it before release.