search

geneontology / noctua

Graph-based modeling environment for biology, including prototype editor and services
http://noctua.geneontology.org/
BSD 3-Clause "New" or "Revised" License
37 stars 13 forks source link

Add missing inferences to output (add has-ligand to the whitelist in Minerva) #505

Open dosumis opened 7 years ago

dosumis commented 7 years ago

Expecting to see inference in this model: http://noctua.berkeleybop.org/editor/graph/gomodel:59cde02c00000003?

FZD1 enables 'transmembrane signaling receptor activity' has_ligand WNT3

I've tested this locally with the editor's version of GO and + minimal model with individuals. The appropriate rule is present in the import chain.

But I don't see any sign of this either in the inferred extensions or in the inference explanation tab.

How is the list of relations selected for display chosen? Will it start showing up once has_ligand is used in the ontology?

Cheers, David

balhoff commented 7 years ago

In your model, I see positively regulates. But the rule for has ligand seems to require directly positively regulates. Do you expect that to be inferred as well?

dosumis commented 7 years ago

Ooops. That fixed it for the inference report:

transmembrane signaling receptor activity#59cde02c00000006 has_ligand WNT3 Hsap#59cde02c00000005

Still not in extensions though. Presumably down to whitelisting. Does that work on the set of relations used in GO?

balhoff commented 7 years ago

@dosumis the allowed extensions relations are in here: https://github.com/geneontology/minerva/blob/master/minerva-converter/src/main/resources/org/geneontology/minerva/legacy/sparql/gpad-extensions.rq

We can add to it if desired.

kltm commented 6 years ago

@dosumis Is there anything actionable on this item?

cmungall commented 6 years ago

Not sure when @dosumis will have a chance to respond.

Even if has-ligand is not in whitelist, it's superproperty has-input is, but I don't see it. Perhaps redundancy filtering is done before whitelist application.

Let's just go ahead and add has-ligand to the whitelist, we can remove it later if it causes any issues

balhoff commented 6 years ago

Aha, I think you've hit on a problem with the current code. Redundancy filtering does happen before whitelist application.

dosumis commented 6 years ago

has_input no longer super-property (or shouldn't be). Aim is to separate out regulatory inputs, otherwise they get tangled with effector inputs. e.g. kinase activity has_input ATP. Don't also want to infer that, in case of RTK, ATPase activity has_input ligand.