Wrong Relation Displayed in Inferred Tree View?

[ukemi]

I think there is a bug in displaying relations in the inferred tree view for a GO term. If you search on the ontology term 'phosphatase complex (GO:1903293)' and display the inferred tree, the relation displayed between 'phosphatase complex' and 'phosphatase activity' is 'capable_of part of'. It should be 'capable_of'.

[kltm]

http://amigo.geneontology.org/amigo/term/GO:1903293#display-lineage-tab

View from parent term shows child with only "capable_of": http://amigo.geneontology.org/amigo/term/GO:0016791#display-lineage-tab

[kltm]

Nuts, I get the "correct" answer using the JS browser: http://nakama.berkeleybop.org/amigo/browse There is almost certainly a bug in the perl engine then.

[kltm]

I'm thinking that I'd rather drop the perl version altogether and just have a form of the JS one in its place--I do not want to have to keep aligning these two codebases.

@ukemi , you like the current tree behavior with jumping between pages, right?

[kltm]

Ugh, or maybe not--there is also the [Ancestors and children] view.

[kltm]

Duplicated locally.

[cmungall]

Technically it's not wrong, it's just imprecise (as capable-of-part-of is the parent relation). However, it is confusing. We should show the more specific capable-of.

[kltm]

Okay, I see where this is going. In the transitivity closure, we have the following relations:

                        {
                            "sub": "GO:1903293",
                            "obj": "GO:0016791",
                            "pred": "capable of"
                        },

and

                        {
                            "sub": "GO:1903293",
                            "obj": "GO:0016791",
                            "pred": "capable of part of"
                        },

Both are, of course, right. This looks like the shadow of #161, with the JS grabbing one, and perl the other. This could be "solved" by adding these relations to the weighing system.

[cmungall]

Bit ad-hoc, would be better to have better redundancy filtering upstream, but the weighting is at least a good and easy fallback

On 3 Mar 2016, at 15:15, kltm wrote:

Okay, I see where this is going. In the transitivity closure, we have the following relations:

                        {
                            "sub": "GO:1903293",
                            "obj": "GO:0016791",
                            "pred": "capable of"
                        },

and

                        {
                            "sub": "GO:1903293",
                            "obj": "GO:0016791",
                            "pred": "capable of part of"
                        },

Both are, of course, right. This looks like the shadow of #161, with the JS grabbing one, and perl the other. This could be "solved" by adding these relations to the weighing system.

---

Reply to this email directly or view it on GitHub: https://github.com/geneontology/amigo/issues/316#issuecomment-192013346