Cell and anatomy relations in BP only template are creating incorrect models

[vanaukenk]

In the BP only version of the form, there are two entries for Cell and Anatomy, both of which use the occurs_in relation.

This is the correct relation to use between a BP term and either a cell or anatomy term, but when the corresponding annotations appear in the graph editor, the cell type is connected to the anatomy term via an 'occurs in' relation which is incorrect.

One possible solution for this might be:

Keep both occurs_in relations for cell and anatomy, but also add a slot for a part_of relation to qualify the cell type with an anatomy term. The occurs_in cell and anatomy qualifiers would be at the same 'level' in the form view, but the part_of anatomy slot would only further qualify the cell type.

This would allow curators to make a statement that a BP occurs_in a cell that is part_of an anatomy term, or just say that a BP occurs_in a cell OR that a BP occurs_in an anatomy term.

@sabrinatoro @krchristie @tmushayahama

[krchristie]

When I have both a cell type and an anatomy term, I think I usually want to say that the cell type is 'part_of' the anatomy term. I think Kimberly's proposal would work for me.

[vanaukenk]

New behavior: if a curator enters a value into the Cell field, then the Anatomy relation will become part_of.
If no value is entered into the cell field, then the Anatomy relation stays as occurs_in.

[tmushayahama]

@vanaukenk @thomaspd Postponing this problem, by making the default bpOnly assumes everything is filled out. So is it okay to label this (changed occurs in to part of Anatomy)

[vanaukenk]

@tmushayahama I think this works fine.

The only thing is that the graph editor is still making an occurs_in relation between the cell and anatomy term when both fields have a value:

When both fields are filled in, the cell type and anatomy fields should be connected via the 'part of' relation in the graph editor.

Thx.

[tmushayahama]

Testing using http://68.181.125.145:8910/

Thanx @vanaukenk I have fixed the graph created on the . Can you check again together with the summary table table. So far it doesn't work when Cell Type is skipped

Also I have updated the test pre and post check cases on that file https://docs.google.com/spreadsheets/d/1cN2NbikNsp5_-kAAJTCuIuynp1v7g52YNrcX9dGRzmU/edit#gid=256881405

[vanaukenk]

Hi @tmushayahama

Yes, the relations in the graph editor are now correct for me when both CC and Anatomy are filled in.

And when CC is skipped, the relation is not correct, i.e. it is BP part_of instead of BP occurs_in. But this was working correctly earlier in the day, I thought, when I was doing some other testing. Is that possible?

[vanaukenk]

@tmushayahama - just checking on the status of this ticket wrt skipping the cell type and only adding an anatomy term. Note that when a curator skips the cell type and only enters an anatomy term, it is possible to save the annotation without adding evidence to the anatomy term.
We need to also make sure that no annotation can be saved without evidence on each assertion.

[vanaukenk]

@tmushayahama There is still a bug when curators enter an anatomy term, but not cell type on the BP only form. The relation between the process and the anatomy term needs to be occurs_in rather than part_of when no cell type is entered. Here is what currently happens:

This is a high priority fix, so please let me know if you want to chat about it. Thx.

[vanaukenk]

Needs to be addressed in current specs:

https://github.com/geneontology/GO_Shapes/blob/master/shapes/go-cam-shapes.shex