Draft 1: Update confirmed subclass of evidence in Mondo

[matentzn]

This PR

• Adds a workflow for updating confirmed subclass of evidence
• Updates mondo-edit.obo accordingly

Issues of dispute

Against our previous agreement, the current version of the "confirmed" subclassOf axioms covers "Case 2 (SCR is direct in source, but indirect Mondo)". So, in other words, we include subclassOf axioms that are implied by Mondo (not asserted in the edit file!) and confirmed by an external source.

Example:

[Term]
id: MONDO:0000266
name: pulmonary aspergilloma
def: "A aspergillosis that involves the lung." [MONDO:patterns/location]
...
intersection_of: MONDO:0005657 ! aspergillosis
intersection_of: disease_has_location UBERON:0002048 ! lung

results in this:

[Term]
id: MONDO:0000266
name: pulmonary aspergilloma
def: "A aspergillosis that involves the lung." [MONDO:patterns/location]
...
is_a: MONDO:0005657 {source="DOID:0050153"} ! aspergillosis
intersection_of: MONDO:0005657 ! aspergillosis
intersection_of: disease_has_location UBERON:0002048 ! lung

https://docs.google.com/document/d/1H8fJKiKD-L1tfS-2LJu8t0_2YXJ1PQJ_7Zkoj7Vf7xA/edit#heading=h.9hixairfgxa1

Chris would probably be very against this, and I must say, I am also against it, but weakly. Indeed, I am like 55% against it, and 45% think this could be a good idea. The advantage is that we see all the evidence as Mondo curators - every positive subsumption. When we make classification decisions in the future, we can see them all in the edit file (not just the non-redundant ones). Note - all of these are implied, so they are all correct. Today.

The clear disadvantage is though that changing our logical definitions / patterns (equivalent class statements) will no longer be sufficient to change classification - if you really want to get rid of some subsumption moving forward, you must physically delete both the equivalence class axioms and the asserted subclass axiom. Here is a particularly bad scenario: We decide a pattern no longer works and change it, which leads to changes in classification. The old classification will not be removed, as the subclass of axioms have now been baked into the class hierarchy. I dont know though how much of a problem this is in reality.

I would suggest you

1. Consult Chris on slack about his opinion on this. Its a big, PI-level, change.
2. If he says "no", we need to change the source for the "confirmed subclass pipeline" to "mondo-edit". Its not a huge deal, but may require some thinking.
3. If he says "yes", we still need to determine if there is a bug in this PR or not - I spot-checked 10 random examples and they are all correct (correct in the sense, mondo entails the subclass axiom).

[sabrinatoro]

I don't mind the redundancy. In fact, I think it is helpful as it adds a level of confidence to the term classification (which is always good since Mondo integrates multiple sources).

We could consider removing this redundancy in the released version of the ontology (actually we might already do it), but I personally would keep it.

[sabrinatoro]

I reviewed a few changes in the mondo-edit.obo file, and these seems as expected.
I found new source evidence on existing SC annotations, and new SC annotations+source.

@twhetzel, please take a look, especially for the changes in the subclass.yaml and mondo.Makefile files. Please merge if you approve. Thank you!

[twhetzel]

Two suggested changes/commits

[matentzn]

thx @twhetzel, i will wait for your approval to merge!

[twhetzel]

lgtm