parts vs is_a parents

[vanaukenk]

This is a follow-up to discussions we had on the 2020-11-30 ontology editors call.

Molecular functions can have different 'parts' or sub-functions, but it seems that sometimes these parts are is_a parents in the ontology and sometimes not.

For MFs like NTP-dependent molecular motors or NTP-dependent helicases (specifically I'm thinking of DEAD-box helicases), the gene products do enable an NTPase activity but NTP binding and protein or nucleic acid binding are critical parts of the overall function.

From: https://pubmed.ncbi.nlm.nih.gov/19748356/

In this example of a DEAD-box helicase, could the ATP binding and dsRNA binding also legitimately be 'is a' parents for the RNA helicase MF?

A possible GO-CAM representation of the DEAD-box helicase activity:

http://noctua.geneontology.org/editor/graph/gomodel:5fb9cc0600000424?model_id=gomodel:5fb9cc0600000424

[pgaudet]

I think (but I am not sure) that we were proposing to have all binding activities as 'has_part' of a more descriptive function; weren't we ?

[ValWood]

My understanding (mainly from reading in between the lines, but this is what seems to be a consensus)

1. We aren't modelling 'conformations change' which is important for the activity but represents mechanistic detail below the resolution of GO MF
2. So, in the above example the activity would be a helicase?. This MF would have part ATP binding

so isn't this just helicase Catalysis of the reaction: ATP + H2O = ADP + phosphate, to drive the unwinding of a DNA or RNA helix.

3. I also thought the plan was for binding to be has_part of more descriptive functions.

When I was reviewing the tracker last week I found a ticket where @thomaspd confirmed this. I will see if I can locate it. and I found another ticket where I also asked the question.

[pgaudet]

@vanaukenk Can we ~you~ say this is resolved with the decision to use 'is a' for ATPases'?

[pgaudet]

@thomaspd your input would be useful

[pgaudet]

Discussion on ontology jamboree

• @thomaspd Do you need to capture ATPase as evidence ? This seems 'obvious' information
• @krchristie feels is important to capture the where the data come from
• @thomaspd for helicases, part of its activity is an ATPase
• talk about the suggestion we had previously, to have two versions of the ontology, one for function (helicase has_part ATPase), one for classification (ie a helicase is_an ATPase) - but we had decided on an ontology call a few weeks ago to make all functions using ATP is_a ATPase activity

[pgaudet]

May be solved with #20876

• to be discussed again once we look at the PR

[hattrill]

tagging @sjm41

[pgaudet]

Hi @vanaukenk

Now that we created the distinction between ATP hydrolysis activity and ATPase activity, can this close?

20876

Thanks, Pascale

[vanaukenk]

@pgaudet - yes, I'll close it. Thx.