#11776 NTR: Thymidylate synthase inhibitor binding

[rach_huntley]

Request on behalf of Hadil Alrohaif;

Can we have a new term "thymidylate synthase inhibitor binding" as a child term of drug binding (GO:0008144)?

Thymidylate synthase inhibitor is in ChEBI (http://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI:63720), but we couldn't request this term via TermGenie as Heiko said TS inhibitors are roles rather than distinct chemicals.

Evidence is from Figure 4 in PMID:19116913.

Please add GOC references;
GOC:BHF
GOC:hal

[David Hill]

Chris,

Assigning to you. I think this is ok if we can get a GOCHE ID for it.

-D

[rach_huntley]

Hi Chris, David,

Is is possible to get this done ASAP because my student needs to get her project finished in the next week or so.

Thanks very much!
Rachael.

[Chris Mungall]

A few issues:

1. CHEBI:63720 ! EC 2.1.1.45 (thymidylate synthase) inhibitor is not a subclass of drug role in CHEBI, so we wouldn't have a justified assertion in GO. (however, all of the children of this biological role are also children of drug in CHEBI, specifically, 'antineoplastic agent'). Is the classification under drug crucial for the student project?
2. I don't think we should be using the CHEBI biological role classes at all, these are redundant with GO. If the goal is to show that the binding affects the thymidine synthase inhibition activity then this is best done with a lego diagram, perhaps following this template:

[binding] -> [ ] -| [thymidine synthase activity]

But it depends what we're trying to capture from the paper. If it's just the raw binding data, is GO the best place for that, and if so, should we not capture the specific compounds (e.g. BCG 945) and not their class? In fact if this is for annotation of FRbeta, I think the paper only demonstrates binding for some subclasses of some folate-based thymidylate synthase inhibitors.

However, it looks like the point of the paper is to model a therapeutic intervention, targeting of FRbeta+ cells in synovial tissue to inhibit rheumatoid arthritis, via binding of a compound to the folate receptor, which then takes the compound up into the cell, where it is toxic to the cell (I think; I skimmed). We could capture this in a LEGO model. Should we? It's fairly different from traditional GO annotation, but maybe actually capturing the therapeutic model is of far more utility to translational medicine than traditional GO annotation?

I don't want to hold up a student project so perhaps we can just manually add the term to the ontology for now with a note saying it may be deprecated in future?