JW – Shirts lab folks (Anika and Tim) are working with polymers:
MS: I'm sure Tim Bernat would be happy to help out with the template design/recognition work/testing, and I think Anika has some covalently bonded ligands to test.
JM – Would be good to see more use cases if they can provide them. Do either have branched polymers?
JW – Tim probably would. I’ll ask them for their examples.
JM – Currently substructure matching is only within residues. So this will be hard.
JW – That’s OK, if it’s impossible we can keep the old code going
JM – If there’s a topology with (in order) protein A, B, and C, and there’s a crosslink between A and C, then we can’t create an OpenFF Topology with B having atom indices in the middle of A and C (Since AC is a single molecule, it must be contiguously indexed in the topology). Currently the behavior will just be that the atoms of C get indexed right after A, and B will go at the end of the topology.
JW – What an interesting case. I think that’s the correct behavior given our other constraints. Would be good to emit a warning when this happens and store original indices in atom metadata
JM – Warning sounds good, and I already store original index in metadata.
JM – Reviews on pablo?
JW – I don’t think we should do reviews on pablo while you’re doing a lot of code writing whole-cloth. But we will want to move to a model of reviews once changes become more modest.
JM – And it’d be good to get an overall architectural review once things stabilize.