Confgen: add option to use symmetry when doing RMS pruning #3813
Conversation
| @@ -1050,6 +1051,49 @@ void embedHelper_(int threadId, int numThreads, EmbedArgs *eargs, | |||
| delete positions[i]; | |||
| } | |||
| } | |||
|
|
|||
| std::vector<std::vector<unsigned int>> getMolSelfMatches( | |||
There was a problem hiding this comment.
I think it would be useful to make this a utility class, something like:
bool symmetry=true;
RMSDSet symm(mol, symmetry);
if (symm.rmsd(conf)) < 0.5) {
symm.add(conf);
}
Perhaps not for this PR.
| if (ssr < ssrThres) { | ||
| return false; | ||
| for (const auto &match : selfMatches) { | ||
| for (auto confi = mol.beginConformers(); confi != mol.endConformers(); |
There was a problem hiding this comment.
Should you loop through confs -> matches or matches -> confs?
There was a problem hiding this comment.
I don't think it matters? Am I missing something?
There was a problem hiding this comment.
The self matches are contiguous in memory and will most likely be faster to access in the inner loop.
There was a problem hiding this comment.
Ah, makes sense.
Given the call to the alignment code that's in the loop, I think this is unlikely to be a bottleneck here. I can change it if you think it'll make a difference
There was a problem hiding this comment.
It makes my premature optimization sensor happy.
|
I don't see any red flags here. This does seem suited for a help class for RMSD which could be generally useful. |
|
@greglandrum Fix the conflicts, and I'll go ahead and merge. I'll test the speed and if there is any hint of a speedup all do a PR |
This adds an option (and makes it the default) to use symmetry when doing RMS pruning (the equivalent of using
getBestRMS()) in the conformation generation code.In order to avoid "uh oh" moments where the symmetry pruning takes an incredibly long time, setting this option causes Hs to be ignored while doing the RMSD calculations (ignoring Hs is the default anyway).