nelse003
commented
6 years ago Nicks Adjustments
So I have made the necessary changes and they’re now in the “hotfix” branch on bitbucket. I will test them tomorrow morning with Elliot before the meeting (if possible).
Essentially:
- giant.merge_conformations:
- I realised there was a better way so the default occupancy multipliers for both minor and major states are now 1.0 and there shouldn’t really be any reason to change them AT ANY POINT IN XCE (see the split_conformations point below).
- if reset_occupancies is False (the default), then after merging, an residue with greater than 1.0 occupancy will be scaled down to 1.0 (four conformers of 0.4, 0.6, 0.4, 0.6 from merging go to 0.2, 0.3, 0.2, 0.3 which is sensible).
- if reset_occupancies is True, all alternate conformer occupancies set to 1/n_conformers.
- in all cases, blank conformer atoms are set to unitary occupancy after merging
- giant.split_conformations:
- if reset_occupancies is False (the default), then on splitting, the occupancies are left untouched. thus, after modelling, the re-merging can be done with occupancy multipliers of 1.0.
- if reset_occupancies is True, then if tries to sensibly identify a multiplier to bring the occupancy to approximately one — this one shouldn’t be used much if ever, so probably not that important. maybe a chemist will care.
So… under this schema the input occupancies from pandda.inspect are respected, as it any modification done during modelling. so this should be the minimum amount of fudging required.
Current fix of #91 is a temporary fix that is a non optimal solution to the refinement paradigm. Nick Pearce has implemented changes to the giant.merge_conformation scripts which should allow a better handling of conformers.
Elliot: To test and re implement the hot-fix in XCE