Closed bas-rustenburg closed 6 years ago
Merging #115 into master will decrease coverage by
1.17%
. The diff coverage is86.87%
.
@@ Coverage Diff @@
## master #115 +/- ##
==========================================
- Coverage 64.91% 63.74% -1.18%
==========================================
Files 37 37
Lines 4923 4967 +44
==========================================
- Hits 3196 3166 -30
- Misses 1727 1801 +74
Impacted Files | Coverage Δ | |
---|---|---|
protons/tests/test_ligands.py | 41.46% <ø> (+10.48%) |
:arrow_up: |
protons/tests/test_explicit.py | 94.44% <100%> (+4.94%) |
:arrow_up: |
protons/tests/test_implicit.py | 22% <100%> (-69.9%) |
:arrow_down: |
protons/app/driver.py | 85.16% <100%> (-0.41%) |
:arrow_down: |
protons/app/proposals.py | 67.11% <74.39%> (+17.11%) |
:arrow_up: |
protons/app/analysis.py | 7.53% <0%> (-0.32%) |
:arrow_down: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update bb75527...3c82e3f. Read the comment docs.
This still needed to be merged into master, since it is needed for production runs.
Implementation of ion selection from Chen-Roux 2015 paper, extended to deal with ligands/charge changes larger than one unit. Key thing is that alchemical charges are only added/deleted in one direction, as recommended by Chen and Roux, as opposed to creation/destruction of opposite charges.
The proposal mechanism is called
OneDirectionChargeProposal
.Also adds tests that actually ensure that accumulation of ions doesn't happen. Large improvement over old test is that the ion scheme is tested without actually having to run a simulation. Much faster now, and doesn't require CUDA (or OpenMM really). It uses a fictional MC chain of protonation states from realistic residue types and keeps track of what salt would have been added/removed to maintain charge neutrality.
Should be good to merge if builds pass.