Skip to content

Openmm pypi#24

Open
tristanic wants to merge 36 commits into
masterfrom
openmm-pypi
Open

Openmm pypi#24
tristanic wants to merge 36 commits into
masterfrom
openmm-pypi

Conversation

@tristanic
Copy link
Copy Markdown
Owner

No description provided.

In the original set of ligand AMBER parameterisations done
by Moriarty & Case, the -NH2 group of guanine was
incorrectly assigned as NH3. This commit corrects that for
all current ligands in the components dictionary containing
"guanosine" in their name. The underlying issue is a wider
one affecting various other ligands; correcting this
completely needs a more considered, general approach.
Forcefield loading was previously happening in a separate thread
managed by `ThreadPoolExecutor` to reduce impact on main-thread
performance. For unknown reasons, this seems to *very* rarely lead
to a segmentation fault when loading the pickled force field. The
good news is that the hit to performance is minimal - I added this
threaded approach at a time when all the ligand definitions were
pre-loaded into the main force field making it very slow to
instantiate. Now that ligands are only added as-needed, startup
time is essentially negligible.
Documentation says that it resets back to the original ffXML files, but
it was only resetting back to the cached *pickle* file.
The RCSB ligand-expo server recently stopped serving these templates describing various amino acid residue protonation variants. Since they're currently used by ISOLDE's code for rebuilding incomplete residues, I'm bundling in an offline version of this template subset.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant