| | |
|---|
PCM support | Josh | PCM support now in QCSubmit, validation included, release cut release includes updates to address Hyesu’s basis set issues TG: Hyesu’s submission no longer blocked; will give it a final review and submitted 0.1.0 is the version, has basis set exchanged included as dependency
|
BCC refit study
| Trevor | Need to resolve question on getting back density from wavefunction components TG: in Daniel’s snippet, taking out all the occupied orbitals, multiplying together to get the densities; if you want to get the entire wavefunction, may be out of scope; if we just want to use Simon’s recharge, should have everything we need TG: once you have dA, can use Daniel’s other script; make sure not using symmetry JH: will send notebook I was using; will work on this together
|
ANI1x, ANI1cxx | Josh | |
Optimization local executor | David+Trevor | Trevor’s Optimization local executor in QCSubmit? QCEngine cli? Trevor will make a PR with the script against QCsubmit; this will start discussion on CLI components, where it can go TG: Does it make sense for this debugging functionality (pulling e.g. an Optimization from a live QCA instance to run locally). BP: first instinct it’s a separate package TG: Pulls ids, runs geometric doesn’t touch the generated task, so no auth required if we just pulled the task, would be a much smaller script
JH: would it make sense to generate the task locally as part of this script? BP: def run_proc_id(proc_id):
task = client.query_tasks(base_result=proc_id)
res = qcengine.compute_procedure(*task.spec.args)
return resÂ
TG: related: Lee-Ping would like a way to get geometric inputs from the corresponding QCArchive tasks specs
|
Enamine REAL | Trevor | 2% of molecules that couldn’t be built some prototypes exist to do the coverage TG: one validation that might become important is a chirality check; we’re seeing some downstream failures when we pull molecules later JH: that’s strange; recent dataset? TG: get a bunch on older datasets; chirality undefined in one case, when you generate the CMILES tags, puts chirality on one center, but then when you load in with OpenEye says it’s non-chiral
JH: no longer require nitrogen stereochemistry; are these nitrogens JH: next step is allow users to provide the CMILES and the molecule separately JH: we could maybe quickly load the CMILES in and try to make the conformers for it TG: if we need to iterate over molecules, do a [m for m in dataset.molecules] (versus Molecule.from_qcschema() from the toolkit via contacting the server directly)
|
Protomer/tautomer enumeration | Josh | John suggested this is important; Pavan will spearhead preparing this with help from Josh One potential dataset for this is the Genentech one Bayer set used for fitting Josh will drop a question into benchmarks-partners for clarification on Bayer and Genentech datasets; former had bits used for fitting, latter is being prepared for submissions (is it the same set they want to use as their public contribution for benchmarking?)
|
STANDARDSv3 | Trevor | |