JW – Local packaging. Qcsubmit basis-set-exchange pip install issues. Meetings about OpenMM conda-forge migration. Contracting a conda-forge GPU dev. Working on nitrogen stereochemistry. Working w/ Cerutti on FF port. Docker example of distributed QCF compute for Galileo folks.
JRG – Best thing to do is submit basis-set-exchange to conda-forge – JW can do this, but we can as BP first.
MT – Maintenance and toolkit work. FB publication on C-F. Not much progress on System. Close to doing vacuum energy. Electrostatics are a bit strange – Instead of holding a “parameter” object for charges, it just holds a number.
DD – Automating lifecycle for dataset submission to QCA (showed to Josh + Ben + Trevor). Uses state-based approach. Will be working with JH to add validation workflow. TG + JH are working on validators for protein component submissions. Goal is to remove human intervention.
DH – Working on manuscripts, Victoria Lim’s QM structure comparison, and LiveCOMS benchmarking paper. Also new targets for PL benchmark.
SB – Working on evaluator publication. Nitrogen BCCs are almost totally mapped into new repo. Lots of difficulty wrt nitrogen bonds because many deal with electron delocalization. Process of generating ESPs is nearing completion. Using Psi4 to generate ESPs, sqllite to hold data, pytorch to fit. We get optimized conformers from OE, which are usually based on ELF10 and wouldn’t optimize much.
JW – QCA ESP grid gen?
SB – Should be suitable for that, but I’m not making it specifically for that purpose.
JRG – Done with teaching for summer semester. Started building RDKit for PowerPC, should be out next release. Progress with OpenMM conda-forge migration, relying on making a blocking/warning mechanism for janky builds. Working on dockerizing projects. for ML models in my lab. Could use molecule class inheritance issue fix in 0.7.1 if possible.
DH – Will System object handle FE calcs?
MT – I’d like it to handle that, but I don’t know specifically what it’ll need to suppoer
DD – Could take a look at alchemlyb
DH – Given a protein w/ two ligands (where we’ll sample transformations from one to the other), it would be good to have the System understand the parameters for both.
SB – Some sort of container object, either IN System class, or CONTAINING two System classes, with the ability to provide lambda states, will cover a ton of use cases. Can see YANK/PERSES for example of this.
MT – DH, please add any more hints or guidance you can to system architecture doc.
Sprint Planning
Task nomination
JW
Zenodo is broken
Conformer gen script
Benchmarking ….?
Vsites validation+friendly user period
Docs link fixes
JRG
Molecule class inheritance issue fix in 0.7.1 if possible
Add Comment