2024-05-28 BW/LW Check-in

General updates and discussion on projects

• Torsion splitting in 2.2

  • Still where it was before OMSF workshop

  • Torsion shapes next steps?

    • Comparing functional form to QM data

• Fragment dataset curation code

  • Ported to Python, found some bugs, re-generating the database

• LW – let’s plan projects and effort allocation over the next year

  • Roadmap is https://docs.google.com/presentation/d/1CYhur-N-M31WKa41VyGZa-3vlV6sinWepod2V-tumNI/edit?usp=sharing

  • Could you please have a look and figure out what you’re interested in by next meeting?

  • Also include infrastructure projects in looking – best way to frame it is specific repos or projects

• Benchmarking

  • BW – differences between AMI – probably from not pruning outliers. AMI was using my scripts. I re-ran Sage 2.2 and TM FFs with most recent YAMMBS version (edit: although before the May revision so still subject to the Molecule.from_inchi bug)

  • LW – will raise in internal-benchmarking the possibility of a shared repo like qca-dataset-submission

• Plans for next week?

  • BW – try to wrap up the dataset code

  • BW – try to put together a dataset

    • BW – computed coverage for new split parameters (TM FF). ~60 not covered by existing TD datasets, roughly same not covered in opt datasets. Planning to cover more of both. Last search only uncovered 20 torsions. Planning to augment by creating molecules.

      • Should be fine to have them in same dataset, but would be nice to label origin of molecule

    • Also helping Lexie with sulfur molecule dataset

• SureChembl – ways to search patent molecules

  • LW – is this a subset of ChEMBL?

  • BW – it’s larger than chembl (~14M compounds) so probably not

• Check on OMSF expenses

• Continue keeping eye out for interesting conferences (domestic preferred)