RDKit SMILES discussion JW: for the future there’s a context manager that changes the GLOBAL_TOOLKIT_REGISTRY https://github.com/openforcefield/openff-toolkit/blob/74cfe94e8cda29e933866b8d895849073ba00689/openff/toolkit/utils/toolkit_registry.py#L466 JW: Does Genentach just need a pile of structures or is this for a forcefield that needs SMILES? It seems like this work is focused on the SMILES. LW: The former is the bare minimum, but we are looking for making this dataset useful in the future JW: OpenFF doesn’t want to handle metals. You should focus on an external pipeline. How are the SMILES handling metal bonds? JAC: Mostly “<-“ sometimes radical JW: I’m in favor of using the “->” for dative bonds, (with low confidence) this seems like the best option for eventually being compatible with smirnoff-land … JW: happy to pair-program tomorrow on this
QCSubmit keys discussion LW: In Genentech discussion it came up that our pipeline is set up to include many many things like inchiley and 4 different types of smiles, is any/everything used for indexing or important things? JW: We included everything just in case, but no one normally used it. When submitting the dataset just have a disclaimer that the dataset isn’t standard JW: have the pdf and list of smiles and let the qca-dataset-submission simply be note taking LW: I think our CI is independent of qcsubmit… I think it is JW: I think it might do some validation thing… you can try…but it might be easier not to. LW: Jen let us know if you need help submitting directly. LW: we should probably just update QCA-DS CI to not go through QCSubmit – it’s broken in a couple other ways as well, anyway
TMFF dataset discussion PVC discussion
|