DeepChem Minutes
There was no Americas/Europe/Africa/Middle East call this week due to the Thanksgiving holiday.
India/Asia/Pacific Call
Date: November 26th, 2020
Attendees: Bharath, Peter, Vignesh
Summary:
Bharath did a little bit of work updating the interaction fingerprints, but this week has been slow with the Thanksgiving holiday.
Peter has finished the project of migrating the moleculenet loaders to the new style after a series of PRs. He’s also done some work on updating the tutorials.
Vignesh has put up a PR with the LCNN platinum adsorption dataset and implementing the featurizer for it. Vignesh has also been experimenting with the model with a Pytorch alone and DGL implementation as well. The Pytorch model seems to match the performance of the authors implementation with 4.00 rmse error. But the dgl model seems to be giving a 0.4 rmse error ( using the same splits). Bharath mentioned that it would be good to check number on the exact split the authors used and Vignesh said he would double check on the authors split.
With roundtable updates completed, we moved to general discussion. Peter mentioned that we should have a discussion of post 2.4.0 priorities with a larger group, likely on a Github thread. Bharath mentioned that physical deep learning might be a useful focus area given that a lot of packages in that ecosystem seem to be in Julia and not in Python.
Joining the DeepChem Developer Calls
As a quick reminder to anyone reading along, the DeepChem developer calls are open to the public! If you’re interested in attending either or both of the calls, please send an email to X.Y@gmail.com, where X=bharath, Y=ramsundar.