Bharath Ramsundar
Bharath Ramsundar
There are a number of important molecular and scientific models that I would love to see implementations for in DeepChem. Implementations of any of these models would be a great...
- [ ] A tutorial on DeepChem dataloaders and how a DataLoader can be used to load in a new dataset - [ ] A tutorial on deep microscopy, focusing...
The DeepChem tutorial suite is emerging as a powerful resource for community members. We should make each DeepChem tutorial citeable by adding a bibtex with the author(s) of the tutorial...
The following code snippet: ``` model = dc.models.GraphConvModel(12, batch_size=50, mode='classification') mols = ['C', 'CCCC'] # Train the model here feat = dc.feat.ConvMolFeaturizer() X = feat.featurize(mols) model.predict(dc.data.NumpyDataset(X)) ``` fails with the...
TDC (Therapeutic Data Commons) https://github.com/mims-harvard/TDC is a new dataset collection that open sources a collection of therapeutically relevant datasets. It would be very nice to have a tutorial introducing how...
https://github.com/DeepGraphLearning/torchdrug is a new Torch framework for drug discovery. It might be interesting to make a tutorial for the tutorial series introducing TorchDrug and how it can be used in...
Contributing a model to DeepChem requires a good bit of work. Adding a guide that walks new developers through the process of contributing a model starting from an open source...
We're currently using hyperopt in the MoleculeNet hyperparameter search code (https://github.com/deepchem/moleculenet/blob/master/examples/gnn.py#L136). This offers some nice capabilities that DeepChem's current hyperopt code doesn't support. We should add a Hyperopt based hyperparameter...
A common issue DeepChem users have is that they wrote code for an old version of DeepChem which no longer runs on new DeepChem. We've taken a lot of effort...
DeepChem has many featurizers and many models. It would be really useful to add a table to the docs which lists which featurizers can be used with which models. This...