TDC module integration with DeepChem

[kexinhuang12345]

Describe the problem DeepChem has many useful drug discovery ML models. It would be great to implement TDC dataloader in DeepChem such that users in DeepChem can directly utilize TDC datasets.

Describe the solution you'd like DeepChem has dataset class for many MolNet datasets (e.g. https://github.com/deepchem/deepchem/blob/master/deepchem/molnet/load_function/bace_datasets.py). We can modify them to create a meta deepchem.tdc/deepchem.molnet.tdc dataloader and use TDC at the backend to retrieve the csv file in molnet format and feed them to DeepChem data loader. We can first limit our scope to the ADMET, and some other single instance prediction since they have similar format as MolNet.

Additional context Ideally, it would be great to have someone who is familiar with DeepChem to help on this.

[ncfrey]

Hey @kexinhuang12345, I'm a DeepChem dev and I'd be happy to help with this! Let me know if you have any questions about DeepChem, and feel free to tag me in a work-in-progress PR if you want any feedback.

[kexinhuang12345]

Sounds good, thanks so much for offering to help @ncfrey !

[bibs2091]

Any news about this ?

[kexinhuang12345]

@bibs2091 hi we are busy with other big features (coming soon!) so didn't have the bandwidth to do this currently. if you would like to help us, let us know! thanks!