Bharath Ramsundar
Bharath Ramsundar
Summarizing an in person discussion. Potential features that it would be useful to add to the UI: - Export results from the docking as CSV file. - Display an options...
There is a potential error in the QM8 dataset from the original MoleculeNet paper caused by duplicate columns (possibly due to a pandas data processing error). https://github.com/deepchem/deepchem/issues/2747 We are still...
Will try to do this over the next week or two. CC @ncfrey
A really nice paper just came out that curates 66 therapeutically relevant datasets: https://arxiv.org/abs/2102.09548. They have documentation and an API up at https://tdcommons.ai/.
A number of the MoleculeNet datasets have PAINS compounds and other compounds that detract from their usefulness as benchmarking datasets. Let's use this issue to brainstorm the set of datasets...
A number of MoleculeNet datasets have obvious false positives attached. For example, the HIV dataset has a number of cytotoxic compounds attached as hits. There are likely a number of...
MoleculeNet has some limitations in its original choice of metrics. Here are a couple issues: - Currently MoleculeNet only tests RMS for its regression tasks. For some tasks, RMS isn't...
We should update the MoleculeNet website with thorough details on how to contribute a dataset to MoleculeNet. Perhaps we could even create a template for the contribution. Most of the...
Here's a cool new dataset from the following paper https://arxiv.org/abs/2006.05531. This dataset would be a really nice contribution to MoleculeNet **GEOM: Energy-annotated molecular conformations for property prediction and molecular generation**...
One of the biggest limitations of the original MoleculeNet was that there was no hidden test set. This means that many of the papers that have used MoleculeNet datasets test...