cyrusmaher

Benchmarking confidence intervals

1

For my dataset, I tried correlating the CIs to absolute error on the test set, and didn't find a relationship. I do get a relationship if I use the standard...

Running inference outside of pre-computed datasets

4

Dear authors, Congratulations on Uni-Mol! This is truly exciting work 🥇 I'm writing to express interest in running inference outside of the pre-computed datasets. For example: 1.) Running docking from...

Type requirements are overly specific

1

Requires 64 bit values when 32 bit would work fine. I believe this is a legacy issue from missing functionality in Julia v0.1...

String kernels can exploit biases in SMILES string format, skewing performance metrics

4

Just a heads up on this issue: https://github.com/deepchem/moleculenet/issues/15 I propose that string-based classifiers canonicalize smiles prior to processing to prevent confounded performance, CI, etc. estimates.

Following up on a conversation with Meng Liu, I wanted to link this bug. I confirmed it for ClinTox, but it may be present for other datasets: https://github.com/deepchem/moleculenet/issues/15 One set...

Non-canonical smiles confound string-based classifiers

10

Running a string kernel classifier on the clintox dataset, I can obtain an AUROC of 0.96. When I canonicalize the smiles, my AUROC drops to 0.69. This implies that there...