Vignesh Ram Somnath

Should try testing DeepMHC on these datasets.

Hi, Thanks for reaching out! -- I think some classes used to be named differently before, and the data was processed with that. Let me have a look at this...

Yes, I think if you specified `prot_mode` as `surface2backbone` this should construct the full hierarchical graph.

Hi, sorry for the very delayed response here. Are you referring to the latest version on arxiv (https://arxiv.org/abs/2302.11419) for Eq. 5 and Eq. 6? We use Eq. 5 instead of...

I have to give some thought to the first issue suggestion, but for the second one, I was doing something similar during GSOC last year. The ChEMBL dataset had 1M...

@rbharath: If this is not taken up yet, I could do it. I work with the group that developed this, and we are trying to use this for a GAN...

@dannykwells: Sounds good, but it would take a while before I get it ported to DeepChem :)

Just realised, that CellCNN was written in Python 2.7. The actual model would still be easy to get working in Python 3, but the loading of .fcs files is a...

We used the MaSIF code as is, and just updated the data preprocessing for the binding affinity task. Instead of MaSIF, I would recommend trying out dMaSIF(https://github.com/FreyrS/dMaSIF), which uses PyTorch...