Jeff Ruffolo

These values are calculated using the [MMseqs2 tool](https://github.com/soedinglab/MMseqs2) to find the closest matches between the generated sequences and the protein databases. We report the identity to the top database hit...

Can you share an example of a fasta file that produces these errors? I can try to reproduce the errors and update the code accordingly.

Hi @hima111997 just wanted to give a quick update. I ran your sequences in the Colab notebook and am getting strange results as well. When I simply remove the linker,...

Hello, you should be able to train using the code here: `deepab/models/AbResNet/train.py`.

Hello, this may be a result of Colab updating to python3.8 in the last couple days. I am working on a fix for this currently.

Hi, any chance you can share the sequence you are trying to predict? I expect that if the notebook is failing to produce decoys then Rosetta is unable to find...