Reproducing CASP results

HeliXonProtein / OmegaFold

OmegaFold Release Code

Apache License 2.0

532 stars 75 forks source link

Reproducing CASP results #31

Open lhatsk opened 1 year ago

lhatsk commented 1 year ago

Hi,

I'm wondering how to reproduce your results in the paper. Might be connected to #14

For example, predicting T1056 with python main.py T1056.fasta output and release1.pt results in a TM-score of 0.54, compared to 0.95 reported in the paper.

Thanks!

lhatsk commented 1 year ago

So I found one issue. You start numbering residues at 0, I was expecting numbering to start at 1. TM-score is now 0.8226, which is still lower.