Open Germanki opened 1 year ago
I also have a similar, although slightly different, NaN issue when i feed it a pdb file and corresponding sequence in fasta file format. I get all NaN in scores. not sure how to debug it.
Update: it was an issue with pdb files at my end. resolved.
Hi, Thank you so much for your work on ProteinMPNN! I have the examples scripts working and I was modifying them (submit_example_4.sh in particular) with my own inputs. However, in the parsed_pdbs.json file that is created my different sequences have both N_chain_A and O_chain_A and every value on the O_chain_A has NaN for all the coordinates. ProteinMPNN still runs and generates outputs for all the sequences but the score, global score, and seq_recovery all = nan (which I am assuming is connected to the O_chain_A=NaN). I imagine that this is due to my pdb files. Please would you be able to explain why this might be happening, what the difference between N_chain_A and O_chain_A is, and how I might be able to fix it? Please do direct me towards any further reading as well! Thank you very much! Adam