Bozitao Zhong
Bozitao Zhong
We will add it soon.
We did an overall update for AlphaFold v2.1, so I removed most of the modified scripts to avoid contradiction to the new system. It will be available very soon.
When using multimer models, sometimes a collapsed structure may cause the running time and amber relaxation time very long. Did you look into the predicted computed structure? Is it correct...
you can just look at the output structure, is it looked reasonable? like does it have reasonable secondary structure? also, you can check which step took longest time? is it...
5305 should be the sum of all MSA search time (actually include the PDB search, but that's fast), in the article, we state that the HHblits is the most time-consuming...
If you mean the CPU accerlation, acutally we have not yet add that to this repo, I will upload that soon
I'm not sure about this. Maybe it's the jax version issue as you said, but I didn'tmet this before.
Use the `run_alphafold.py` to replace the same file in your directory, then have another try. If there's still this error, maybe you can send the code for running `run_alphafold.sh` here
> Actually, I recently wrote a scripte to create an empty MSA: https://github.com/Zuricho/ParaFold_dev/blob/main/parafold/create_fakemsa.py To use this, you can use the --use_precomputed_msas to use these empty MSA files
Could you send me your cuda version and jax/jaxlib version? I think you are correct that "jax/jaxlib versions may not be compatible for the CUDA"