Urinx / alphafold_pytorch

An implementation of the DeepMind's AlphaFold based on PyTorch for research
Apache License 2.0
392 stars 92 forks source link

How can I get the pkl file? #17

Open ibr1996 opened 3 years ago

ibr1996 commented 3 years ago

Hi, in order to run a prediction for other target using the alphafold.sh script, I need as input a pkl file, how can I get it?

LeopoldWizard commented 3 years ago

You can look for your aminoacid sequence in any protein repository (such as UniProt) to get a FASTA format file such as: https://www.uniprot.org/uniprot/P07724.fasta. Download and write it to a file named "protein_name.seq". Then that is the ".seq" file you need as input for alphafold.sh to generate a ".pkl" file, which is the input for alphafold.py, not for alphafold.sh.

ibr1996 commented 3 years ago

Thanks for your reply, but, the script feature.sh just give me a .npy file, no .pkl o .tfrec file is generated, but it seems like the alphafold script accept a .npy file

yuxulingche commented 3 years ago

I have the same doubt, I want to know how to generate .pkl and .tfrec file, thanks.

elephantpanda commented 1 year ago

You can look for your aminoacid sequence in any protein repository (such as UniProt) to get a FASTA format file such as: https://www.uniprot.org/uniprot/P07724.fasta. Download and write it to a file named "protein_name.seq". Then that is the ".seq" file you need as input for alphafold.sh to generate a ".pkl" file, which is the input for alphafold.py, not for alphafold.sh.

Hi @LeopoldWizard I am currently implementing this in Unity. What I am lacking is a few very small examples of the order about 64-100 amino acids. I am looking for someone who has managed to generate any pkl files. Or even have a Google Colab running that can run the feature.sh file to generate more examples. I have got everything else working, it's just generating the pkl files which has me stumped. Any help would be welcome thanks!

My colab code so far: https://colab.research.google.com/drive/16cFl10rETmCXTdeq91sdqE81IROiEQQY?usp=sharing