search

nrbennet / dl_binder_design

MIT License
203 stars 49 forks source link

Is it possible to inpaint a missing part of a protein and designing sequences using the dl_binder_design? #43

Closed wangdoc2002 closed 10 months ago

wangdoc2002 commented 11 months ago

Hi all,

I am inpainting and repairing part of a protein. Previously, this was performed using RFDesign (https://github.com/RosettaCommons/RFDesign). Now the repo suggests using RFdiffusion instead. So I inpainted 14 residues of a protein using RFdiffusion and generated a bunch of structures. They look beautiful. Now I need to generate AA sequences of the structures using ProteinMPNN-FastRelax protocol to do sequence design, as suggested by the RFdiffusion repo.

After fed the pdb files to the "dl_interface_design.py", I got error messages like this: PyRosetta-4 2023 [Rosetta PyRosetta4.conda.linux.cxx11thread.serialization.CentOS.python39.Release 2023.38+release.52c4cf62073872b07f5cf0623533318b147b5910 2023-09-19T15:49:21] retrieved from: https://urldefense.com/v3/__http://www.pyrosetta.org__;!!K-Hz7m0Vt54!g-kftWD_6n-wmi1E_dFBo61k6nznk11FFjQKUk5LBifU2FiEtos7BqGHbuL4NkDIC2o1zjgBESZeu-qCf1hK6gc$ (C) Copyright Rosetta Commons Member Institutions. Created in JHU by Sergey Lyskov and PyRosetta Team. Found GPU will run ProteinMPNN on GPU Attempting pose: /dssg/home/acct-clsljd/clsljd/bin/RFdiffusion/examples/example_outputs/YbtE/0925/8.pdb Struct with tag /dssg/home/acct-clsljd/clsljd/bin/RFdiffusion/examples/example_outputs/YbtE/0925/8.pdb failed in 1 seconds with error: <class 'IndexError'>

These are the commands I used to run the program: module load miniconda3 source activate dl_binder_design

python /dssg/home/acct-clsljd/clsljd/bin/dl_binder_design/mpnn_fr/dl_interface_design.py \ -pdbdir /dssg/home/acct-clsljd/clsljd/bin/RFdiffusion/examples/example_outputs/YbtE/0925 -outpdbdir /dssg/home/acct-clsljd/clsljd/bin/RFdiffusion/examples/Design/YbtE/0925

Thanks for reading this issue thread.

nrbennet commented 10 months ago

This repository is specifically designed to work with protein complexes and does not work with protein monomers. You are getting an error related to this. I would suggest looking into ColabFold (https://colab.research.google.com/github/sokrypton/ColabDesign/blob/v1.1.1/rf/examples/diffusion.ipynb) if you would like an easy way to run ProteinMPNN and AlphaFold2 on your designs.