ClusterProteinDesign

This is a personal repository for running protein design scripts on Slurm. The scripts are designed to use various tools such as RFdiffusion, ProteinMPNN, AlphaFold2, Rosetta, etc..

Usage:

To use the scripts from this repository, one must install several different things:

• RFdiffusion
• ColabDesign

What worked for me was installing cuda before colabdesign:

conda create -n colabdesign python=3.9 pip
conda activate colabdesign
pip install --upgrade "jax[cuda11_pip]" -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
pip -q install git+https://github.com/sokrypton/[email protected]

#Downgrade to match cluster version of cuda
pip install nvidia-cuda-nvcc-cu11==11.7.99
pip install nvidia-cuda-runtime-cu11==11.7.99

• PyRosetta/Rosetta
• ColabFold
• RoseTTAfold2

Note: I linked my fork that has some additional scripts

• ESMfold

Pipelines

• ColabFold AF2 predictions

  • Predict single fasta/a3m file
  • Predict pdb files and compare

• Protein design with ProteinMPNN and AF2

  • MPNN-AF2 pipeline:

    • ProteinMPNN
    • AF2 prediction (ColabFold)
    • Structure analysis

  • Fix conserved positions MPNN-AF2 pipeline:

    • based on article: Sumida2024 - Improving Protein Expression, Stability, and Function with ProteinMPNN

• Binder design with RFdiffusion, ProteinMPNN and AF2

  • RF-MPNN-AF2 pipeline:
    • Binder docking:
      • Random binder docking
      • RFdiffusion scaffold docking
      • RFdiffusion random binder scaffold docking
    • Binder optimization
    • Binder analysis
    • Binder filtering

• Binder design with partial diffusion

• Motif scaffolding