Converts PDF documents to clean, structured Markdown.
npx skills add https://github.com/PKU-YuanGroup/OpenAI4S --skill proteinmpnn-pku-yuangroup
Please help me install this Agent Skill. Skill: proteinmpnn Source: https://github.com/PKU-YuanGroup/OpenAI4S/tree/main/skills/proteinmpnn Command: npx skills add https://github.com/PKU-YuanGroup/OpenAI4S --skill proteinmpnn-pku-yuangroup
This Skill allows for the inverse-folding of protein backbones into amino-acid sequences using ProteinMPNN, aiding in sequence design and redesigning specific chains of a PDB while holding interface residues fixed.
Run the proteinmpnn skill with the following command:
python protein_mpnn_run.py --pdb_path backbone.pdb --pdb_path_chains "A" --out_folder out --num_seq_per_target 16 --sampling_temp "0.1"
High-intent search queries and answers about installing and using this skill.
To inverse fold a protein backbone into an amino-acid sequence, run the script with your PDB file path and target chain. ProteinMPNN processes the structure to generate corresponding sequences for downstream folding.
Yes, you can redesign a specific protein chain while keeping interface residues fixed. The ProteinMPNN script allows you to specify target chains in your PDB file to preserve critical interaction interfaces during sequence design.
To generate multiple sequence variants for downstream folding, use the sampling temperature parameter. ProteinMPNN creates temperature-swept sets of sequences, allowing you to specify the number of sequences generated per target.
Protein sequence design requires a PDB file format as input. You provide the backbone structure via the pdb path argument, and ProteinMPNN reads this structural data to perform the inverse folding.
Yes, ProteinMPNN works with RFdiffusion backbones. You can run sequence design directly on the generated structures by passing the RFdiffusion output PDB files to the inverse folding script.
To run ProteinMPNN, you need PyTorch and NumPy installed in your environment. These dependencies support the message-passing network operations required for protein structure prediction and sequence generation.
InternSciencePredict full-atom protein sidechain conformations from backbone PDB files.
Lord1EgyptPredict 3D protein conformations from amino acid sequences using AlphaFold2 and ESMFold.
mims-harvardDesign protein therapeutics with RFdiffusion, ProteinMPNN, and ESMFold validation.
adaptyvbioDesign protein sequences from structural backbones using ProteinMPNN inverse folding.