Design protein sequences for given backbone structures using ProteinMPNN.
npx skills add https://github.com/junior1p/ProteinClaw --skill proteinmpnn-junior1p
Please help me install this Agent Skill. Skill: proteinmpnn Source: https://github.com/junior1p/ProteinClaw/tree/main/skills/proteinmpnn Command: npx skills add https://github.com/junior1p/ProteinClaw --skill proteinmpnn-junior1p
This Skill automates designing high-quality protein sequences for provided backbone structures, reducing manual iteration and accelerating binder and protein engineering workflows.
Generate 8 sequences for backbone.pdb with sampling temperature 0.1 and save outputs to the output folder.
High-intent search queries and answers about installing and using this skill.
To design protein sequences for a backbone PDB structure, you can use inverse folding to generate multiple candidate sequences. This process allows adjustable sampling temperature and sequences-per-target, outputting FASTA files and designed PDBs.
Inverse folding is the process of identifying optimized amino acid sequences that fold into a given 3D backbone structure. You need it for de novo backbone generation, sequence redesigns with fixed-residue constraints, and ligand-aware or solubility optimization scenarios.
Yes, you can fix specific residues during protein sequence redesign. The Skill supports flexible design modes including fixed-residue constraints, per-chain targeting, multi-state design, and negative design within a single PDB input workflow.
Yes, you need a CUDA-enabled GPU and Python 3.8 or higher for performant execution. While the inverse folding task is computationally intensive, the Skill provides guidance for common CUDA out-of-memory errors to help manage GPU resources.
To handle chain ID mismatches during backbone PDB processing, the Skill provides specific troubleshooting guidance. It supports per-chain targeting to ensure the inverse folding model correctly interprets your multi-chain PDB inputs and generates accurate sequences.
Sampling temperature controls the diversity of generated protein sequences; lower values like 0.1 produce conservative designs while higher values increase variability. You can configure this alongside the sequences-per-target count to optimize expression and structural stability.
InternScienceRebuild missing atoms in protein PDB structures for docking and simulation preparation.
PKU-YuanGroupPredict protein, nucleic-acid, and small-molecule complex structures with Boltz-2.
zongtingweiDesign solubility-optimized protein sequences for a given backbone using SolubleMPNN.
jaechang-hitsGenerate protein sequences, predict structures, and create embeddings with ESM models.
