AlterLab AlphaFold (via ColabFold)
Führt AlphaFold2 via ColabFold aus, um proteins/complexes zu falten und pLDDT/pTM/ipTM/PAE confidence zu interpretieren.
Funktioniert mit Setup
Was es kann
Predicts 3D protein structures from a FASTA sequence using AlphaFold2 through ColabFold's MMseqs2-accelerated MSA pipeline, covering single-chain monomers and AlphaFold2-Multimer complexes. Explains how to read per-residue and per-interface confidence metrics and how to validate a designed sequence via self-consistency refolding. Part of a larger AlterLab suite; explicitly defers to sibling skills for ligand co-folding, antibody complexes, embeddings, or database lookups. Requires a CUDA GPU and ColabFold install.
Testbericht
Correctly routes a ProteinMPNN-designed peptide into the self-consistency fold-and-check loop and hands the GPU step off to alterlab-remote-compute -- but actual folding needs a CUDA GPU plus multi-GB AlphaFold2 weights that this sandbox doesn't have, so only the guidance layer could be verified.
Getestet am: 2026-07-14 · Claude Code 2.x (agent harness)
Installation
git clone https://github.com/AlterLab-IEU/AlterLab-Academic-Skills cd AlterLab-Academic-Skills mkdir -p ~/.claude/skills cp -r skills/bioinformatics/alterlab-alphafold ~/.claude/skills/alterlab-alphafold
Befehle & Beispiel-Prompts
/alterlab-alphafoldFührt AlphaFold2 via ColabFold aus, um proteins/complexes zu falten und pLDDT/pTM/ipTM/PAE confidence zu interpretieren.
Skills reagieren auf normale Anfragen — keine Slash-Befehle nötig. Nach der Installation aktivieren Prompts wie diese den Skill (auf Englisch):
Fold this FASTA sequence and give me pLDDT confidence scores.Predict this protein complex structure with AlphaFold2-Multimer.I designed this peptide with ProteinMPNN, check if it folds back.