Does this skill require a GPU to run?

Yes, LigandMPNN requires a CUDA-enabled GPU (minimum 8GB VRAM) for local runs, or it can be executed via Modal to access remote T4 GPUs.

How do I ensure my ligand is recognized by the model?

Ensure your ligand is properly formatted as HETATM records within the input PDB file and has a valid residue name that matches the model's expectations.

What types of ligands are supported by this skill?

It supports any entity represented as HETATM records in a PDB, including small molecules, metals (Zn, Fe, Mg, Ca), cofactors (NAD, FAD, ATP), and nucleic acids.

When should I use LigandMPNN instead of ProteinMPNN?

Use LigandMPNN when your design target includes non-protein elements like small molecules, metal ions, cofactors, or DNA/RNA. Use ProteinMPNN for protein-only scaffolds.

Ligand-Aware Protein Design

Name: Ligand-Aware Protein Design
Author: adaptyvbio

byadaptyvbio

•

Data Science & ML

Designs optimized protein sequences for binding sites and enzymes using Ligand-aware Message Passing Neural Networks.

LigandMPNN is a specialized sequence design tool that enhances protein engineering by incorporating non-protein context—such as small molecules, metal ions, and cofactors—into the design process. It is essential for researchers and developers working on enzyme engineering, drug discovery, or metal coordination, as it allows for the high-confidence optimization of binding pockets and active sites that standard protein-only models might overlook. By leveraging Claude Code, users can automate the generation of thousands of sequences tailored to specific chemical environments, significantly accelerating the early stages of protein-ligand design.

Key Features

01Ligand-aware sequence generation for small molecules and cofactors

0275 GitHub stars

03Precise metal coordination site and active site optimization

04Seamless integration with Modal for serverless GPU scaling

05High-throughput design generating 50-100 sequences per minute

06Native support for DNA/RNA and HETATM records in PDB files

Use Cases

01Creating protein scaffolds for metal ions like Zinc or Iron

02Optimizing enzyme active sites to improve catalytic efficiency

03Designing high-affinity binding pockets for pharmaceutical small molecules

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add adaptyvbio/protein-design-skills ligandmpnn

For use in Claude.ai and ChatGPT

Download Skill