What are the requirements for using this skill?

You need access to NVIDIA NIM microservices for tools like RFdiffusion and ProteinMPNN, which typically require specific API keys.

Does it provide sequences ready for laboratory testing?

Yes, the workflow generates FASTA files containing the designed sequences and ranked candidate reports including quality metrics like pLDDT and pTM.

Can I design binders for specific therapeutic targets?

Yes, the skill supports target-guided binder design by identifying epitopes from PDB or EMDB structures and applying design constraints.

What AI models does this skill use for protein design?

It utilizes RFdiffusion for backbone generation, ProteinMPNN for sequence design, and ESMFold or AlphaFold2 for structural validation of the designs.

Does it support membrane protein design?

Yes, the skill specifically checks EMDB for cryo-EM structures, which are highly valuable for designing binders for membrane proteins and large complexes.

AI Protein Therapeutic Designer

Name: AI Protein Therapeutic Designer
Author: mims-harvard

bymims-harvard

•

1,038

•

Data Science & ML

Designs novel protein therapeutics, binders, and enzymes using state-of-the-art AI models like RFdiffusion and ProteinMPNN.

The AI Protein Therapeutic Designer skill enables the de novo design of functional proteins through a rigorous, multi-phase workflow. It leverages NVIDIA NIM microservices—including RFdiffusion for backbone generation, ProteinMPNN for sequence optimization, and ESMFold or AlphaFold2 for structural validation—to create high-confidence candidates for therapeutic applications. By integrating target characterization from PDB and EMDB with developability assessments, this skill provides a comprehensive pipeline for generating actionable protein sequences ready for experimental testing.

Key Features

01De novo backbone generation using RFdiffusion via NVIDIA NIM

02Sequence optimization and likelihood scoring with ProteinMPNN

03High-accuracy structural validation using AlphaFold2 and ESMFold

041,038 GitHub stars

05Automated reporting with ranked candidates, quality metrics, and FASTA outputs

06Target characterization via PDB and EMDB (cryo-EM) data sources

Use Cases

01Engineering novel enzymes or protein variants for specific biochemical reactions

02Designing high-affinity protein binders for specific therapeutic targets

03Creating stable protein scaffolds for functional motif embedding

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mims-harvard/tooluniverse tooluniverse-protein-therapeutic-design

For use in Claude.ai and ChatGPT

Download Skill