How does BindCraft compare to RFdiffusion?

BindCraft is preferred for production-quality campaigns where joint sequence optimization and built-in AF2 validation are required, whereas RFdiffusion is primarily used for backbone-only generation.

Can I choose between speed and design quality?

Yes, the skill supports 'fast', 'default', and 'slow' protocols, allowing you to choose between quick initial screening and high-precision production runs.

Does this skill require specialized hardware?

Yes, BindCraft requires high-performance GPUs with significant VRAM (recommended 48GB L40S or A100) due to the intensive AlphaFold2 validation steps.

What metrics are used to validate designs?

The skill outputs several key metrics including pLDDT (confidence), ipTM (interface quality), and RMSD, allowing researchers to rank and select the best candidates for synthesis.

What is the primary use of the BindCraft skill?

BindCraft is used for the automated end-to-end design of protein binders, optimizing both the protein backbone and sequence simultaneously for high binding affinity.

BindCraft Protein Binder Design

Name: BindCraft Protein Binder Design
Author: adaptyvbio

byadaptyvbio

•

データサイエンスとML

Generates high-affinity protein binders using end-to-end AlphaFold2 hallucination and rigorous structural validation.

BindCraft is a specialized Claude Code skill for computational biologists and researchers focusing on protein-protein interaction design. By leveraging AlphaFold2-based hallucination, it performs joint optimization of both protein backbones and sequences, significantly increasing the likelihood of experimental success compared to traditional methods. The skill manages complex workflows, from defining target hotspots and binder lengths to executing multi-tiered protocols—fast, default, or slow—and providing deep analytical metrics like pLDDT and ipTM for every design trajectory.

主な機能

01End-to-end binder design with joint backbone and sequence optimization

02Built-in AlphaFold2 validation for high experimental success rates

03Comprehensive metrics reporting including pLDDT and ipTM scores

0475 GitHub stars

05Three tiered design protocols (fast, default, slow) for speed vs. quality

06Automated hotspot targeting and binder length configuration

ユースケース

01Designing production-quality binders for therapeutic target proteins

02Running high-throughput binder design campaigns with automated QC

03Generating validated protein structures for experimental synthesis

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add adaptyvbio/protein-design-skills bindcraft

For use in Claude.ai and ChatGPT

Download Skill