Can I run these models locally?

Yes, the skill supports local usage of models like esm3-sm-open-v1 and esmc-300m, though larger models require the Forge API for inference.

What is inverse folding?

Inverse folding is the process of designing a protein sequence that will fold into a specific, pre-defined 3D structure.

ESM3 is a generative multimodal protein model capable of designing proteins across sequence, structure, and function tracks simultaneously.

Does this skill support batch processing?

Yes, it provides patterns for efficient batch processing using the Forge API's async executor for high-throughput tasks.

Evolutionary Scale Modeling (ESM)

Name: Evolutionary Scale Modeling (ESM)
Author: Activer007

byActiver007

数据科学与机器学习

Provides comprehensive tools for protein sequence generation, structure prediction, and representation learning using ESM3 and ESM C models.

关于

The ESM skill integrates state-of-the-art protein language models directly into the Claude environment, enabling researchers and developers to perform complex biological tasks such as designing novel proteins, predicting 3D structures from sequences, and generating high-dimensional embeddings. By supporting both local inference for experimentation and the cloud-based Forge API for production-scale tasks, this skill streamlines workflows in protein engineering, drug discovery, and functional annotation, allowing for sophisticated 'chain-of-thought' protein design and multimodal property conditioning.

主要功能

High-efficiency protein embeddings using ESM C models up to 6B parameters
Cloud-based inference via Forge API for scalable protein engineering
Chain-of-thought generation for iterative design and refinement
Inverse folding capabilities to design sequences from target 3D structures
Multimodal protein design with ESM3 across sequence, structure, and function
0 GitHub stars

使用场景

Predicting 3D coordinates and structural pdb strings from amino acid sequences
Designing novel proteins with specific functional properties like fluorescence
Extracting sequence embeddings for downstream machine learning and classification

关于

主要功能

High-efficiency protein embeddings using ESM C models up to 6B parameters
Cloud-based inference via Forge API for scalable protein engineering
Chain-of-thought generation for iterative design and refinement
Inverse folding capabilities to design sequences from target 3D structures
Multimodal protein design with ESM3 across sequence, structure, and function
0 GitHub stars

使用场景

Predicting 3D coordinates and structural pdb strings from amino acid sequences
Designing novel proteins with specific functional properties like fluorescence
Extracting sequence embeddings for downstream machine learning and classification