skills.rest
PKU-YuanGroup
PKU-YuanGroupOFFICIAL
Explore all skills by this author →
WebsiteGitHub

fair-esm2

Embed protein sequences with Meta AI's ESM-2 model.

288|34|Updated Jul 6, 2026
Search Tags:#bioinformatics#drug discovery#ESM-2#protein embeddings#protein contact prediction
One-click install
npx skills add https://github.com/PKU-YuanGroup/OpenAI4S --skill fair-esm2-pku-yuangroup
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: fair-esm2
Source: https://github.com/PKU-YuanGroup/OpenAI4S/tree/main/skills/fair-esm2
Command: npx skills add https://github.com/PKU-YuanGroup/OpenAI4S --skill fair-esm2-pku-yuangroup

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires torch, esm, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill allows you to embed proteins with Meta AI's ESM-2, enabling you to perform tasks such as extracting embeddings for ML models, scoring masked-LM likelihoods, and predicting contacts from sequences.

Core Features & Use Cases

  • Protein Embeddings: Generate embeddings for proteins using the ESM-2 model from Meta AI.
  • Masked-LM Scoring: Calculate likelihood scores for masked-LM scenarios and mutation effects.
  • Contact Prediction: Predict contacts between residues in a protein sequence.
  • Use Case: If you're working on a drug discovery project and need to understand the 3D structure of a protein, you can use this skill to generate embeddings for the protein and analyze its potential binding sites.

Quick Start

Embed a protein sequence 'ubq' into embeddings using the fair-esm2 skill.

Frequently Asked Questions about fair-esm2

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract protein embeddings using ESM-2 for machine learning?▼

Extract protein embeddings by passing amino acid sequences through Meta AI's ESM-2 model to generate numerical representations for downstream ML tasks like function prediction. This requires a CUDA-enabled GPU with 8 GB or more VRAM.

Can I predict protein contacts from sequences using ESM-2?▼

Predict protein contacts by processing amino acid sequences with the ESM-2 model to infer residue interactions. This functionality supports structural analysis for applications like drug discovery and binding site identification.

What GPU and environment do I need to run ESM-2 protein contact prediction?▼

Running ESM-2 for protein contact prediction requires a CUDA-enabled environment and a GPU with at least 8 GB of VRAM. The underlying dependencies include PyTorch and the ESM package.

Can I score masked-LM likelihoods and mutation effects with ESM-2?▼

Score masked-LM likelihoods and mutation effects by evaluating sequence positions within the ESM-2 framework. This provides quantitative metrics for analyzing protein sequence variations and functional impacts.

How does ESM-2 compare to other bioinformatics tools for drug discovery?▼

ESM-2 provides advanced protein embeddings and contact predictions directly from sequences using Meta AI's transformer model. It is specifically designed for structural analysis and function prediction in bioinformatics workflows.

Why does ESM-2 protein embedding require at least 8 GB of VRAM?▼

ESM-2 protein embedding requires 8 GB or more of VRAM because the transformer model processes complex sequence data and computes deep representations that demand significant CUDA-enabled GPU memory resources.

Related Skills

Explore More Education & Research Skills→
moltis-orgmoltis-org
2.8k

heartmula

Generate music locally from lyrics and tags using HeartMuLa models.

Official
Advanced
Lord1EgyptLord1Egypt
2

protein-structure-prediction

Predict 3D protein conformations from amino acid sequences using AlphaFold2 and ESMFold.

Community
Advanced
jonnabiojonnabio
6

diffdock

Predict binding poses and confidence scores for protein-ligand interactions.

Community
Advanced
adaptyvbioadaptyvbio
151

chai

Predict protein structures from FASTA sequences using Chai-1.

Official
Advanced