protenix

Name: protenix
Availability: InStock
Author: LiorZ

Fold biomolecular complexes from sequences and ligand specifications into CIF structures.

2|Updated May 12, 2026

Search Tags:#diffusion-model #apptainer #protein-design #structure-prediction #alphafold3 #biomolecular-complexes #msa-templates

Ready to run

npx skills add https://github.com/LiorZ/protein-design-skills --skill protenix-liorz

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: protenix
Source: https://github.com/LiorZ/protein-design-skills/tree/main/skills/protenix
Command: npx skills add https://github.com/LiorZ/protein-design-skills --skill protenix-liorz

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Protenix provides end-to-end all-atom structure prediction for biomolecular complexes, enabling users to fold proteins, DNA/RNA, ligands, and ions in a single pass from descriptive inputs.

Core Features & Use Cases

Multi-entity folding: protein–DNA/RNA–ligand complexes from sequences and ligand specs.
PTMs, covalent bonds, and soft constraints to model realistic chemistry and interfaces.
MSA/template/RNA-MSA support with presets and ranking via AF3-style confidence scores.

Quick Start

Run a GPU-accelerated prediction from an input JSON using a Protenix model with default parameters.

Frequently Asked Questions about protenix

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

FAQPage Schema

How do I predict biomolecular complex structures from sequences and ligand specifications?▼

Protein-DNA/RNA-ligand folding is handled by processing an AlphaFold-Server-style input JSON through a Protenix model. This end-to-end all-atom prediction outputs CIF structures and per-sample confidence scores for the entire biomolecular complex.

What is an AlphaFold3-style structure folding model?▼

An AlphaFold3-style structure folding model is a trainable diffusion-model architecture that predicts all-atom structures for biomolecular complexes. It applies MSA templates and soft-pocket constraints to generate realistic chemical interfaces and CIF structures.

Do I need a containerized runtime to fold protein-DNA-RNA-ligand systems?▼

Yes, folding protein-DNA-RNA-ligand systems requires a containerized runtime like Apptainer to execute GPU-accelerated predictions. This environment isolates dependencies required by the diffusion model to process input JSON and generate CIF structure outputs.

Can I model post-translational modifications and covalent bonds in protein-ligand complexes?▼

Yes, you can model post-translational modifications and covalent bonds in protein-ligand complexes. The folding pipeline explicitly supports these realistic chemistry features alongside soft constraints to accurately predict structural interfaces and CIF outputs.

Does the structure prediction pipeline support MSA templates and RNA-MSA inputs?▼

Yes, the structure prediction pipeline supports MSA templates and RNA-MSA inputs. It uses these evolutionary cues with specific presets to rank the generated all-atom structures via AF3-style confidence scores and output CIF files.