protflow

Orchestrate protein-design pipelines across multiple tools via a unified API.

2|Updated May 12, 2026
One-click install
npx skills add https://github.com/LiorZ/protein-design-skills --skill protflow
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: protflow
Source: https://github.com/LiorZ/protein-design-skills/tree/main/skills/protflow
Command: npx skills add https://github.com/LiorZ/protein-design-skills --skill protflow

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

ProtFlow provides a unified orchestration layer to build and run end-to-end protein-design campaigns by coordinating multiple design tools behind a single API, tracking results, and enabling reproducible workflows.

Core Features & Use Cases

  • Orchestrates multi-step design pipelines across RFdiffusion, LigandMPNN/ProteinMPNN/SolubleMPNN, ESMFold, ColabFold (AF2), AlphaFold3, Boltz, Caliby, Frame2Seq, ProteinGenerator, PLACER, AttnPacker, SigmaDock, Rosetta, GROMACS, DSSP, FPocket, Propka, TM-align, RMSD/biopython metrics behind a single Poses + JobStarter + Runner API.
  • Supports SLURM array jobs and local execution, with per-pose tracking in a pandas-backed Poses object and a consistent on-disk layout for inputs, intermediate outputs, and scores.
  • Tracks residue motifs across generative steps using ResidueSelection and remapping after diffusion, enabling motif persistence through design campaigns.
  • Extends pipelines by subclassing Runner to add new tools, wiring tool paths and Python environments via a config system, and robust debugging wrappers for failure recovery.

Quick Start

Load input poses and run a sample multi-tool pipeline (RFdiffusion → LigandMPNN → ESMFold) with a unique prefix and a configured jobstarter.

Frequently Asked Questions about protflow

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

FAQPage Schema
How do I orchestrate a multi-tool protein design pipeline?

You orchestrate a multi-tool protein design pipeline by unifying multiple design tools behind a single API. This skill coordinates tools like RFdiffusion, LigandMPNN, and ESMFold while tracking results in a pandas-backed Poses dataframe for reproducible workflows.

Can I run protein design workflows on a SLURM cluster?

Yes, you can run protein design workflows on a SLURM cluster. The orchestration layer supports SLURM-based array jobs alongside local execution to scale protein-design campaigns across high-performance computing environments.

How do I track residue motifs across diffusion steps in protein design?

To track residue motifs across diffusion steps in protein design, the pipeline uses ResidueSelection and remapping after diffusion. This enforces motif persistence through the various generative steps of a design campaign.

What is the best way to add new bioinformatics tools to an existing pipeline?

The best way to add new bioinformatics tools to an existing pipeline is by subclassing the Runner class. This extension method allows you to wire tool paths and Python environments via a config-driven path resolution system.

Does this pipeline orchestration support AlphaFold3 and Boltz?

Yes, this pipeline orchestration supports AlphaFold3 and Boltz. It unifies these structure prediction tools with other design utilities behind a single Poses, JobStarter, and Runner API for standardized score merging.

How do I recover from failures during a protein design campaign?

To recover from failures during a protein design campaign, the orchestration layer provides robust debugging wrappers. These wrappers help manage standardized score merging and handle errors encountered during multi-step tool execution.