protein-structure

name	protein-structure
description	Use when planning or staging protein-structure prediction or structure-design work in this workspace, especially when deciding between missing local predictors and the repos that are actually present.
allowed-tools	["Read","Write","Bash(python:*)","Bash(docker:*)"]
context	fork
agent	general-purpose

protein-structure

Workspace-level gateway for structure-oriented work. In this repo, the strongest local assets are the checked-out RFdiffusion and evo2 repositories plus Docker, not a ready-to-run AlphaFold/ColabFold desktop stack.

Quick Start

• Available local assets: /home/vimalinx/Projects/bio_studio/repositories/active/RFdiffusion, /home/vimalinx/Projects/bio_studio/repositories/active/evo2, /usr/bin/docker
• Fast availability check: command -v colabfold_batch pymol chimera chimerax foldseek mmseqs fpocket
• Current practical path: use repo-backed Docker workflows rather than assuming local GUI prediction tools exist

When To Use This Tool

• Deciding how to approach a protein-structure task in this workspace
• Routing between structure design (RFdiffusion) and sequence-model analysis (evo2)
• Auditing whether local structure-prediction tools are actually installed before planning around them
• Staging follow-on structure workflows once prerequisites are satisfied

Common Patterns

# Check which structure tools are really installed
command -v colabfold_batch pymol chimera chimerax foldseek mmseqs fpocket

# Use the local RFdiffusion repo for generative design
cd /home/vimalinx/Projects/bio_studio/repositories/active/RFdiffusion
./test_rfdiffusion.sh

# Inspect the local Evo 2 repo for sequence-model-based analysis
cd /home/vimalinx/Projects/bio_studio/repositories/active/evo2
python examples/run_evo2.py

Recommended Workflow

1. Verify the required predictor or viewer is actually present before promising a structure workflow.
2. If you need generative backbone or binder design, start from the local RFdiffusion repo.
3. If you need sequence-model scoring or embeddings on DNA-scale inputs, evaluate whether Evo 2 is the right supporting asset.
4. Install missing prediction/visualization tools explicitly before planning around ColabFold, PyMOL, or Chimera-based steps.

Guardrails

• colabfold_batch, pymol, chimera, chimerax, foldseek, mmseqs, and fpocket are not currently on PATH in this workspace.
• RFdiffusion is present locally, but its Docker image is not prebuilt yet.
• The local Evo 2 repo exists, but direct import evo2 currently fails because vortex is missing.
• Do not present RFdiffusion as a native fold-prediction replacement; it is primarily a generative structure-design workflow.