Protein Folding Workspace

This project contains the setup and maintenance scripts for a high-performance, CUDA-accelerated protein folding environment using GROMACS and NCBI tools.

Prerequisites

• UNIX distro (tested on Ubuntu 25.10)
• NVIDIA GPU (RTX/CUDA capable)
• NVIDIA Drivers (580+)

Notice

I use wildcards (*) in the README because that is the most realistic use case when running repeated experiments with numbered scripts. These scripts use official sources and forego external package-managers, but it is always good practise to check scripts for safety before running.

Quick Start

Make scripts executable

chmod +x *.sh

1. Install Dependencies

Installs nvidia-cuda-toolkit, ncbi-entrez-direct, gnuplot, and build tools.

./01*

2. Install GROMACS (Source Compile)

Clones the official repository, configures for CUDA, and installs locally to ./install.

./02*

3. Maintenance (Optional)

Run the maintenance script to verify your driver, CUDA, and GROMACS versions are correctly detected.

./03*

4. Test the Pipeline

Run the automated test script to fold Lysozyme (1AKI) and verify CUDA acceleration.

./04

5. Visualize Results

Generate a graph of the potential energy minimization and launch the 3D web viewer.

./05*

https://xaoscience.github.io/folding/test_lysozyme.html

6 + 7. Variant Discovery Pipeline

Generate and screen novel protein variants (fragments, mutants, or clones).

Step 1: Generate Variants Creates a batch of PDB files.

• Default: Appends to the latest experiment folder for that protein (e.g., experiments/1AKI/2025.11.30_...).
• --new: Forces a new timestamped experiment folder.

./06*
#or
./06* --new

Step 2: Screen Variants Runs the GROMACS pipeline on the generated variants and ranks them by potential energy.

• Auto-detects the latest experiment if no argument is provided.
• Skips already processed variants (incremental screening).

./07* [experiment_dir]