PIPS-Design: Phosphorylation Switch Design

A computational pipeline for designing phosphorylation switches using structural bioinformatics tools.

Prerequisites

1. Install STRIDE

STRIDE is used for secondary structure identification.

# Download from: https://webclu.bio.wzw.tum.de/stride/
# Then modify the stride path in preprocess.py

2. Install ProteinMPNN

git clone https://github.com/dauparas/ProteinMPNN.git
# Modify the main_path in design.py: 'main_path': '/path/to/your/ProteinMPNN'

3. Install ColabFold

git clone https://github.com/sokrypton/ColabFold.git

4. Setup Conda Environment

Create and configure the environment (same as ProteinMPNN requirements):

conda create --name mlfold  # creates conda environment called mlfold
conda activate mlfold      # activate environment

# Install PyTorch (adjust CUDA version based on your system)
conda install pytorch torchvision torchaudio cudatoolkit=12.4 -c pytorch

Usage

Step 1: Preprocessing

python ./preprocess.py \
    --input ./input_pdbs \
    --output ./example_output \
    --stride_path /path/to/your/stride

Step 2: Design and Prediction

python ./design.py ./example_output/rename.csv

Step 3: Filtering and Analysis

python ./filt_analysis.py

Pipeline Overview

1. Preprocessing: Identifies secondary structures using STRIDE
2. Design: Uses ProteinMPNN for protein design
3. Prediction: Utilizes ColabFold for structure prediction
4. Analysis: Filters and analyzes the results

File Structure

├── preprocess.py          # Preprocessing script
├── design.py             # Design and prediction script
├── filt_analysis.py      # Filtering and analysis script
├── input_pdbs/           # Input PDB files directory
└── example_output/       # Output directory

Configuration

Remember to update the following paths in the respective scripts:

• stride_path in preprocess.py
• main_path (ProteinMPNN location) in design.py