Skip to content

GETTING_STARTED.md

Latest commit

 

History

History
118 lines (86 loc) · 3.27 KB

GETTING_STARTED.md

File metadata and controls

118 lines (86 loc) · 3.27 KB
Raw

Getting Started with PROTACXtend

This guide will walk you through setting up PROTACXtend on your system, installing required dependencies, and executing your first AI-guided PROTAC design workflow.

📋 Prerequisites

  • OS: Linux (Ubuntu 20.04+ recommended) or macOS
  • Python: 3.9, 3.10, or 3.11
  • Conda / Mamba: Recommended for RDKit and PyTorch package isolation
  • Optional: Docker / Singularity (for full P4ward ternary complex 3D docking runs)

⚡ Quick Installation

Option 1: Install from GitHub (Recommended)

Clone the canonical repository (Ahuja Lab organization home):

# Clone the repository
git clone https://github.com/the-ahuja-lab/PROTACXtend.git
cd PROTACXtend

# Create a virtual environment
conda create -n protacxtend python=3.10 -y
conda activate protacxtend

# Install key chemistry & machine learning dependencies
pip install -r requirements.txt

# Install PROTACXtend in editable mode (PyPI publishing is on the roadmap)
pip install -e .

Option 1b: Docker

docker build -t protacxtend https://github.com/the-ahuja-lab/PROTACXtend.git

Option 2: Repository-Local CLI Execution (No Installation Required)

If you are working inside the workspace directory, you can run the executable wrapper immediately:

cd /storage/saveena/protacpilot
./PROTACXtend --help
# Or lowercase alias
./protacxtend status

🧪 Verifying the Environment

Run the self-diagnostic check to ensure all chemistry modules, RDKit sanitization routines, and agent tools are correctly initialized:

protacxtend status

Example output:

PROTACXtend System Diagnostic [v0.3.0]
--------------------------------------
Python Version   : 3.10.12
RDKit Version    : 2023.09.1 [OK]
Chemprop Engine  : Active [OK]
P4ward Docker    : Ready [OK]
ChEMBL Lookup    : Connected [OK]
Status           : All 23 workflow nodes operational.

Test molecular sanitization with a sample SMILES:

protacxtend validate --smiles "O=C1NC(=O)C(N2C(=O)c3ccccc3C2=O)CC1"

🚀 Running Your First Workflow

1. Interactive CLI Command

Design PROTACs targeting BRD4 with CRBN E3 ligase:

protacxtend design "Design CRBN PROTAC candidates targeting BRD4 degradation"

2. Launching the Local Workbench UI

Launch the interactive science workbench interface:

protacxtend serve

Open your web browser at http://localhost:8501.

3. REST API Server Mode

Start the FastAPI REST backend for programmatic API calls:

python -m uvicorn protacxtend.backend.api_routes:get_app --factory --host 0.0.0.0 --port 8001

Send a POST request to design PROTAC candidates:

curl -X POST "http://localhost:8001/design" \
  -H "Content-Type: application/json" \
  -d '{"request": "Design 16 PROTACs for HMGB2 with ICM warhead and CRBN E3"}'

📂 Next Steps

  • Explore the complete Architecture Guide to understand the 23-node agentic workflow graph.
  • Learn about slash commands in Workflows & Slash Commands.
  • Set up your GitHub collaboration environment in GitHub Setup.