VortexDock: Distributed Molecular Docking System

VortexDock is a Python-based distributed molecular docking system consisting of a distribution server, a monitoring server, and compute nodes. It supports parallel computation across multiple nodes, utilizing idle resources to improve virtual screening efficiency. Currently, it supports AutoDock Vina, with plans to add more docking software.

Architecture

Distribution Server

Manages the task database, distributes receptor, ligand, and docking parameters, and collects docking results (ligand_out.pdbqt) from compute nodes.

Monitoring Server

Monitors the task database and visualizes task execution status.

Compute Node

Fetches tasks from the distribution server, performs docking, and submits results back to the server.

Installation

1. Clone the repository.
2. Install MySQL (current must, but we'll support SQLite soon)
3. Install dependencies:

   pip install -r requirements.txt
   

Distribution Server Setup

1. Navigate to the distribution_server directory.
2. Edit config.py:
   • Database connection
   • Server port
   • Other parameters
3. Initialize the database:

   python cli.py
   

Compute Node Setup

1. Navigate to the compute_node directory.
2. Edit config.py:
   • Server connection
   • Task parameters
   • System settings
3. Ensure the vina program has execution permissions.

Usage

Start Distribution Server

cd distribution_server
python server.py

Task Management

# List all tasks
python cli.py -ls

# Create a new task
python cli.py -zip <task_file.zip> -name <task_name>

task_name.zip
├── receptor.pdbqt
├── parameter.txt
│   ├── center_x = 233.7
│   ├── center_y = 155.8
│   ├── center_z = 127.7
│   ├── size_x = 35.7
│   ├── size_y = 41.2
│   ├── size_z = 45.2
│   ├── num_modes = 9
│   ├── energy_range = 9
│   └── cpu = 8
└── ligands
    ├── 001.pdbqt
    ├── 002.pdbqt
    ├── 003.pdbqt
    └── ...

# Pause/Resume a task
python cli.py -pause <task_id>

# Delete a task
python cli.py -rm <task_id>

Start Monitoring Server

Python-based monitoring server:

cd monitor_server
python app.py  # Default: localhost:9000

PHP-based monitoring server: Set the web root to monitor_server_PHP.

Start Compute Node

cd compute_node
python client.py

Features

• Distributed computation with multi-node support
• Task resumption after interruption
• Node heartbeat detection
• Automatic retry for failed tasks
• Temporary file management