The goal of this project is to develop heterogenous parallel program to accelerate molecular docking simulation. Some features include;

• AutoDock 4 implementation for semiempirical energy function.
• Historical genetic algorithm for conformational search.
• Python implementation using OpenCL as the accelerator.

Documents:

• NTU Dissertation
• NTU Dissertation Slides
• Paper for IEEE Life Sciences Grand Challenges Conference 2013 in Singapore
• Poster for IEEE Life Sciences Grand Challenges Conference 2013 in Singapore

Python-OpenCL Implementation

Tested on:

• Python 2.7.3
• NumPy 1.7.1
• PyOpenCL 2013.1

To run:

• Go to PyNeuroDock directory
• Execute python NeuroDock.py

Benchmark:

• On Unix shell (Bash), PyNeuroDock directory, execute: ./Benchmark.sh
• Python: Sequential processing run on 2.3GHz Intel Core i7 (1 thread)
• Python-OpenCL GPU: Parallel processing run on NVIDIA GeForce GT 650M 1GB (384 CUDA cores)
• Python-OpenCL CPU: Parallel processing run on 2.3GHz Intel Core i7 (4 cores, 8 threads)