Modeling of Biomolecular Systems Interactions, Dynamics, and Allostery: Bridging Experiments and Computations - September 10-14, 2014, Istanbul, Turkey

Modeling of Biomolecular Systems Interactions, Dynamics, and Allostery Poster Session II

67-POS Board 20 The MEGADOCK Project: High-Performance Protein-Protein Interaction Prediction Tools on Supercomputing Environments Masahito Ohue 1,2 , Takehiro Shimoda 1 , Yuri Matsuzaki 3 , Nobuyuki Uchikoga 4 , Takashi Ishida 1 , Yutaka Akiyama 1,2 . 1 Tokyo Institute of Technology, Tokyo, Japan, 3 Tokyo Institute of Technology, Tokyo, Japan, 2 Japan Society for the Promotion Science, Tokyo, Japan, 4 Chuo University, Tokyo, Japan. Background Protein-protein interaction (PPI) plays a core role in cellular functions. In recent years, PPI prediction methods based on protein docking have been developed and have been applied for large-scale PPI network prediction based on tertiary structures. However, such network prediction requires much computing resources, and a faster PPI prediction method is eagerly demanded. Results We have developed a high throughput PPI prediction system based on rigid-body protein docking, “MEGADOCK”. MEGADOCK can perform faster docking based on its original scoring function. Recently, MEGADOCK has been accelerated by using the general purpose graphics processing unit (GPGPU) technique and it is now released as MEGADOCK-GPU. MEGADOCK was also parallelized for massively parallel supercomputing environment using the hybrid parallelization (MPI/OpenMP) technique. The system, named MEGADOCK-K, achieved an excellent scalability on supercomputing environments, such as K computer, which has 705,024 Fujitsu SPARC64 VIIIfx CPU cores. We have already applied MEGADOCK system to a number of interactome analyses such as bacterial chemotaxis pathway, human apoptosis pathway and human epidermal growth factor receptor related pathway. Conclusion We present a new protein-protein docking engine aimed at exhaustive docking of millions of protein pairs. The system was shown to be scalable when running on thousands of nodes and multiple GPUs.

120