GPU Accelerated Protein Docking Software with Flexible Refinement

DESCRIPTION (provided by applicant): Protein-protein interactions are involved at multiple points in virtually all biological pathways. Understanding such interactions is also important for the design of biologics that can target extracellular receptors with high affinity and specificity. Since determining the structure of protein complexes by X-ray crystallography is expensive and slow, it is important to develop computational docking methods that, starting from the structures of component proteins or homology models, can determine the structure of their complexes. Accordingly, there is increasing demand for protein docking methods in the pharmaceutical industry. Based on the results of CAPRI (Critical Assessment of Predicted Interactions), a worldwide protein docking competition, PIPER, developed at Boston University and licensed to Acpharis, is the best protein-protein docking program currently available. A major problem is that the flexible refinement of the PIPER-generated structures requires computational resources that are generally not available in industry. The general goal of this proposal is to develop efficient flexible refinement methods, and to implement the computationally expensive steps on GPUs. The refinement will employ two novel algorithms. First, given a putative interface defined by a cluster, Acpharis will develop a program to identify the key variable side chains in the interface and their potential conformational states. Second, we will develop a Monte Carlo minimization algorithmfor flexible refinement, which combines search in the space of the selected side chain rotamers with an innovative minimization method in the rotational/translational space based on manifold concepts. A number of the resulting structures will be subjectedto further refinement involving backbone relaxation. In addition to the use of more powerful flexible refinement algorithms, further speed-up will be achieved by implementing the time consuming components of both docking and refinement on GPUs. Profiling the algorithms we have found two such components, namely (1) correlation calculations that use fast Fourier transforms (FFTs) in docking, and (2) the non-bonded energy evaluation in the flexible refinement step. For the docking step we will perform rotationand grid assignment on the CPU while the FFT and filtering will be computed on the GPU. Acceleration of the energy evaluation steps will require changing the underlying data structures and statically mapping the work onto GPU threads in a way that allowsparallel energy evaluations. With the above algorithmic and architectural speed-up, we can expect that a docking and refinement problem that previously required several hours on a 128 CPU cluster will be solved in the same amount of time by a single CPU and 2 NVIDIA Fermi GPU cards. Such a system can currently be assembled for 3500, which is clearly within reach for small pharmaceutical start-up companies or computational chemistry units. PUBLIC HEALTH RELEVANCE: Understanding protein-protein interactions is crucial for discovery of certain drugs and biologics. The goal of this proposal is obtaining the information by novel computational methods implemented on cost effective graphic processing units. (GPUs).