摘要：Summary Graphics processing unit (GPU) accelerated computing has pioneered a new direction of research for various combinatorial optimization problems. One such problem which requires huge computation is protein structure prediction (PSP). PSP is NP-complete problem. Computational prediction of protein native structure from its primary amino acid sequence is termed as ab initio PSP problem. Till date, wet lab experiments conducted on PSP indicate that existing methods take lots of experimentation time and expensive. As a consequence, only 1% of the sequence's structures are known. This work presents a parallel programming approach with GPU computing for PSP using 2D triangular hydrophobic-polar (HP) lattice model. The implementation of proposed approach is tested on the set of HP benchmark sequence of a length ranging from 25 to 100. The experimental result shows that the proposed approach has significantly improved the performance of prediction with immense drop in computation time.
