Research On An Algorithm Of Protein Structural Alignment Optimization With CUDA

Bioinformatics is the use of mathematics, statistics, informatics and computerscience to research on biological problems. Bioinformatics research an important criteriais that structure determines function. Similar structures tend to have similar functions.The current number of known structure of the protein is much less than the sequencenumber of proteins and proteins of known structure and function of the less. Thus, by thecalculation method to determine a similar structure pushed to the functional biologystudies.TM-align protein structure is an effective comparison tool, through rotation andtranslation, to find the largest overlapped portion of the TM-score. The algorithminvolves a large number of TM-score calculations. CUDA technology is a unified thehigh concurrent computing architecture, the TM-score can effectively solvecomputational problems.The research for this article include:(1) study the protein structure that usetraditional serial than traditional algorithm implementations, CPU structure alignmentalgorithm;(2) to study the global multiple protein structures using CUDA accelerationalgorithm, including the method of separation of unified computing and storagealgorithm;(3) study the CUDA and CPU combined calculation method, which makes theacceleration of protein structure may become;(3) by CUDA Toolkit debugging methodsto optimize storage and algorithms.Optimization by the CPU and CUDA comparison of the results it can be concludedin the same computational efficiency in parallel certain size, or larger than the length ofthe protein reaches a certain threshold, the efficiency of the GPU can be larger than theCPU.