Efficient computation of Hash Hirschberg protein alignment utilizing hyper threading multi-core sharing technology

CAAI Transactions on Intelligence Technology

article

Muhannad Abu-Hashem¹  Adnan Gutub² 

[1] Department of Geomatics, Faculty of Architecture and Planning, King Abdulaziz University;Department of Computer Engineering, College of Computer & Information Systems, Umm Al-Qura University

关键词: computational biology;    high-performance computing;    Hyper Threading;    pairwise sequence alignment;    parallel design;    sequence alignment;    shared-memory;   

DOI  :  10.1049/cit2.12070

学科分类：数学（综合）

来源: Wiley

【 摘 要 】

Due to current technology enhancement, molecular databases have exponentially grown requesting faster efficient methods that can handle these amounts of huge data. Therefore, Multi-processing CPUs technology can be used including physical and logical processors (Hyper Threading) to significantly increase the performance of computations. Accordingly, sequence comparison and pairwise alignment were both found contributing significantly in calculating the resemblance between sequences for constructing optimal alignments. This research used the Hash Table-NGram-Hirschberg (HT-NGH) algorithm to represent this pairwise alignment utilizing hashing capabilities. The authors propose using parallel shared memory architecture via Hyper Threading to improve the performance of molecular dataset protein pairwise alignment. The proposed parallel hyper threading method targeted the transformation of the HT-NGH on the datasets decomposition for sequence level efficient utilization within the processing units, that is, reducing idle processing unit situations. The authors combined hyper threading within the multicore architecture processing on shared memory utilization remarking performance of 24.8% average speed up to 34.4% as the highest boosting rate. The benefit of this work improvement is shown preserving acceptable accuracy, that is, reaching 2.08, 2.88, and 3.87 boost-up as well as the efficiency of 1.04, 0.96, and 0.97, using 2, 3, and 4 cores, respectively, as attractive remarkable results.

【 授权许可】

CC BY|CC BY-ND|CC BY-NC|CC BY-NC-ND