GPU Accelerated Maximum Likelihood Analysis for Phylogenetic Inference

Abstract

With the advancement of biology and computer science, the amount of DNA sequences has grown at a rapid rate giving rise to the analysis of phylogenetic trees with many taxa. The maximum likelihood analysis is commonly considered as the best approach in phylogenetic analyses, which is extremely intensive for computation. Availability of computer resources and the application of modern technologies are key factors that determine the use of such analyses. The paper presents a parallel implementation of a GPU accelerated maximum likelihood inference of phylogenetic trees on DNAml program of the PHYLIP package. The improved DNAml program uses both GPU and CPU processing to perform compute-intensive tasks in phylogenetic analyses. The evaluation results show a speedup of x2.94 for the GPU accelerated DNAml program than the existing program. As the results show the proposed system saves the processing time increasingly against the current system with the number of taxa.