Comptes rendus de l’Acade'mie bulgare des Sciences, Vol 69, No4, pp.489-496

Performance Evaluation of Parallel DNA Multigene Sequence Analysis

Luka Filipović, Danilo Mrdak, Božo Krstajić


Computational phylogeny and DNA sequence analysis are challenging even for the most powerful supercomputers. RAxML is the fastest and the most accurate software in the area of phylogenetic analysis, mostly used for sequential and parallel maximum likelihood based inference of large phylogenetic trees. In this paper we present scalability analysis of multigene DNA sequence analysis using RAxML on the high performance cluster. We analyzed five different genes, two real and additional three designed genes, in order to test reliability of the constructed census phylogeny tree. We have proved validity of parallelism using MPI for analyzed dataset with the best efficiency results during execution on up to 32 cores.

Key words: parallel and distributed computing, performance analysis, computational efficiency, DNA computing, bioinformatics