Reconstruction of Maximum Likelihood Phylogenetic Trees in Parallel Environment Using Logic Programming

With rapid increase of nucleotide and amino acid sequence data, it is required to develop reliable and exible application programs to infer molecular phylogenetic trees. The maximum likelihood method is known to be robust among many methods for reconstruction of molecular phylogenetic trees, however, this method requires extremely high computational cost. Although parallel computation is a good solution to realize complicated inference such as the maximum likelihood method, generally speaking, it is not so easy to implement parallel programs. In actual data analyses, furthermore, it is often needed to modify or expand application programs. In other words, there is no perfect application program for all data analyses. Logic programming is a good candidate to implement such data analysis programs in natural science elds, because programs in logic programming are easy to write, easy to modify, and easy to implement in parallel environment. We thus have developed an experimental system for reconstruction of phylogenetic trees in parallel environment in logic programming as a part of molecular evolutionary analysis system DeepForest. We propose the core algorithm for parallel execution of the maximum likelihood and show its veri cation according to simulation using amino acid data.