Perturbative formulation of general continuous-time Markov model of sequence evolution via insertions/deletions, Part III: Algorithm for first approximation

Abstract

BackgroundInsertions and deletions (indels) account for more nucleotide differences between two related DNA sequences than substitutions do, and thus it is imperative to develop a stochastic evolutionary model that enables us to reliably calculate the probability of the sequence evolution through indel processes. In a separate paper (Ezawa, Graur and Landan 2015a), we established an ab initio perturbative formulation of a continuous-time Markov model of the evolution of an entire sequence via insertions and deletions. And we showed that, under a certain set of conditions, the ab initio probability of an alignment can be factorized into the product of an overall factor and contributions from regions (or local alignments) separated by gapless columns. Moreover, in another separate paper (Ezawa, Graur and Landan 2015b), we performed concrete perturbation analyses on all types of local pairwise alignments (PWAs) and some typical types of local multiple sequence alignments (MSAs). The analyses indicated that even the fewest-indel terms alone can quite accurately approximate the probabilities of local alignments, as long as the segments and the branches in the tree are of modest lengths.\n\nResultsTo examine whether or not the fewest-indel terms alone can well approximate the alignment probabilities of more general types of local MSAs as well, and as a first step toward the automatic application of our ab initio perturbative formulation, we developed an algorithm that calculates the first approximation of the probability of a given MSA under a given parameter setting including a phylogenetic tree. The algorithm first chops the MSA into gapped and gapless segments, second enumerates all parsimonious indel histories potentially responsible for each gapped segment, and finally calculates their contributions to the MSA probability. We performed validation analyses using more than ten million local MSAs. The results indicated that even the first approximation can quite accurately estimate the probability of each local MSA, as long as the gaps and tree branches are at most moderately long.\n\nConclusionsThe newly developed algorithm, called LOLIPOG, brought our ab initio perturbation formulation at least one step closer to a practically useful method to quite accurately calculate the probability of a MSA under a given biologically realistic parameter setting.\n\n[This paper and three other papers (Ezawa, Graur and Landan 2015a,b,c) describe a series of our efforts to develop, apply, and extend the ab initio perturbative formulation of a general continuous-time Markov model of indels.]