Prokaryote phylogeny without sequence alignment: from avoidance signature to composition distance

A new and essentially simple method to reconstruct prokaryotic phylogenetic trees from their complete genome data without using sequence alignment is proposed. It is based on the appearance frequency of oligopeptides of a fixed length (up to K=6) in their proteomes. This is a method without fine adjustment and choice of genes. It can incorporate the effect of lateral gene transfer to some extent and leads to results comparable with the bacteriologists' systematics as reflected in the latest 2001 edition of the Sergey's manual of systematic bacteriology. A key point in our approach is subtraction of a random back-groundby using a Markovian model of order K-1 from the composition vectors to highlight the shaping role of natural selection.