Genome Divergence Based on Entropic Segmentation of DNA

Abstract

The concept of a genome signature broadly refers to characteristic patterns in DNA sequences that enable the identification and comparison of species or individuals, often without requiring sequence alignment. Such signatures have applications ranging from forensic identification of individuals to cancer genomics. In comparative genomics and evolutionary biology, genome signatures typically rely on statistical properties of DNA that are species-specific and carry phylogenetic information reflecting evolutionary relationships. We propose a novel genome signature based on the compositional structure of DNA, defined by the distributions of strong/weak, purine/pyrimidine, and keto/amino ratios across DNA segments identified through entropic segmentation. We observe that these ratio distributions are similar among closely related species but differ markedly between distant ones. To quantify these differences, we employ the Jensen–Shannon distance—a symmetric and robust measure of distributional dissimilarity—to define a genome-to-genome distance metric, termed Segment Compositional Distance (D). Our results demonstrate a clear correlation between D and species divergence times, and also that this metric captures a strong phylogenetic signal. Our method employs a genome-wide approach rather than tracking specific mutations; thus, D offers a coarse-grained perspective on genome compositional evolution, contributing to the ongoing discussion surrounding the molecular clock hypothesis.