Algebraic Distribution of Segmental Duplication Lengths in Whole-Genome Sequence Self-Alignments

Abstract

Distributions of duplicated sequences from genome self-alignment are characterized, including forward and backward alignments in bacteria and eukaryotes. A Markovian process without auto-correlation should generate an exponential distribution expected from local effects of point mutation and selection on localised function; however, the observed distributions show substantial deviation from exponential form--they are roughly algebraic instead--suggesting a novel kind of long-distance correlation that must be non-local in origin.