Identifying Positional Orthologs by Linear Programming

Studying the evolution of gene orders is essential to our understanding of the link between gene regulation and phenotype. Comparative studies concerning the full complement of genes in large numbers of genomes are now possible, thanks to the advent of affordable long-read sequencing and whole chromosome assembly, combined with automated genome annotation. The study of gene orders remains cumbersome, however, due to a lack of a streamlined, and standardized, process meant for the identification of positional orthologs . Within a family of homologous genes, these are those pairs of orthologous gene copies that descend from the same locus in their most recent ancestor, not having been produced by an intervening duplication event. This article approaches the detection of positional orthologs by way of the MAXIMUM COLORFUL GRAPH PARTITION Problem. We present two novel integer linear programming solutions to this problem, evaluate their efficacy on simulated and real data, while comparing them to several new and existing heuristics, as well as to the Double Cut and Join based ILP called FFGC .