A probabilistic framework for inferring ancestral genomic orders

Abstract

We introduce a probabilistic framework for inferring contiguous ancestral regions. Our previous work, the inferCARs algorithm, is a method based on adjacencies between synteny blocks. However, the local parsimony procedure has the limitation that it ignores many adjacencies that are potentially possible to exist in the ancestors. In this paper, we introduce a probabilistic method for reconstructing ancestral orders. The essential part of this method is to predict the posterior probability of an adjacency occurring in the ancestor based on an extended Jukes-Cantor model for breakpoints. We implemented a program called inferCARsPro to reconstruct contiguous ancestral regions. Both simulation and real data application results are discussed.