DCJ Median Problems on Linear Multichromosomal Genomes: Graph Representation and Fast Exact Solutions

  • Authors:
  • Andrew Wei Xu

  • Affiliations:
  • School of Computer and Communication Sciences, Swiss Federal Institute of Technology (EPFL), EPFL IC LCBB, Station 14, Lausanne, Switzerland CH-1015

  • Venue:
  • RECOMB-CG '09 Proceedings of the International Workshop on Comparative Genomics
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Given a set of genomes $\mathcal{G}$ and a distance measure d , the genome rearrangement median problem asks for another genome q that minimizes $\sum_{g\in \mathcal{G}} d(q,g)$. This problem lies at the heart of phylogenetic reconstruction from rearrangement data, where solutions to the median problems are iteratively used to update genome assignments to internal nodes for a given tree. The median problem for reversal distance and DCJ distance is known to be NP-hard, regardless of whether genomes contain circular chromosomes or linear chromosomes and of whether extra circular chromosomes is allowed in the median genomes. In this paper, we study the relaxed DCJ median problem on linear multichromosomal genomes where the median genomes may contain extra circular chromosomes; extend our prior results on circular genomes--which allowed us to compute exact medians for genomes of up to 1,000 genes within a few minutes. First we model the DCJ median problem on linear multichromosomal genomes by a capped multiple breakpoint graph , a model that avoids another computationally difficult problem--a multi-way capping problem for linear genomes, then establish its corresponding decomposition theory, and finally show its results on genomes with up to several thousand genes.