Simultaneous Identification of Conserved Regions and Inversions through Global Optimization

We propose a new method for the comparative analysis of two long DNA sequences. Our method is the first that takes into account biological events like local inversions when searching for highly conserved regions. Our method introduces new ideas to the usual concept of alignments, that detects events like insertions, deletions and mutations but not inversions. It proposes a weighted bipartite-matching approach to account for inversions. There is a polynomial solution to this approach, but the obtained solution may produce biologically unrealistic number of inversions. If cost for inversions is added, no polynomical solution is known. We propose a model that supposes nonoverlapping inversions that leads to a polynomial algorithm. This method has been successfully applied to public genomic data availabel for Xylella fastidiosa and Pseudomonas aeruginosa.