Section: New Results

Algorithms for genome rearrangements

Participants : David James Sherman, Macha Nikolski [ correspondant ] , Géraldine Jean.

We developed an improved algorithm, SyDIG, for identifying synteny in distant genomes. It is designed for widespread cases where existing methods, such as filtered genome alignments (e.g. GRIMM-Synteny ), or profile-based iterated search (e.g. i-AdHoRe ), do not work. This in turn has led to improvements in our method for identifying super-blocks of syntenic segments[14] , improving on and building a bridge between competing methods defined by Sankoff and by Bourque and Pevzner. Super-blocks represent the semantics of the ancestral architecture, and provide a piecewise approximation to this architecture that provides a reasonable upper bound on the sum of rearrangement distances between contemporary genomes and the theoretical median. Super-blocks have been successfully identified for a range of species in the Hemiascomycetous yeasts[18] .

Using a new formulation in terms of optimization, we devised a new algorithm, FAUCILS, using techniques from optimization by local search and metaheuristics [40] . The algorithm maintains a population of configurations, modified depending on the set of architectures, and evaluated using the rearrangement distance. The result is a robust approach that converges rapidly, and obtains better results that those reported elsewhere. Compared with competing algorithms currently used, this new algorithm takes only a few minutes, compared to several hours; does so on tens of genomes, compared to a maximum of three; and includes biological constraints such as centromere presence and gene super-block conservation, which competing algorithms do not. A follow-up to FAUCILS uses any colony swarming to identify pairwise rearrangement scenarios[23] .