Detection of Mosaic Structure of Syntenic Segments in Marine Cyanobacterial Genomes

In case of closely related bacterial genomes, segments of conserved gene order that referred to synteny blocks are observed by mapping orthologous genes (or simply called orthologs) to each genome. In many cases, they share the stable structure consisting of synteny blocks, which are often called ‘genomic core’. This structure is a marker for the structural evolution in these genomes. They also possess variable genomic islands consisting of laterally transferred genes [2]. Bacterial genomes are, therefore, envisaged as a mosaic of genomic core interspersed with genomic islands. Such mosaic structure of the bacterial genome has been well studied in Escherichia coil [3] and cyanobacterial genomes among others. To find synteny blocks, we need a multiple genome alignment in advance. Existing software to find them are based on the logic of graph search originally proposed by Sankoff [4], but alignment quality of the results by these algorithms depends on gap penalty at the process of alignment reduction. In spite of the efforts to introduce many heuristics, It is still a problem because no systematic method of estimating gap penalties for particular genomes is known. Even the case of globally optimized alignments by maximum matching approach, local optimization is not guaranteed [1]. In these algorithms, optimization strategy is combinatorial. Now we propose a statistical approach to detect synteny blocks using multidimensional scaling and clustering techniques and show an application result for marine cyanobacteria.