A new algorithm for cluster analysis of genomic methylation: the Helicobacter pylori case.

Abstract

The genomic methylation analysis is useful to type bacteria that have a high number of expressed type II methyltransferases. Methyltransferases are usually committed to Restriction and Modification (R-M) systems, in which the restriction endonuclease imposes high pressure on the expression of the cognate methyltransferase that hinder R-M system loss. Conventional cluster methods do not reflect this tendency. An algorithm was developed for dendrogram construction reflecting the propensity for conservation of R-M Type II systems. The new algorithm was applied to 52 Helicobacter pylori strains from different geographical regions and compared with conventional clustering methods. The algorithm works by first grouping strains that share a common minimum set of R-M systems and gradually adds strains according to the number of the R-M systems acquired. Dendrograms revealed a cluster of African strains, which suggest that R-M systems are present in H.pylori genome since its human host migrates from Africa. The software files are available at http://www.ff.ul.pt/paginas/jvitor/Bioinformatics/MCRM_algorithm.zip.