Bioinformatic methods of gene function prediction fall into 3 classes:(1) homology;(2) non-homology;(3) structure.Non-homology based approaches include:(1) domain fusion,(2) phylogenetic profile,(3) conserved gene position and(4) related gene expression.Generally,the first three methods are collectively referred as genomic context.The replication of bacterial genomes initiates at a specific region,named origin of replication(ori).Up to now,only a small number of complete sequenced bacterial genomes have been annotated experimentally with their ori.The main tools for ori prediction are DNA walk, GC-skew and Z-curve.DNA walk and GC-skew are not efficient for all genomes;while Z-curve need to utilize additional biological features for a final prediction.In this paper,an ori prediction method,called Familiar-Mark,was proposed.The method used a combination of homology and genomic context.Without using more biological features, the method preformed very well among 4 well-annotated Escherichia coli genomes.When applying the method to more E.coli genomes,Familiar-Mark got the same results with Z-curve.Familiar-Mark has good flexibility.It could be of a helpful tool for comparative genomic researches.
|