Abstract
The Next Generation Sequencing (NGS) technology substantially increases the number of DNA sequences of various species. In the DNA sequences of bacteria, scientists especially want to know whether foreign genetic materials integrate with, which promote the adaptation of bacteria and may even make superbugs happen. So when we annotate, we also want to locate foreign genetic materials. Foreign genes often form a cluster called genomic island (GI). GI is the evidence of horizontal origins and its sequence composition different from the host. The methods for detecting GIs are mainly divided two: first one is based on the sequence compositional variations; the second is that searching for the different regions in closely related species by comparative genomic. We obtained reliable GIs from IslandPick which detected by comparative genomics, and then analyzed several composition indices of the GIs to train a good classifier that could predict the unknown sequence. And we combined other features of GIs to design a pipeline for GIs detecting. We integrated above GI prediction pipeline into our platform. This platform additionally included genome annotation. Furthermore, ongoing genome can be predicted and annotated by our platform, so scientists will understand the information of the species in advance. We compared the results of prediction between finished genome and ongoing genome, and then we found the accuracies of ongoing genome were similar to finished genome that means ongoing genome can also obtain high accuracy. The proposed approach helps analyze the evolution of bacteria and the study of resistance and pathogenicity.