A sensitive, support-vector-machine method for the detection of horizontal gene transfers in viral, archaeal and bacterial genomes.

摘要: In earlier work, we introduced and discussed a generalized computational framework for identifying horizontal transfers. This relied on gene's nucleotide composition, obviated the need knowledge of codon boundaries database searches, was shown to perform very well across wide range archaeal bacterial genomes when compared with previously published approaches, such as Codon Adaptation Index C + G content. Nonetheless, two considerations remained outstanding: wanted further increase sensitivity detecting transfers also be able apply method increasingly smaller genomes. discussion that follows, present method, Wn-SVM, show it exhibits significant improvement in approaches. Wn-SVM uses one-class support-vector machine can learn using rather small training sets. property makes particularly suitable studying small-size genomes, similar those viruses, typically larger We experimentally new results superior performance organisms improves even upon our own by an average 10% all examined As small-genome case study, analyze genome human cytomegalovirus demonstrate correctly identifies regions are known conserved prototypical beta-herpesvirinae, have been acquired horizontally from host and, finally, had not up now suspected transferred. Atypical region predictions many eukaryotic including α-, β- γ-herpesvirinae, 123 made available online at http://cbcsrv.watson.ibm.com/HGT_SVM/.