Prediction of Horizontal Gene Transfer Using Machine Learning Techniques in Escherichia coli

Abstract

Horizontal Gene Transfer(HGT) also known as Lateral Gene Transfer is a process where an organism acquires genetic material from another organism without being a descendant of that organism. Until recently Horizontal gene transfer was not considered as important as vertical transfer; however now researchers have become aware of the important role played by HGT in the evolution of organisms. Horizontal gene transfer is said to be the predominant method of evolution in prokariyotic organisms and is identi ed as the most common method of bacteria acquiring drug resistance. This study is focused on constructing a method that employs genome comparison and semi supervised learning to identify genes that are horizontally transferred to Escherichia coli O157:H7 and attempting to nd a link between these genes and other organisms that display pathological behavior. E.coli O157:H7 is compared to E.coli K-12 which is a harmless strain of the same organism. This comparison yields the set of genes that has not originated from the same ancestor(non-homologous) and is the possible cause of its pathogenic properties. It has been identi ed in earlier works that horizontally transferred genes contain di erent compositional feature values than that of the host genome. With this in mind, a supervised self organizing map was constructed to classify the non-homologous genes as either horizontally or vertically transferred. Most of the obtained horizontally transferred genes have shown a striking similarity to other pathological bacteria and archaea. The results have indicated that, while it is possible to discern the mode of transfer of a gene based on compositional feature to a certain degree, it is better to combine several other features to further re ne the ndings.