A New Compositional-feature Method For The Detection Of Horizontal Gene Transfer

Horizontal gene transfer(HGT), also known as lateral gene transfer(LGT), refers to the exchange of genetic material between different individual organisms, i.e. from outside the individual’s own genetic material obtained in the process. Different individual organisms can be of the same organism but a different individual organisms containing genetic information, it can also be distant or even no genetic relationship. Horizontal gene transfer is proposed by relative to the vertical gene transfer(parental passed to offspring), current knowledge indicates that horizontal gene transfer is an important phenomenon, which breaks the boundaries of kinship, so that made genes flow is possible.Most computational methods for the distinguish of horizontal gene transfer have been proposed, the majority of them used parametric method detect horizontally transferred genes in bacterial genomes, and usually select single feature or simply combine several features for the detecting of single bacterial genome or artificial genomes. As we all know, different features characterizing different gene sequences encoding information, different gene sequences can be expressed by different features, therefore, different gene sequences maybe expressed by different optimal feature or combination of features, and the combination of plurality features maybe to expression more accurately than single feature. Accordingly, the detection of horizontally transferred genes in several bacterial genomes by single feature or simply combine several features may affect the accuracy and reliability of prediction.To solve this problem, we propose a new compositional-feature method for detection of horizontal gene transfer. First, we select 17 features that be used frequently and have better performance, base on support vector machine model weighted array of them, using genetic algorithm to optimize combined features, improve the prediction accuracy. Second, we select 396 genuine bacterial genomes for detected, it can more comprehensive and reliable analysis the influence of different features in different bacterial genomes for gene functional expression that makes our results more convincing.In this paper, our combinational features method is the method for detecting of horizontal transferred genes that classification by SVM and using GA to optimization, also in-depth study of the parameter methods. By comparison with other prediction methods, our proposed combinational features method is more effective to improve the prediction accuracy and reduce the time complexity. For biology researchers, use of more effective and better prediction method to deal with practical problems, can save a lot of time and unnecessary expenses, such as the study of bacterial and human disease etc. Our method is through to handle large amounts of genome data and further observation and analysis, to provide a reference for the detecting of horizontal transferred genes in prokaryotes, to lay the foundation for biological research. In future studies, we will be more careful analysis of each type of bacterial genomes, to classification by functional expression of proteins, and predict horizontal transferred genes of protein coding in different functional expression.