In-silico Gene Annotation Prediction Using the Co-expression Network Structure

Abstract

Identifying which genes are involved in particular biological processes is relevant to understand the structure and function of a genome. A number of techniques have been proposed that aim to annotate genes, i.e., identify unknown biological associations between biological processes and genes. The ultimate goal of these techniques is to narrow down the search for promising candidates to carry out further studies through in-vivo experiments. This paper presents an approach for the in-silico prediction of functional gene annotations. It uses existing knowledge body of gene annotations of a given genome and the topological properties of its gene co-expression network, to train a supervised machine learning model that is designed to discover unknown annotations. The approach is applied to Oryza Sativa Japonica (a variety of rice). Our results show that the topological properties help in obtaining a more precise prediction for annotating genes.