Gene Essentiality Prediction Using Topological Features From Metabolic Networks

Author(s)

Nagai, James S. ; Sousa, Herio ; Aono, Alexandre H. ; Lorena, Ana C. ; Kuroshu, Reginaldo M. ; IEEE

• Document type: Journal article
• Source: 2018 7TH BRAZILIAN CONFERENCE ON INTELLIGENT SYSTEMS (BRACIS); v. N/A, p. 6-pg., 2018-01-01.

Abstract

Fundamental questions such as what are the genes that are really necessary for the survival of cells have motivated many studies to investigate the essentiality of genes in different species. Initial efforts have attempted to address this problem through exhaustive knockout experiments in simple bacteria. Recently, results obtained in these studies have also been applied to the emerging field of synthetic biology with possible implications in many other fields such as health and energy. Motivated by the evolution of DNA sequencing technology and high-throughput biological data generation, many recent efforts have also been made for building and understanding biological networks. In particular, metabolic networks represent the set of known biochemical reactions within a cell. Essential genes are expected to play a key role in these networks, as they must be involved in vital metabolic pathways. Even though some studies investigated the correlation between essential genes and biological network information, different types of networks and other biological information were usually combined and the effect of each of them in the obtained results was not stressed. This paper describes an attempt to predict essential genes using solely topological features from metabolic networks. The networks were built from a common repository, the KEGG database, ensuring data uniformity. Experimentally, considering different prediction scenarios and reference organisms, the use of topological features from metabolic networks achieved mean AUC of about 70% in the prediction of gene essentiality. This reveals that more factors affect essentiality and should indeed be considered in order to obtain more accurate predictions. (AU)

• FAPESP's process: 12/22608-8 - Use of data complexity measures in the support of supervised machine learning
• Grantee: Ana Carolina Lorena
• Support Opportunities: Research Grants - Young Investigators Grants