Exploring Epigenetic Drug Discovery Using Computational Approaches

Abstract

The misregulation of epigenetic mechanisms has been linked to disease. Current drugs that treat these dysfunctions have had some success, however many have variable potency, instability in vivo and lack target specificity. This may be due to the limited knowledge of epigenetic mechanisms, especially at the molecular level, and their association with gene expression and its link to disease. Computational approaches, specifically in molecular modeling, have begun to address these issues by complementing phases of drug discovery and development, however, more research is needed on the relationship between genetic mutation and epigenetics and their roles in disease. Gene regulatory network models have been used to better understand diseases, however inferring these networks poses several challenges. To address some of these issues, a multi-label classification technique to infer regulatory networks (MInR), supplemented by semi-supervised, learning is presented. MInR's performance was found to be comparable to other methods that infer regulatory networks when evaluated on a benchmark E. coli dataset. In order to better understand the association of epigenetics with gene expression and its link with disease, MInR was used to infer a regulatory network from a Kidney Renal Clear Cell Carcinoma (KIRC) dataset and was supplemented with gene expression and methylation analysis. Gene expression and methylation analysis revealed a correlation between 5 differentially methylated CpGs and their matched differentially expressed transcripts. Incorporating this information into a network allowed for the visualization of potential systems that may be involved in KIRC. Future analysis of these systems alongside the drug discovery and development process may lead to the discovery of novel therapeutics.