Visual Analysis for MicroRNA and mRNA Expression Profile Data

Abstract

MicroRNAs (miRNA) are short nucleotides that down-regulate its target genes. Various miRNA target prediction algorithms have used sequence complementarity between miRNA and its targets. Recently, other algorithms tried to improve sequence based miRNA target prediction by exploiting miRNA-mRNA expression profile data. Some web-based tools are also introduced to help researchers predict miRNAs targets from miRNA-mRNA expression profile data

However, there is still a demand for a miRNA-mRNA visual analysis tool that include quality miRNA prediction algorithms and more interactive visualizations. We presented two techniques for miRNA-mRNA interaction visualizations, Bipartite Treemap and enhanced node-link diagram. Bipartite Treemap is a new visualization technique for miRNA-mRNA interaction network that resolves occlusion problem. Enhanced node-link diagram provides interaction techniques that help users to explore miRNA-mRNA interaction network easily. We designed and implemented miRTarVis, which is an interactive visual analysis tool that predicts miRNA targets by integrating sequence based and miRNA-mRNA expression profile based miRNA target prediction algorithms, and visualizes the resulting miRNA-mRNA interaction network. miRTarVis has intuitive interface design in accordance with the analysis procedure of load, filter, predict, and visualize. It predicts miRNA targets by adopting Bayesian inference and MINE analyses, as well as conventional correlation and mutual information analyses. It visualizes a resulting miRNA-mRNA network in an interactive Bipartite Treemap as well as ehanced node-link diagram. Using miRTarVis, we analyzed miRNA-mRNA expression profile data from an experiment over asthmatic and non-asthmatic fibroblasts exposed to obese visceral exosomes. In addition, we applied miRTarVis to miRNA-mRNA expression profile data from breast cancer cell lines data to show its efficacy. miRTarVis verified its efficacy by helping its users execute miRNA target prediction easily and gain insights from miRNA-mRNA expression profile data by its interactive visualization.