Unlocking the complexity of genomic data of RMS patients through visual analytics

This paper presents a novel visual analytics technique that enables effective analysis of large and complex genomic and biomedical data. A comprehensive prototype has been developed to support the analysis process The system consists of multiple components, including an automated gene selection, a three-dimensional visualization for analyzing patient's relationship, and an interactive Heatmap visualization. These visualizations provide not only the meaningful and easy interpretable views to medical analysts, but also a user-centric adjustment in the analytical reasoning (feature selection) phase of the model through visual interaction. Therefore, the results of analytic reasoning can be adjusted accurately through human involvement. We demonstrate our techniques on a case study of a dataset of Rhabdomyosarcoma (RMS) patients which is the most common soft tissue childhood sarcoma. Two major histological subtypes of RMS are Alveolar (ARMS) and Embryonal (ERMS) with ERMS patients having a more positive prognosis. This study aims to discover genes from the gene expression microarray dataset that can differentiate between ERMS and ARMS patients.