Using Visualization to Illustrate Machine Learning Models for Genomic Data

Massive amounts of genomic data are created for the advent of Next Generation Sequencing technologies. Visualizing these complex genomic data requires not only simply plotting of data but should also invite a decision or a choice. Machine learning has the ability to make prediction and aid in decision-making. Machine learning and visualization are both effective ways to deal with big data but focus on different purposes. Machine learning applies statistical learning techniques to automatically identify patterns in data to make highly accurate predictions while visualization can leverage the human perceptual system to interpret and uncover hidden patterns in big data. Clinicians, experts and researchers intend to use both visualization and machine learning to analyze their complex genomic data, but it is a serious challenge for them to understand and trust machine learning models in the medical industry. This paper overcomes this problem by combining intelligent and interactive visualization with machine learning models. Our prototype not only visualizes the complex genomics data in a meaningful 3D similarity space, but also illustrates the machine learning models and the real-time prediction results. Interactions and connections between the machine learning model and the 3D scatter plot are also developed and illustrated.