Data-Driven Analytics for High-Throughput Biological Applications

High dimensional and complex biological data continues to burgeon, making the development and automation of data-driven algorithms and workflows ever-more important. Focusing on graph the-oretical methods, we study graph construction and analytics for two foundational problems. In the first, we explore techniques for the thresholding of simple, undirected, edge-weighted biologicalgraphs. In the second, we build resting state brain graphs from magnetoencephalographic data, on which we use a number of graph metrics to compare individuals, brainwaves and epoch lengths.In a separate effort, we move down the evolutionary ladder and take a look at the functional and metabolic differences between Escherichia coli phylotypes. Throughout, we develop novel data-driven methodologies and focus on exposing underlying assumptions of previous data-analysis workflows.