Tuesday, July 19 2022
2:00pm - 4:00pm
PhD Thesis Presentation
Statistical Tools for Longitudinal Microbiome data: Simulation Frameworks, Method Comparisons, and Development of Association Tests for Longitudinal Microbiome Data

Understanding omics, the molecules, and processes that enable life to function is a central component of advancing human health. In particular, studying relationships between the microbiome and health over time can illuminate the impact of environmental factors on individuals. Recent advancements in high-throughput sequencing technologies and human health research have created a need for statistical methods that identify biomarkers within microbiomes associated with longitudinal changes in human health. In this dissertation, we create and compare statistical methods for simulating longitudinal microbiome data and testing hypotheses of association between the microbiome and health in longitudinal studies. We create the simulation software SimMiL to simulate longitudinal microbiome data allowing for the creation of flexible and realistic biological structures using real-world microbiome data sets. Three statistical tests commonly used to test the association between health and microbiome in longitudinal studies are then compared across a variety of simulation scenarios. Finally, modified kernels are introduced and used to test more precise hypotheses for longitudinal microbiome data that were not previously testable using existing kernel association tests.
Speaker:Nicholas Weaver

Download as iCalendar