Dimension reduction in structured biological data

Biological data often has sample structure that requires consideration when performing dimension reduction. For example, in case-control data, how can dimension reduction methods take best advantage of the controls to identify variation specific to the cases? In this talk, I will discuss a number of methods we have developed that exploit the structure of the data when performing dimension reduction, including spatiotemporal data and case-control experimental data. I will highlight the benefits of using these methods in hypothesis development and feature selection beyond standard methods in this space across a number of relevant experimental datasets.