Statistics Department Seminar Series: Johann Gagnon-Bartsch, Assistant Professor, Department of Statistics, University of Michigan

High-dimensional data, such as genomics and neuroimaging data, are plagued by unwanted variation -- systematic errors introduced by uncontrolled variations in experimental conditions. This unwanted variation is often stronger than the variation of interest, making analysis of the data challenging, and severely impeding the ability of researchers to capitalize on the promise of the technology.  One of the biggest challenges to removing unwanted variation is that the factors causing the variation are unmeasured or simply unknown.  This makes the unwanted variation difficult to identify; the problem is essentially one of unobserved confounders.  In my talk, I will discuss the use of negative controls and replicates to help solve this problem.  A negative control is a variable known a priori to be unassociated with the factor of interest.  Replicates are repeated measurements of a single experimental unit.  Both negative controls and replicates allow researchers to partly identify unwanted variation, and separate it from the biological variation of interest.  I will present examples using microarray data, nanostring data, and diffusion tensor imaging data.