"Statistical Modeling and Inference for Temporally Dependent RNA-seq Data"

Abstract: Scientists study messenger ribonucleic acid (mRNA) levels to track gene activity in biological systems.  Tens of thousands of gene mRNA levels can be measured simultaneously in a biological sample using RNA sequencing (RNA-seq) technology.  By observing how mRNA levels change across samples of different types or across samples taken from plants or animals receiving different treatments in an experiment, scientists gain clues about how genes function together in biological systems.  In some experiments, samples are collected from each experimental unit at multiple time points.  The RNA-seq measurements corresponding to samples extracted from a single experimental unit tend to be correlated.  This talk will introduce a statistical modeling and inference approach that can be used to account for such correlation when drawing conclusions from RNA-seq experiments.  Although the number of samples is often small, we do our best to capitalize on data from many genes to learn appropriate correlation structures and estimate correlations.  We use a parametric bootstrap approach for inference that allows for the control of false discovery rate when testing genes for changes in mRNA levels caused by treatments of interest.  We will discuss the analysis of example experiments involving pig response to immune system stimulation and barley response to fungal infection.

This talk will present joint work with Meiling Liu (Iowa State) and Yet Nguyen (Old Dominion).

Host: Nan Lin

Tea @ 3:45 in room 200