Claudio Fuentes, Ph.D., Assistant Professor, Department of Statistics, Oregon State University

"Applications of a Bayesian Hypothesis Testing Framework to the  Analysis of Genetic Data"

In this talk, I will briefly discuss two different applications of a Bayesian hypothesis testing framework to the analysis of genetics data. First, I will focus on the overdispersion problem that is often exhibited in RNA-seq data and introduce a marginal maximum likelihood estimator for the dispersion parameter in a negative binomial model.  Then, we will use this new estimator to propose a Bayesian hypothesis test for detecting differentially expressed genes.

Second, we will address the problem of determining weather the number of clusters found in given data set is statistically siginificant.  Specifically, we make use of a Bayesian framework to examine the hypotheses H0 : ?=1 vs. H1 : ?=k, where ? denotes the true number of clusters in a given population. We illustrate the methodology using NIR spectroscopy data coming from a study in maize genetics, where the existence of clusters in the data may suggest the presence of mutant genes.

Stay connected TwitterFacebookLinkedInYouTubeInstagram