A Semiparametric Bayesian Framework for Identifying Up or Down Regulated Genes in Subjects with Neurocysticercosis (NCC) Associated Epilepsy

• Activity Number: 338
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Michael Anderson*+ and Cheuk H. Leung and Suzanne R. Dubnicka and Douglas A. Drevets and Vedantam Rajshekhar and Anna Oommen and Prabhakaran Vasudevan and Josephin Justin Babu and Ramajayam Govindan and Helene Carabin
• Companies: University of Oklahoma and OUHSC and Kansas State University and OUHSC and Christian Medical College and Christian Medical College and Christian Medical College and Christian Medical College and Christian Medical College and University of Oklahoma Health Sciences Center
• Keywords: micro-array ; Bayesian methods ; neurocysticercosis ; Bayesian classification ; semi-paramtric methods ; epidemiology

Abstract

Neurocysticercosis (NCC) is a neurological inflammatory process caused by Taenia solium larvae cysts in the brain and is one of the most common causes of seizures in developing countries. Subjects reporting to the Department of Neurological Sciences at the Christian Medical College in Vellore, India were recruited for participation in a study to determine inflammation relevant gene expression profiles specific to NCC in peripheral blood monocytes. While technology to measure and describe gene regulation has become quite sophisticated, statistical methods for analyzing such data have remained somewhat pedestrian. Traditional micro-array analysis often relies on unwarranted assumptions of normality and a battery of several thousand t-tests to identify significantly up or down regulated genes between groups. In this work we propose a semi-parametric Bayesian framework as a robust, probability based alternative to the t-test approach. Combined posterior change measures based on suitable priors and interpreted as the probability of correct group affiliation, are used to identify deferentially expressed genes. We compare the top genes identified by both methods in the NCC data set.