Modeling Dependence in Reverse Phase Protein Arrays

• Activity Number: 26
• Type: Topic Contributed
• Date/Time: Sunday, August 1, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Donatello Telesca*+ and Peter Muller and Yuan Ji
• Companies: University of California, Los Angeles and MD Anderson Cancer Center and MD Anderson Cancer Center
• Keywords: Mixture Models ; Markov Random Fields ; MCMC ; Trans-dimensional problems

Abstract

Reverse Phase Protein Arrays (RPPA) is a high-throughput functional proteomic technology that provides a quantification of the expression for targeted proteins selected from molecular pathways. Statistical inference centers on identifying over-abundant proteins and patterns of molecular interaction between proteins. We propose a probability model that represents molecular interactions as hidden relations of conditional dependence between normal and over-abundance states. The proposed stochastic scheme builds on the causal interpretation of the target pathway to define an informative prior on the hidden conditional dependence structure. Our inference identifies relevant sub-pathways in relation to the unfolding of a biological process under study.