Phage display is a biological process to screen random peptide libraries for ligands that bind to a target, e.g., an organ, with high affinity. Based on a count data set from an innovative multi-stage phage display experiment, we propose a class of Bayesian mixture models to classify peptides into three groups that exhibit different trends across stages. The peptides with an ascending pattern in their counts over the stages are of particular interest and selected for further validation by the biological investigators. The proposed Bayesian models are specifically designed to describe the high-dimensionality and complex correlation structure in the phage data; however, they can be easily extended to other settings. We present a simulation study to examine the properties of the proposed methodology. After implementing the proposed model for the practical phage data set, we provide a list of discovered peptides.
Ji, Yuan; Yin, Guosheng; Tsui, Kam-Wah; Kolonin, Mikhail G.; Sun, Jessica; Arap, Wadih; Pasqualini, Renata; and Do, Kim-Anh, "Bayesian Mixture Models for Complex High-Dimensional Count Data in Phage Display Experiments" (December 2005). UT MD Anderson Cancer Center Department of Biostatistics Working Paper Series. Working Paper 13.