Using Finite Poisson Mixture Models for Gene Differential Expression Analysis in RNA-Seq

• Activity Number: 571
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Han Wu*+ and Yu Zhu
• Companies: Purdue University and Purdue University
• Keywords: RNA-Seq ; Poisson mixture model ; Differential expression ; EM algorithm

Abstract

RNA-Seq is a powerful technique in genome research. As much as the improved sensitivity and coverage, RNA-Seq also brings challenges for data analysis. The massive amount of sequence reads data, excessive variability, uncertainties, and bias stemming from multiple sources all make the analysis difficult. In this talk, we propose to use finite Poisson mixture models for gene differential expression analysis based on RNA-Seq data. Finite Poisson mixture models were recently used to characterize base-level RNA-Seq reads count data and quantify gene level expression, and demonstrated excellent performance and great potential. These models allow us to quantitatively assess various variability in RNA-Seq data and in particular to properly assess accuracy in gene expression quantification. Moreover, it is straightforward to incorporate other covariates into finite mixture models. Therefore, these models also allow us to account for additional information and correct for different sources of bias. We applied the proposed method for detecting differentially expressed genes or transcripts in two real RNA-Seq data sets. The results will also be reported in this talk.