Zero-Inflated Poisson Regression for Identification of Essential Genes with Tn-Seq Data

• Activity Number: 603
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Fangfang Liu*+ and Chong Wang and Peng Liu
• Companies: Iowa State University and Iowa State University and Iowa State University
• Keywords: Tn-seq ; Zero-inflated Poisson ; EM algorithm ; FDR

Abstract

Identification of genes essential to the survival of bacteria is very important for understanding the function of genes and detecting potential drug targets for antimicrobial treatment. Creation of genome-wide random transposon mutant libraries followed by next-generation sequencing results in transposon-sequencing (Tn-seq) technology that generates high-throughput data for detecting essential genes for bacteria growth. Tn-seq data consist of sequence reads around each transposon insertion sites, and such measurements are used to infer the functions of each gene on bacteria growth. In this paper we propose a zero-inflated Poisson regression method for analyzing these large and complex Tn-seq data. We derive an Expectation-Maximization (EM) algorithm to obtain parameter estimates, and we propose a multiple testing procedure that categorizes genes into each of the three states, growth-impaired, non-influential, and growth-enhanced, while controlling false discovery rate. We also apply the proposed method to a real Tn-seq dataset for a bacterial pathogen Campylobacter jejuni.