CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: Nonparametric empirical Bayes mixture models in Hi-C peak calling and Allelic Imbalance detection from ChIP-seq Authors:  Qi Zhang - University of Nebraska Lincoln (United States) [presenting]
Abstract: Over-dispersion is a common phenomenon observed in NGS data. If not properly modeled, it may severely influence the accuracy of statistical inference, e.g., peak calling, detecting differentially expressed genes or allelic imbalance. Commonly used parametric mixture models fail to capture such overdispersion due to the parametric constraint. We propose an non-parametric empirical Bayes framework for modeling NGS data, and we apply it to the problems of detecting allelic imbalance from ChIP-Seq data and calling peaks from HiC data.