A1168
Title: Differential gene expression via local false discovery rate with nonparametric maximum likelihood
Authors: Sangkon Oh - Pukyong National University (Korea, South) [presenting]
Geoffrey McLachlan - University of Queensland (Australia)
Abstract: Two-component mixture models are particularly useful for identifying differentially expressed genes, but their performance can deteriorate markedly when the alternative distribution departs from parametric assumptions or symmetry. A semiparametric mixture model is proposed in which the null component follows a standard normal distribution, while the alternative is modeled as a skew-normal scale mixture with an unspecified scale-mixing distribution. This formulation accommodates skewness and heavy-tailed behavior, providing a flexible and computationally tractable framework for differential gene expression analysis under a nonparametric maximum likelihood paradigm, without imposing restrictive distributional assumptions. Identifiability and consistency of the proposed model are established and an efficient estimation algorithm is developed that incorporates nonparametric maximum likelihood estimation of the scale distribution. Numerical studies demonstrate notable improvements over existing parametric and nonparametric approaches for modeling the alternative distribution, and applications to colon cancer and leukemia datasets illustrate reduced false discovery and false negative rates.
