A0300
Title: Regularized t distribution: Definition, properties and applications
Authors: Tiejun Tong - Hong Kong Baptist University (Hong Kong) [presenting]
Abstract: Omics data analysis plays an important role in biological research. An important task for gene expression data analysis is to identify genes that are differentially expressed between two or more groups. Nevertheless, as biological experiments are often measured with a relatively small number of samples, how accurately estimating the variances of gene expression becomes a challenging issue. To tackle this problem, a regularized t distribution is introduced, and its statistical properties are derived, including the probability density function and the moment generating function. The noncentral regularized t distribution is also introduced for computing the statistical power of hypothesis testing. For practical applications, the regularized t distribution is applied to establish the null distribution of the regularized t statistic and then formulate it as a regularized t-test for detecting the differentially expressed genes. Simulation studies and real data analysis show that the regularized t-test performs much better than the Bayesian t-test in the limma package, in particular when the sample sizes are small.