Some methods and models for analyzing time-series gene expression data

Abstract

Experiments in a variety of fields generate data in the form of a time-series. Such time-series profiles, collected sometimes for tens of thousands of experiments, are a challenge to analyze and explore. In this work, motivated by gene expression data, we provide several methods and models for such analysis. The methods developed include new clustering techniques based on nonparametric Bayesian procedures, and a confirmatory methodology to validate that the clusters produced by any of these methods have statistically different mean paths.