Inference of Temporally Varying Bayesian Networks

TL;DR

This paper introduces a nonparametric Bayesian method for inferring dynamic gene regulatory networks over time, capturing changes in network structure using hidden states modeled by a Hierarchical Dirichlet Process Hidden Markov Model.

Contribution

The paper presents a novel approach combining hierarchical Dirichlet process HMMs to infer temporally varying network structures without predefining the number of states.

Findings

Effective on real microarray data

Performs well on simulated data

Captures dynamic network changes

Abstract

When analysing gene expression time series data an often overlooked but crucial aspect of the model is that the regulatory network structure may change over time. Whilst some approaches have addressed this problem previously in the literature, many are not well suited to the sequential nature of the data. Here we present a method that allows us to infer regulatory network structures that may vary between time points, utilising a set of hidden states that describe the network structure at a given time point. To model the distribution of the hidden states we have applied the Hierarchical Dirichlet Process Hideen Markov Model, a nonparametric extension of the traditional Hidden Markov Model, that does not require us to fix the number of hidden states in advance. We apply our method to exisiting microarray expression data as well as demonstrating is efficacy on simulated test data.