Clustering gene expression data using continuous Markov models

The Coxian phase-type distribution is a special type of Markov model that represents a process as consisting of phases or states which change as time progresses. It is this phase-type representation of the process that makes the Coxian phase-type distributions appealing to use by easing the complexity of the system and in some cases further reducing the amount of cumbersome numerical calculations required. The main aim of this paper is to introduce a new clustering technique based on the Coxian phase-type distribution. The modelling technique is still formed using a continuous Markov model but in addition allows the modelling of similar characteristics within each cluster. Such a technique will provide insights into the data by identifying clusters whereby each cluster may be described as behaving in a certain way or cluster members behaving in similar ways. The clustering technique in this paper may be used to model the behaviour of gene expression data over continuous time.