Dynamic Clustering of Gene Expression Data Using a Fuzzy Approach

Abstract

The amount of gene expression data gathered in the last decade has increased exponentially due to modern technologies like micro array and next-generation sequencing, which allow measuring the levels of expression of thousands of genes simultaneously. Clustering is a data mining technique often used for analysing this kind of data, as it is able to discover patterns in genes that are very important for understanding functional genomics. To study biological processes which are dynamic by nature, researchers must analyse data gradually, as the processes evolve. There are two ways to achieve this: perform re-clustering from scratch every time new gene expression levels are available, or adapt the previously obtained partition using a dynamic clustering algorithm, which is more efficient. In this paper we propose a fuzzy approach for dynamic clustering of gene expression data and we prove its effectiveness through a set of experimental evaluations performed on a real-life data set.