ABSTRACT
Clustering methods have become some of the most prevalent exploratory approaches for evaluating gene expression experiments. Methods such as hierarchical clustering provide a relatively concise visualization of the whole experiment, allowing us to see which genes and samples tend to “cluster” together. From a broader perspective, the goal of cluster analyses is pattern recognition: to identify concise systematic patterns amongst the large amount of data produced by microarray experiments. It must be stressed, however, that clustering methods should be considered exploratory approaches to the analysis of gene expression arrays. In-depth evaluation of the resulting cluster structure or implementation of other hypothesis-driven analyses are critical steps that follow cluster analyses.
