Clustering techniques can be used in microarray analysis to (i) facilitate visual display and interpretation of experimental results, and (ii) suggest the presence of subgroups of objects (genes or samples) that behave similarly. The input of a cluster analysis are the gene expression values of the samples in an experiment, with no additional phenotype information. Depending on the approach, the output can be a list of subgroups, or a visualization that simplifies manually establishing subgroups. In some applications, unsupervised methods are used even though phenotype information is available. The goal is often to see how the clusters of samples that arise from an unsupervised approach compare to the known phenotypes.

0 0

Post a comment