6 - Statistical analysis of array data: Dimensionality reduction, clustering, and regulatory regions

Summary

Problems and approaches

Differential expression is a useful tool for the analysis of DNA microarray data. However, and in spite of the fact that it can be applied to a large number of genes, differential analysis remains within the confines of the old one-gene-at-a-time paradigm. Knowing that a gene's behavior has changed between two situations is at best a first step. In a cancer experiment, for instance, a significant change could be associated with a direct causal link (activation of an oncogene), a more indirect chain of effects (signaling pathway), a non-specific related phenomenon (cell division), or even a spurious event completely unrelated to cancer (“noise”).

Most, if not all, genes act in concert with other genes. What DNA microarrays are really after are the patterns of expression across multiple genes and experiments. And to detect such patterns, additional methods such as clustering must be introduced. In fact, in the limit, differential analysis can be viewed as a clustering method with only two clusters: change and nochange. Thus, at the next level of data analysis, we want to remove the simplifying assumption that genes are independent and look at their covariance, at whether there exist multi-gene patterns, clusters of genes that share the same behavior, and so forth.