Finding Characteristic Biology Patterns in Cancer Microarrays

Genetic and environmental differences are known to affect gene expression. The natural variance of expression of many genes affects the control system in any tissue. A finite set of controls must respond to these perturbations, causing regular patterns of altered gene expression characteristic of the system. In order to examine these ideas, a simple method the summarise and test the observed patterns is presented. In this paper, a classified microarray data is used to find relationships between genes. The quantitative data from microarrays can be classified as up or down, allowing estimation of significance by Monte Carlo methods. Spectral analysis and singular value decomposition were also used to study the gene expression pattern. Successive SVD vectors can identify obvious clusters of related genes.