ABSTRACT
Microarray data analysis involves analyzing the data from the gene expression matrix to sort out differentially expressed genes and recognition of patterns.1,2 The comparisons are done either in rows of the matrix or in the column of the matrix. Genes with similar expression values may be coregulated while the same gene with different expression values under different experimental conditions may be differentially expressed. Once these similarities or differences are observed, further data analysis can be carried out in supervised and unsupervised fashion.3-5
Supervised analysis mainly focuses on classifying the data on the basis of previously known classes. This type of analysis includes class comparison experiments and class prediction experiments. The analysis methods used for such an analysis include support vector machines (SVM) and regression trees which use a prede ned class to compare and analyze the new data.5 In unsupervised analysis, there is no application of prede ned classes and the data are clustered without any prior knowledge about them. Hierarchical clustering and k-mean clustering are examples of some of the clustering methods that are used for unsupervised analysis of class discovery.5 Apart from these, there is one more kind of microarray experiment called the mechanistic analysis, which looks for the mechanism behind the expression of genes. MAPPFinder, EASE, and GOMiner are some of the software that are used for the prediction of mechanism. This approach is still in the budding stage and requires a lot more work for it to be recognized as an established method of analysis. Quackenbush5 has summarized the types of analysis, and the various algorithms used for each analysis are shown in Table 25.1.
