This supervised analysis presumes that any meaningful differences are between the predetermined groups of samples. An unsupervised analysis uses no prior knowledge about how the samples are related. As an example, global hierarchical clustering was used to discover the interferon signature in the blood of some but not all SLE patients. Closer integration of biological knowledge of genes with the analysis of expression data can enable more detailed examination of the patient samples. Gene Set Enrichment Analysis is a knowledge-based method to identify genes differentially expressed that share common biological functions or are in the same biochemical pathways. This type of analysis with sets of genes that are specifically expressed indifferent Orbifloxacin immune cell subsets can be used to identify the presence of these subsets in disease blood or tissue. However, the results are only qualitative, and systematic analysis of relative proportions or activation states of these subsets is not possible by this method. The deconvolution based on synchronized populations of yeast cells at specific points of the cell cycle predicted the phases occupied by different cell cycle mutants. In Labetalol hydrochloride another application, Wang et al. analyzed mouse mammary tissue and used the residuals of their fit to separate the differential expression due to changes in tissue composition from those due to intrinsic gene regulation. In both these studies expression signatures of homogeneous samples of cells enabled the interpretation of the cellular composition of a complex tissue. A biological sample from a patient with an autoimmune disease typically contains 
various different immune cell subsets, and the process of microarray deconvolution can quantify their relative proportions. Essentially, the expression of each gene in the sample is modeled as a linear combination of the expression of that gene in each of the cells comprising that sample. If the expression signature of each immune cell subset is known, then the fractions of each subset in the sample can be determined by solving a linear equation to best fit the fractions of cell subsets to the whole sample��s expression signature. This first step of experimentally determining the signatures of the constituent parts is critical because it defines the framework of the results of deconvolution. The different cell types present in blood can be purified in order.