Centralization: a new method for the normalization of gene expression data.

摘要: Microarrays measure values that are approximately proportional to the numbers of copies different mRNA molecules in samples. Due technical difficulties, constant proportionality between measured intensities and per cell is unknown may vary for arrays. Usually, data normalized (i.e., array-wise multiplied by appropriate factors) order compensate this effect enable informative comparisons experiments. Centralization a new two-step method computation such normalization factors both biologically better motivated more robust than standard approaches. First, each pair arrays quotient constants estimated. Second, from resulting matrix pairwise quotients an optimally consistent scaling samples computed.