BACKGROUND: Stochastic dependence between gene expression levels in microarray data is of critical importance for the methods of statistical inference that resort to pooling test-statistics across genes. It is frequently assumed that dependence between genes (or tests) is sufficiently weak to justify the proposed methods of testing for differentially expressed genes. A potential impact of between-gene correlations on the performance of such methods has yet to be explored. RESULTS: The paper presents a systematic study of correlation between the t-statistics associated with different genes. We report the effects of four different normalization methods using a large set of microarray data on childhood leukemia in addition to several sets of simulated data. Our findings help decipher the correlation structure of microarray data before and after the application of normalization procedures. CONCLUSION: A long-range correlation in microarray data manifests itself in thousands of genes that are heavily correlated with a given gene in terms of the associated t-statistics. By using normalization methods it is possible to significantly reduce correlation between the t-statistics computed for different genes. Normalization procedures affect both the true correlation, stemming from gene interactions, and the spurious correlation induced by random noise. When analyzing real world biological data sets, normalization procedures are unable to completely remove correlation between the test statistics. The long-range correlation structure also persists in normalized data.

MOTIVATION: Class distinction is a supervised learning approach that has been successfully employed in the analysis of high-throughput gene expression data. Identification of a set of genes that predicts differential biological states allows for the development of basic and clinical scientific approaches to the diagnosis of disease. The Independent Consistent Expression Discriminator (ICED) was designed to provide a more biologically relevant search criterion during predictor selection by embracing the inherent variability of gene expression in any biological state. The four components of ICED include (i) normalization of raw data; (ii) assignment of weights to genes from both classes; (iii) counting of votes to determine optimal number of predictor genes for class distinction; (iv) calculation of prediction strengths for classification results. The search criteria employed by ICED is designed to identify not only genes that are consistently expressed at one level in one class and at a consistently different level in another class but identify genes that are variable in one class and consistent in another. The result is a novel approach to accurately select biologically relevant predictors of differential disease states from a small number of microarray samples. RESULTS: The data described herein utilized ICED to analyze the large AML/ALL training and test data set (Golub et al., 1999, Science, 286, 531-537) in addition to a smaller data set consisting of an animal model of the childhood neurodegenerative disorder, Batten disease, generated for this study. Both of the analyses presented herein have correctly predicted biologically relevant perturbations that can be used for disease classification, irrespective of sample size. Furthermore, the results have provided candidate proteins for future study in understanding the disease process and the identification of potential targets for therapeutic intervention.

BACKGROUND: Affymetrix microarrays are used by many laboratories to generate gene expression profiles. Generally, only large differences (> 1.7-fold) between conditions have been reported. Computational methods to reduce inter-array variability might be of value when attempting to detect smaller differences. We examined whether inter-array variability could be reduced by using data based on the Affymetrix algorithm for pairwise comparisons between arrays (ratio method) rather than data based on the algorithm for analysis of individual arrays (signal method). Six HG-U95A arrays that probed mRNA from young (21-31 yr old) human muscle were compared with six arrays that probed mRNA from older (62-77 yr old) muscle. RESULTS: Differences in mean expression levels of young and old subjects were small, rarely > 1.5-fold. The mean within-group coefficient of variation for 4629 mRNAs expressed in muscle was 20% according to the ratio method and 25% according to the signal method. The ratio method yielded more differences according to t-tests (124 vs. 98 differences at P < 0.01), rank sum tests (107 vs. 85 differences at P < 0.01), and the Significance Analysis of Microarrays method (124 vs. 56 differences with false detection rate < 20%; 20 vs. 0 differences with false detection rate < 5%). The ratio method also improved consistency between results of the initial scan and results of the antibody-enhanced scan. CONCLUSION: The ratio method reduces inter-array variance and thereby enhances statistical power.