The paper by Dudoit, Yang, Callow, and Speed presents statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. The authors propose new methods for image analysis and normalization, which are crucial for pre-processing the data. Given normalized data, the problem of differential expression is framed as a multiple hypothesis testing problem, where the null hypothesis is that there is no association between gene expression levels and the response or covariates of interest. Differentially expressed genes are identified using adjusted p-values from a multiple testing procedure that strongly controls the family-wise Type I error rate and accounts for the dependence structure between gene expression levels. No specific parametric form is assumed for the distribution of the test statistics, and a permutation procedure is used to estimate adjusted p-values. The paper also discusses various data displays for visualizing differentially expressed genes and their features. The methods are applied to microarray data from a study of gene expression in mice with very low HDL cholesterol levels, comparing the results with those from single-slide methods. The key contributions include new image analysis and normalization techniques, a multiple testing procedure, and the use of permutation to estimate adjusted p-values.The paper by Dudoit, Yang, Callow, and Speed presents statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. The authors propose new methods for image analysis and normalization, which are crucial for pre-processing the data. Given normalized data, the problem of differential expression is framed as a multiple hypothesis testing problem, where the null hypothesis is that there is no association between gene expression levels and the response or covariates of interest. Differentially expressed genes are identified using adjusted p-values from a multiple testing procedure that strongly controls the family-wise Type I error rate and accounts for the dependence structure between gene expression levels. No specific parametric form is assumed for the distribution of the test statistics, and a permutation procedure is used to estimate adjusted p-values. The paper also discusses various data displays for visualizing differentially expressed genes and their features. The methods are applied to microarray data from a study of gene expression in mice with very low HDL cholesterol levels, comparing the results with those from single-slide methods. The key contributions include new image analysis and normalization techniques, a multiple testing procedure, and the use of permutation to estimate adjusted p-values.