Assessing Gene Significance from cDNA Microarray Expression Data via Mixed Models

Abstract

The determination of a list of differentially expressed genes is a basic objective in many cDNA microarray experiments. We present a statistical approach that allows direct control over the percentage of false positives in such a list and, under certain reasonable assumptions, improves on existing methods with respect to the percentage of false negatives. The method accommodates a wide variety of experimental designs and can simultaneously assess significant differences between multiple types of biological samples. Two interconnected mixed linear models are central to the method and provide a flexible means to properly account for variability both across and within genes. The mixed model also provides a convenient framework for evaluating the statistical power of any particular experimental design and thus enables a researcher to a priori select an appropriate number of replicates. We also suggest some basic graphics for visualizing lists of significant genes. Analyses of published experiments studying human cancer and yeast cells illustrate the results.

Transcriptomic profiling of the canine tachycardia-induced heart failure model: global comparison to human and murine heart failure

“…We used the significance analysis of microarrays (SAM) software to identify genes differentially expressed between the experimental groups with high statistical significance [27]. The SAM method provides the false discovery rate (FDR) estimate, using permutations of repeated measurements, to a set of gene-specific t-tests (http://www-stat.stanford.edu/~tibs/SAM).A two-stage linear mixed model [28] was used to identify differentially expressed genes in the following comparisons: normal LV vs. normal RV; and HF LV vs. HF RV. In the two-stage linear mixed model, the normalization model was used to remove bias due to experiment-wide systematic effects; a gene-specific model was used for gene-by-gene inference of effects such as individual subjects, ventricular chambers (LV, RV), stage (normal, end-stage HF), and chamber-stage interaction (see supplement methods).…”

Effect of the tumor promoter phenobarbital on the pattern of global gene expression in liver of connexin32‐wild‐type and connexin32‐deficient mice

“…The magnitude of the effects of strain, treatment, and interaction between treatment and strain as well as the corresponding p-values are illustrated in the form of volcano plots 21 in Figure 1, where results for each of the ϳ12,000 transcripts are represented. The levels used for discrimination (p-value Յ0.01 and a ͉log 2 expression ratio͉ Ն 1) are indicated in the figure by horizontal and vertical lines and genes meeting these criteria are represented in the grey areas of the plots.…”

Variability-specific differential gene expression across reproductive stages in sows

“…As highlighted by Wolfinger et al (2001), inference in microarray gene expression analyses is typically focused on HEAT vs. 15d HEAT vs. 45d 45d vs. 15d Figure 1 Plots of the pairwise distribution of log(BF HE=HO i ), s 2 e and p parameters in the uterus.…”

“…The analysis of these genomic data tries to identify differential gene expression (Wolfinger et al, 2001) associated with some phenotype of interest, for example muscle growth (Reecy et al, 2006) or stress tolerance (Collier et al, 2006), although all studies in domestic species typically focus on differences linked to -E-mail: Joaquim.Casellas@uab.cat the mathematical expectation (i.e. mean) of two or more groups of microarrays, without considering alternative patterns of departure.…”