We examined within mouse and between mouse variation in more than 22,000 protein coding genes and identified groups of genes with shared patterns of variation that are enriched for known biolo gical functions. To facilitate exploration of our data, we have created an on line resource that includes graphical displays, test statistics, and gene groupings for all tran scripts characterized in this study indi vidualvariation. shtml. Results We performed a microarray experiment using the Illu mina Sentrix Mouse 6 v1. 1 BeadChip microarray plat form to study transcript variation in 10 Inhibitors,Modulators,Libraries week old male C57BL 6J mice. Six pairs of siblings were co housed from weaning under uniform environmental conditions. From each mouse we obtained duplicate samples of adipose, heart, kidney, and liver tissues by splitting whole organs Inhibitors,Modulators,Libraries or tissues Batimastat prior to homogenization and RNA extraction.

Adipose, heart, and liver tissues were coarsely cut into pieces and divided into two samples that were homogenized sepa rately in order to extract RNA. The left and right kid neys were also homogenized separately. We computed Inhibitors,Modulators,Libraries a decomposition of variance for each probe on the array. The within mouse variance component cap tures biological variance between two dissected tissue samples as well as technical variance due to sample and microarray processing. The between mouse variance component reflects differences between individual mice. We repeated gene expression assays on the liver sam ples, using the Affymetrix Whole Transcript Mouse Gene 1. 0 ST array, to provide validation on a different measurement platform.

Expressed genes and variable genes We declared a gene to be expressed if the probe inten sity was greater than the 95th percentile of the negative control probes for Inhibitors,Modulators,Libraries both samples in at least 1 of the 12 mice. A total of 12657 genes, representing 55% of the annotated probes on the array, were expressed in at least one of the four tissues. Across tissues, the number of expressed genes ranged from 8919 in liver to 11204 in adipose tissue. We computed the total variance, s2, across all samples for each gene in each tissue. Liver and kid ney have relatively few genes of high variability but heart and adipose have many. We tested the hypothesis that the distribution of total variance occurred by chance using a c2 test and found significantly greater variance than expected in each tissue.

We applied coexpression network analysis to the top 2500 genes in each tissue, which we refer to as the vari able genes. We decomposed total variance for each gene into within mouse and between mouse compo nents. The distribution of between mouse variance com ponents was similar across all four tissues. Adipose tissue showed the greatest number of genes with a large within mouse component followed by heart, kidney, and liver.