To demonstrate the package functionality, we obtained previously published data for 171 lymphoblastoid cell lines from 57 individuals from the HapMap CEU population (HapMap, 2003
). Expression studies based on these data have been published (Kwan et al.
) using the Affymetrix Human Exon ST 1.0 arrays [available through GEO (Edgar, et al.
) accession number GSE9372]. In that work, Kwan and co-workers explored the association between genotypic differences and expression values for the entire genome. Leukotriene pathway genes are of interest because they are involved in the inflammatory response, which is a central part of the pathophysiology of cardiovascular disease. The relation between expression levels and a set of SNPs in the leukotriene pathway genes has recently been shown to be directly associated with ischemic stroke (Bevan et al.
). In this example, we combined the knowledge of genotype effects seen in the leukotriene cascade genes with the GSE9372 expression data and the HapMap genotype data for these genes, to perform an analysis with GeneRegionScan
gene encodes for the five-lipoxygenase activating protein which, with 5-lipoxygenase, is required for leukotriene synthesis and is therefore a vital component of the inflammatory response. We investigated all the SNPs used in Bevan et al.
and applied the same 0-1-2, 0-0-1 and 0-1-1 genetic models. 0-1-2 is a codominant model (three genotype groups per SNP separately) in which the heterozygote is valued as 1, and the homozygotes as 0 and 2, respectively. 0-0-1 and 0-1-1 are recessive and dominant models—essentially comparing groupings of heterozygote and homozygote samples with samples of the other homozygote type. A linear regression was fitted to the defined models to test the relation between specific genotype models and expression values. Further description of this algorithm can be found in the Supplementary Materials
, as well as in the software documentation.
The most interesting result was SNP rs3885907, also referred to as FL10 by Bevan et al.
, which evidenced a highly significant expression change when comparing the risk allele AA samples with the heterozygote and the non-risk allele CC samples (). In addition, this SNP was also found to confer a 1.473-fold increased risk of ischemic stroke (Bevan et al.
). shows that: (i) possessing two copies of the risk allele A results in a decreased intensity for all probes across the entire ALOX5AP
gene, and(ii) this effect is not mediated through alternative splicing, since all probes matching to the mRNA show the same trend. These results suggest that the SNP rs3885907 or a linked SNP confers a mechanism for controlling the expression level of the ALOX5AP
transcript, which could have implications in the development of cardiovascular disease.
Fig. 1. Effect of the rs3885907 SNP on the expression level of ALOX5AP. Visualization of the expression levels of all probes with sequences mapping to current ALOX5AP sequence. The data have been stratified by risk allele (AA) or non-risk allele (CC and heterozygote) (more ...)