Methods

Study participants and genotyping
The Korea Association Resource (KARE) study recruited 10,038 participants aged 40 years to 69 years from the rural Ansung and urban Ansan cohorts and has been previously described in detail [6]; 1,196 subjects were excluded due to poor genotyping data, and we also excluded subjects with prehypertensive status (120 mm Hg < SBP < 140 mm Hg and/or 80 mm Hg < DBP < 90 mm Hg). In total, 6,420 participants-1,968 hypertensive cases with hypertensive therapy or SBP ≥ 140 mm Hg or DBP ≥ 90 mm Hg and 4,452 controls with SBP ≤ 120 mm Hg and DBP ≤ 80 mm Hg-were examined for a hypertension case control study.
The Affymetrix Genome-Wide Human SNP array 5.0 (Affymetrix, Inc., Santa Clara, CA, USA) was used to genotype KARE study individuals. The accuracy of the genotyping was examined by Bayesian Robust Linear Modeling using the Mahalanobis distance (BRLMM) genotyping algorithm [7]. The sample and SNP quality control criteria have been described [6]. In brief, samples with accuracies that were lower than 98%, high missing genotype call rates (≥4%), high heterozygosity (>30%), or gender biases were excluded. SNPs were excluded according to filter criteria as follows: SNP call rate > 5%, minor allele frequency < 0.01, and Hardy-Weinberg equilibrium p < 1 × 10-6. After quality control, 8,842 individuals and 352,228 markers remained.

Ascertaining ROS- and hypertension-related genes
The SciMiner [8] web-based literature mining tool was used to obtain gene sets associated with ROS and hypertension. SciMiner was run on a query of "Reactive Oxygen Species" [MeSH] AND "Hypertension" [MeSH], identifying ROS-hypertension articles and genes as of April 24, 2012. We also retrieved genes for these ROS-hypertension articles from NCBI gene2pubmed (ftp://ftp.ncbi.nlm.nih.gov/gene/DATA) data. The newly found genes from gene2pubmed were added to the ROS-hypertension gene set. The positions of genes in the human genome were downloaded from the Ensembl Biomart database (NCBI build 36). Some gene symbols were different from the results of SciMiner and Biomart, such as NOS2A → NOS2 and STN → EEF1A2. The functional analysis tools, such as SciMiner, WebGestalt [9, 10], and DAVID [11, 12], were used for enrichment analysis to find the pathway with ROS-hypertension-associated genes, and the statistical significance of biological functions was calculated with Benjamini and Hochberg-adjusted p < 0.05 as the cutoff.

Statistical analyses
PLINK version v1.07 (http://pngu.mgh.harvard.edu/~purcell/plink) was used to perform the association analysis, and the hypertension case control study was tested by logistic regression analysis. The association tests were based on an additive genetic model and adjusted for recruitment area, age, sex, and body mass index.