Results

Ascertaining ROS and hypertension candidate genes
With the results of SciMiner, queried with "'Reactive Oxygen Species' [MeSH] AND 'Hypertension' [MeSH]", 574 genes were obtained from 903 ROS-hypertension-related articles; 49 genes were found through the NCBI gene2pubmed data with these 903 papers, and only 2 genes out of 49 genes were new to the 574 SciMiner genes. Three hundred seventeen genes (55%) among the 576 ROS-hypertension genes were referenced in only 1 article (Table 1) and were excluded for further analysis, with 259 genes remaining.
Using Ensembl Biomart (NCBI Build 36), we then extracted the position information of 259 genes, and the genes located on chromosomes X, Y, and MT were also excluded. Finally, 237 genes that included SNPs genotype information from KARE data within the gene boundary (±20 kb upstream and downstream of the gene) were selected, and 2,945 SNPs were tested for hypertension association analysis. The frequently mentioned genes (number of papers > 40) in the ROS-hypertension papers are shown in Table 2.

Association analysis of hypertension
We examined 2,945 SNPs of 237 genes for a hypertension case control study by logistic regression analysis; 68 genes were significantly associated with hypertension (p < 0.05) (Table 3). The most significant SNP was rs2889611 within mitogen-activated protein kinase 8 (MAPK8; p = 2.70 × 10-5; odds ratio [OR], 0.82; confidence interval [CI], 0.75 to 0.90), and rs1356415 from PROM1 and rs4536994 from KDR were strongly associated with hypertension (p = 3.45 × 10-4; OR, 1.18; CI, 1.08 to 1.29 and p = 3.73 × 10-4; OR, 1.19; CI, 1.08 to 1.31, respectively).

Functional analysis of ROS-hypertension gene set
The 68 targets that were significantly associated with ROS and hypertension were tested for functional enrichment analysis. Three functional analysis tools, SciMiner, Web-Gestalt, and DAVID, identified 34 significantly over-represented biological functions in the Kyoto Encyclopedia of Genes and Genomes pathway [13, 14]. The most significant biological pathway from the 3 functional analysis tools was focal adhesion, involved in the cell communication pathway group (Table 4). The most frequent pathway group was cancer pathways (n = 9), such as glioma and pancreatic cancer. Eight other pathways related to signal transduction and 11 organismal system pathways (level 1) related to the immune system, endocrine system, and nervous system were significantly identified.