Methods

Study subjects
We downloaded the SNPs data from Haplotype Map (HapMap) phase 3 (http://www.hapmap.org) for CEU (Utah residents with Northern and Western European ancestry), JPT (Japanese in Tokyo, Japan), and YRI (Yoruba in Ibadan, Nigeria). We focused on the gene-based SNP associations in the three ethnicities, because ethnicity is a highly heritable polygenic quantitative trait of biomedical importance. Ethnicity-specific SNPs were obtained by eliminating common SNPs.

Enrichment analysis for SNP-based gene set
Ethnicity-specific HapMap SNPs were mapped to genes from UCSC RefGene (http://genome.ucsc.edu; ver. hg18) [12]. For the mapped genes, gene set enrichment analysis (GSEA) was performed using the Database for Annotation, Visualization, and Integrated Discovery (DAVID ver. 6.7) [13] with Gene Ontology (GO) terms, including biological process (BP), cellular component (CC), and molecular function (MF). The p-values were calculated for the probability of getting a set of genes within a given GO group.

Semantic modeling
To look for diverse interactions of ethnicity-specific SNPs, we constructed a semantic model using BioXM [7], which efficiently manages knowledge, such as complex scientific research data. The model provides semantic networks with useful relationship information between participating entities. Our semantic model consists of seven entities, including "Gene [14]," "Pathway [14]," "Disease [14]," "Chemical [14]," "Drug [15]," "SNP [12]," and "ClinicalTrials (http://www.clinicaltrials.gov)", and 10 relations, including "Pathway-Gene," "Disease-Pathway," "Disease-Chemical," "Gene-Disease," "Gene-Chemical," "SNP-Gene," "Chemical-Pathway," "Chemical-Drug," "ClinicalTrials-Disease," and "Drug-ClinicalTrials." Conversion of all data to entity input format was parsed using Python.