Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Association between Expression Quantitative Trait Loci and Metabolic Traits in Two Korean Populations

  • Kyung-Won Hong ,

    Contributed equally to this work with: Kyung-Won Hong, Seok Won Jeong

    Affiliation Division of Biomedical Informatics, Center for Genome Science, National Institute of Health, KCDC, Cheongwon-gun, Korea

  • Seok Won Jeong ,

    Contributed equally to this work with: Kyung-Won Hong, Seok Won Jeong

    Affiliation Division of Biomedical Informatics, Center for Genome Science, National Institute of Health, KCDC, Cheongwon-gun, Korea

  • Myungguen Chung,

    Affiliations Division of Biomedical Informatics, Center for Genome Science, National Institute of Health, KCDC, Cheongwon-gun, Korea, Division of Molecular and Life Science, Hanyang University, Seoul, Korea

  • Seong Beom Cho

    sbcho@korea.kr

    Affiliation Division of Biomedical Informatics, Center for Genome Science, National Institute of Health, KCDC, Cheongwon-gun, Korea

Association between Expression Quantitative Trait Loci and Metabolic Traits in Two Korean Populations

  • Kyung-Won Hong, 
  • Seok Won Jeong, 
  • Myungguen Chung, 
  • Seong Beom Cho
PLOS
x

Abstract

Most genome-wide association studies consider genes that are located closest to single nucleotide polymorphisms (SNPs) that are highly significant for those studies. However, the significance of the associations between SNPs and candidate genes has not been fully determined. An alternative approach that used SNPs in expression quantitative trait loci (eQTL) was reported previously for Crohn’s disease; it was shown that eQTL-based preselection for follow-up studies was a useful approach for identifying risk loci from the results of moderately sized GWAS. In this study, we propose an approach that uses eQTL SNPs to support the functional relationships between an SNP and a candidate gene in a genome-wide association study. The genome-wide SNP genotypes and 10 biochemical measures (fasting glucose levels, BUN, serum albumin levels, AST, ALT, gamma GTP, total cholesterol, HDL cholesterol, triglycerides, and LDL cholesterol) were obtained from the Korean Association Resource (KARE) consortium. The eQTL SNPs were isolated from the SNP dataset based on the RegulomeDB eQTL-SNP data from the ENCODE projects and two recent eQTL reports. A total of 25,658 eQTL SNPs were tested for their association with the 10 metabolic traits in 2 Korean populations (Ansung and Ansan). The proportion of phenotypic variance explained by eQTL and non-eQTL SNPs showed that eQTL SNPs were more likely to be associated with the metabolic traits genetically compared with non-eQTL SNPs. Finally, via a meta-analysis of the two Korean populations, we identified 14 eQTL SNPs that were significantly associated with metabolic traits. These results suggest that our approach can be expanded to other genome-wide association studies.

Introduction

Recently, large-scale genome-wide association studies (GWAS) that comprised several thousands of samples have reported many novel findings in various diseases and disease-related phenotypes [1]. These findings have been enlightening the path to the identification of disease mechanisms and biomarkers.

Among the human phenotypes, metabolic traits are frequently studied in different populations [2]. In the Korean population, common metabolic traits, such as glucose, cholesterol, and bilirubin levels, have been studied via conventional GWAS [3], [4]. However, little was explained by heritability [3]. This phenomenon, which is termed the missing heritability problem, is hard to resolve by conventional GWAS. It was suggested that the missing heritability might come from the stringent multiple testing correction of GWAS analyses [5]. This multiple testing correction is necessary to exclude false-positive loci, but simultaneously may discard many true-positive loci [6]. In other studies, it was shown that reducing the number of tests is advantageous for GWAS. In that research, the categorization of the genome-wide SNPs into functional categories provided the opportunity to reduce multiple testing and to identify functional variants [7], [8].

In addition to the missing heritability, it should be considered that most of the significant single-nucleotide polymorphisms (SNPs) used in these GWAS lay in intergenic and intron regions and had little association with changes in the protein-coding sequences of genes [1]. Thus, these SNPs likely regulate gene activity at the transcript level directly, or cooperate with other DNA variations that mediate this type of regulation. Based on these facts, expression quantitative loci (eQTL) are being actively studied for elucidating the relationship between changes in genotype and expression dynamics, which will promote the understanding of the results of GWAS [9][15].

eQTL information provides insights into the regulation of transcription and aids in the interpretation of genome-wide association studies [9]. In cases in which the allelic changes of a SNP are significantly correlated with the expression of a gene, the SNP is defined as an eQTL-SNP. Using this information, researchers try to identify trait-associated SNPs that would be otherwise hard to find. For example, Fransen and colleagues reported a GWAS for Crohn’s disease using eQTL-SNP information. Those authors selected eQTL SNPs among the GWAS results for Crohn’s disease, and performed follow-up replication studies [6]. They showed that the eQTL-based preselection for follow-up studies was a useful approach for identifying risk loci from the results of a moderately sized GWAS.

Here, we reanalyzed genome-wide associations between metabolic traits and SNPs using eQTL information. The main goal of this research was to explore metabolic trait-associated variants using an eQTL-based filtering strategy. The major eQTL SNPs used in this study were obtained from the RegulomeDB, and the other eQTL SNPs were obtained from recent reports of liver tissues [13] and from lymphoblastoid cell lines [14]. We collected the genotypes of the eQTL SNPs from the Korean Association Resource (KARE) [3], [16] and examined their association with 10 metabolic traits in two independent Korean cohorts (Ansan and Ansung).

Materials and Methods

Study subjects

The study subjects comprised two population-based cohorts, Ansung and Ansan, which have been examined as part of the Korean Genome and Epidemiology Study (KoGES). The phenotype of the cohort has been described [17]. Briefly, the subjects came from Ansung and Ansan in KyungGi-Do province, near Seoul, Korea. Written informed consent was obtained from all participants, and this research project was approved by the institutional review board of KNIH. A total of 10,038 individuals were recruited for the cohorts, and 8,842 individuals of the KoGES were analyzed by the Korean Association Resource (KARE) consortium to understand their genome-wide association with the surveyed or measured phenotypes [16].

Subjects with genotype accuracies below 98% and high missing genotype call rates (≥5%), high heterozygosity (>30%), or inconsistency in sex were excluded from subsequent analyses. Individuals who had a tumor were excluded, as were related individuals whose estimated identity-by-state (IBS) values were high (>0.80). Based on these criteria, 8,842 samples were selected; these quality-control steps were described in a previous GWAS [16].

We obtained the following information from the cohorts (Ansung and Ansan cohorts): sex; age; past disease history; anthropometric measurements, such as weight and height; and biochemical measurements, such as fasting glucose (GLU0), serum albumin (ALB), blood urea nitrogen (BUN), aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma glutamyl-transpeptidase (GGT), total cholesterol (Tchol), high-density lipoprotein cholesterol (HDLC), and triglyceride (TG) levels. Low-density lipoprotein cholesterol (LDLC) was calculated as per the Friedewald formula: LDL = Tchol – HDLC – (TG/5) (all concentrations in mg/dL). The exclusion criteria of each metabolic trait are described in Table 1.

thumbnail
Table 1. Clinical characteristics of the Ansung and Ansan cohorts and the exclusion criteria for each biochemical trait.

https://doi.org/10.1371/journal.pone.0114128.t001

Genotyping

The genotype data were obtained from KARE, which used the Affymetrix Genome-wide Human SNP array 5.0 and imputed SNPs. The genotype quality-control criteria were as reported in a previous GWAS [16]. Briefly, the criteria for the inclusion of SNPs were a genotype call rate >0.98, a minor allele frequency >0.01, and Hardy–Weinberg equilibrium (HWE) (p>1×10–6). Ultimately, 352,228 SNPs passed the quality-control process. SNP imputation was performed with IMPUTE [18] using the JPT and CHB sets of HapMap Phase 2 as the reference. After removing SNPs with a MAF <0.01 and a SNP missing rate >0.05, we combined the genotyped SNPs and imputed SNPs; a total of 1.8 million SNPs were used in the subsequent study.

Population stratification was tested by a principal component analysis using the EIGENSOFT software [19]. To prevent overrepresentation of regions with more redundant SNPs, we used the indep-pairwise command in PLINK [20] to reduce linkage disequilibrium between the remaining variants by eliminating any SNP that had a pairwise r2>0.3 with any other SNP in a 1500 bp window (step size, 150 bp). This reduced the original dataset to 93,877 SNPs; subsequently, we used smartpca [19].

eQTL-SNP selection

Most of the eQTL-SNP resources that were used in this analysis are available in online databases, such as RegulomeDB (http://regulome.stanford.edu/), including several published resources for various cell types/tissues, such as monocytes [9], human brain [10], lymphoblastoid cell lines [11], [12], and human liver [13]. The RegulomeDB is one of the most comprehensive eQTL-SNP databases, because it contains rich information about the products of the ENCODE project, such as transcription factor binding sites, chromatin structure, histone modification, and eQTL. The RegulomeDB classifies the human SNPs into seven categories [21], of which Category 1 includes eQTL SNPs with other information about the regulatory elements, such as transcription factor binding, chromatin structure, and histone modification. A total of 39,332 eQTL SNPs were downloaded from the RegulomeDB, Category 1. In addition, we also included two recent eQTL reports based on liver (1,078 SNPs) or lymphoblastoid (907 SNPs) tissues [14], [15]. Finally, a total of 41,317 eQTL SNPs were used to extract the same SNPs from our genotype dataset, and we extracted the 25,658 eQTL SNPs from the KARE SNPs using PLINK [20] software.

Estimation of the genetic variance that is explained by eQTL or non-eQTL SNPs

The non-eQTL SNPs were defined as the SNPs that excluded the eQTL SNPs and were in perfect linkage disequilibrium with the eQTL SNPs (r2 = 1.0 and D′ = 1.0). The genetic variances were computed via GCTA v1.24 [22], which is a tool for estimating the proportion of phenotypic variance that is explained by genome-wide SNPs for complex traits [23]. First, we estimated the pairwise genetic relationship using the make-grm option. Next, we estimated the proportion of phenotypic variance that was explained by the eQTL and non-eQTL SNPs, respectively, based on the restricted maximum likelihood (REML) [23].

Statistical analyses

The effect of a genotype was analyzed by linear regression. We calculated the effect size (beta) and standard error (SE) of minor alleles on metabolic traits for each Ansung and Ansan subject. All analyses were adjusted for age, sex, body mass index (BMI), and principal component (PC) 1 and PC2. PLINK v 1.07 [20] was used for all statistical tests. All tests were performed based on the additive model, and we combined the Ansung and Ansan results by an inverse-variance meta-analysis under the assumption of fixed effects using Cochran’s Q test, to assess between-study heterogeneity [24]. In this study, we applied false discovery rates and Bonferroni correction p-values to determine significant associations.

Results

Dataset characteristics

The clinical characteristics and sample exclusion criteria are described in Table 1. The KARE subjects consisted of two Korean populations (Ansung and Ansan), which we used as the replication set of genetic associations based on their differing clinical characteristics. However, the population stratification analysis showed similar genetic structures (S1 Figure), indicating that they constitute a replication set that can be used to identify consistent genetic effects, despite the differences in demographics and environments.

The proportion of phenotypic variance that was explained by eQTL and non-eQTL SNPs

Based on the RegulomeDB and previous reports, we isolated 25,658 eQTL SNPs from the KARE genotype dataset. The proportion of phenotypic variance that was explained by eQTL and non-eQTL SNPs for complex traits is described in Table 2. The non-eQTL sites examined were approximately 66 times more examined SNPs than the eQTL sites; however, compared with those explained by non-eQTL SNPs, the proportions of phenotypic variance that were explained by eQTL SNPs were larger for GLU0 (1.9 in eQTL vs 0.2 in non-eQTL SNPs) and TG (1.3 in eQTL vs 0.1 in non-eQTL SNPs), and similar for BUN (3.4 in eQTL vs 4.0 in non-eQTL SNPs). Moreover, the proportions of phenotypic variance that were explained by eQTL SNPs were relatively large for the remaining phenotypes, with the exception of ALB.

thumbnail
Table 2. Estimated genetic variance explained by eQTL SNPs and non-eQTL SNPs.

https://doi.org/10.1371/journal.pone.0114128.t002

Association study of eQTL-related SNPs

These eQTL SNPs were examined with regard to their association with metabolic traits in the Ansung and Ansan cohorts. Ultimately, we selected 509 eQTL SNPs that had the same effect and p-values<0.05 in both cohorts, and combined the results obtained for each cohort via a meta-analysis. All association results and meta-analysis results for the 509 SNPs are described in S1 Table.

Among the 509 SNPs analyzed, we identified significant associations using adjusted p-values<0.05 for FDR. Twenty-six SNPs for GLU0, eleven SNPs for GGT, two SNPs for Tchol, thirty-five SNPs for LDLC, and two SNPs for TG met our criteria (S1 Table, characters highlighted in bold). Because many associated SNPs were located in the same LD blocks, we selected the most significant SNP in each significant LD block (summarized in Table 3). We then used these SNPs for further annotation and compared them with the previous conventional GWAS results (http://www.genome.gov/gwasstudies). Further information for the significant eQTL positions is described in Table 4, including the genes, cell types, and ENCODE regulatory elements in metabolic-trait-related cell types derived from the blood, liver, or pancreas. No significant association was detected regarding the results of ALB, BUN, AST, ALT, and HDLC.

thumbnail
Table 3. Significantly associated SNPs in the Ansung and Ansan cohorts and meta-analysis results.

https://doi.org/10.1371/journal.pone.0114128.t003

Significantly associated eQTL SNPs

Table 3 shows that two SNPs (rs1535 and rs463302) for FPG, two SNPs (rs2251468 and rs11065774) for GGT, two SNPs (rs780092 and rs4604177) for Tchol, and two SNPs (rs12679834 and rs651821) for TG were significantly associated and passed the Bonferroni correction. The p-values of two SNPs (rs3002007 and rs1431985) for FPG, one SNP (rs13233571) for GGT, and three SNPs (rs11065774, rs9966367, and rs4604177) for TG did not pass the Bonferroni correction, but passed the FDR correction.

In this study, we focused on significantly associated SNPs. SNP rs1535 was a FADS1 and NXF1 eQTL-SNP hot spot, and the results of the meta-analysis of the association between the Ansan and Ansung populations were β = –0.943 and p-value = 3.3×10–8. SNP rs463302 was a B3GALT4 eQTL-SNP hot spot, and the results of the meta-analysis were β = 1.347 and p-value = 8.5×10–7. SNP rs2251468 was a C12orf43 eQTL-SNP hot spot, and the results of the meta-analysis were β = –0.812 and p-value = 2.2×10–7. SNP rs11065774 was an MYL2 eQTL-SNP hot spot, and the results of the meta-analysis were β = –1.009 and p-value = 2.5×10–7. SNP rs780092 was a XAB1 eQTL-SNP hot spot, and the results of the meta-analysis were β = –2.710 and p-value = 4.4×10–7. SNP rs4604177 was a FAM169A eQTL-SNP hot spot, and the results of the meta-analysis were β = –2.492 and p-value = 7.7×10–7. SNP rs12679834 was an eQTL-SNP hot spot of LPL, and the results of the meta-analysis were β = –6.183 and p-value = 4.5×10–9. Finally, SNP rs651821 was an eQTL-SNP hot spot of TAGLN, and the results of the meta-analysis were β = 5.011 and p-value = 4.7×10–9.

Discussion

In this study, we performed a GWAS of 10 biochemical traits using SNPs that were preselected based on the eQTL-SNP lists of RegulomeDB and two recent eQTL papers. The proportion of phenotypic variance that was explained by eQTL- and non-eQTL-related SNPs showed that the eQTL SNPs were more likely to be associated with the metabolic traits than were the non-eQTL SNPs. We identified 14 eQTL SNPs that were associated with metabolic traits in two Korean populations (Ansung and Ansan), in which the p-values met our multiple comparison criteria. The SNPs revealed novel candidate genes for FPG, GGT, Tchol, LDLC, and TG.

SNP rs1535 was reported as being associated with decreased plasma phospholipid levels by an European study [25], and with decreased HDLC by an Indian Asian study [26]. Although FADS1 has been studied extensively in lipid metabolism, recent genetic association studies showed its association with glucose metabolism [27]. Moreover, NXF1, a novel glucose-regulated protein, is elevated under high-glucose conditions [28]. Therefore, the expression of both the FADS1 and NXF1 genes is influenced by the rs1535 genotypes, which might represent a functional element for regulating both glucose and lipid metabolism.

SNP rs463302 has not been reported in association studies; however, our study indicated that rs463302 may contribute to B3GALT4 expression levels. B3GALT4 encodes UDP-Gal:betaGlcNAc beta 1,3-galactosyltransferase, polypeptide 4, which is a member of the beta-1,3-galactosyltransferase protein family. Although B3GALT4 mitigates ganglioside activity in neurodegenerative disorders, such as Huntington disease [29], it was also reported that gangliosides are related to the immunological pathophysiology of type 1 diabetes and to insulin resistance in type 2 diabetes [30], [31]. SNP rs463302 lies 200 bp upstream of B3GALT4, to which RNA polymerase 2A (POLR2A) binds, based on the ENCODE ChIP-seq data. In addition, the ENCODE DNase-seq and histone modification data indicated that the chromatin around this SNP is open and undergoes histone modifications. Taken together, our results and in silico evidence suggest that the rs463302 SNP may be a regulatory factor of B3GALT4 and a modifying factor of fasting glucose level.

In our study, rs2251468 was significantly associated with GGT. The SNP was recently reported to be associated with plasma homocystein levels [32]. Because homocystein concentration is a risk factor for coronary artery disease [32] and is significantly correlated with GGT [33], this SNP might be a novel candidate marker for coronary artery disease.

Although rs12679834 has not been reported in association studies, a SNP that is in high LD with rs12679834 (rs331, r2 = 0.771, D′ = 1.000) has been reported as being associated with plasma lipoprotein concentration [34]. SNP rs12679834 was an eQTL-SNP of LPL (lipoprotein lipase), which is a critical enzyme in lipid metabolism that catalyzes the hydrolysis of TGs. Dysfunction of LPL induces pathophysiological lipid-related disorders, including hyperlipidemia, dyslipidemia [35], and hypertriglyceridemia [36].

Our study had two limitations. First, although the positions of the eQTL SNPs have several evidences of the regulatory elements from ENCODE results, not all eQTL SNPs were experimented in the cell types or tissues that are directly related to the metabolic traits [37]. Thus, we assumed that the eQTL SNPs also played a similar role in the metabolic-trait-related cell types or tissues. For example, the B3GALT4 eQTL-SNP (rs463302) was experimented in cerebellum tissue; however, it is located on several ENCODE regulatory elements, such as transcription factor binding, open chromatin, and active histone modification markers, in metabolic-trait-related cell types, such as the liver and pancreas. The other limitation was that we used HapMap 2 imputed SNPs. Recently, many genome-wide association studies used 1000 Genomes Project-based imputation [38]. Although we could not use 1000 Genomes-based imputation in the current study, we will apply it in future studies.

Previous GWASs of complex traits have primarily examined DNA sequence variants (DSVs) between individuals that contribute to the susceptibility to a disease, clinical outcomes, and response to therapy [9]. However, the mechanisms that govern the expression of a phenotype are embedded not only in the DSVs, but also through their effects on various genomic components that regulate gene expression, variants, and posttranslational modifications of the encoded proteins, in conjunction with environmental factors. Thus, a complicated phenotype is the consequence of complex interactions between many genetic and nongenetic factors.

The results of eQTL-SNP GWAS might be novel targets for the design of future experimental studies. As an example, rs12740374 was identified in European and American GWAS for LDLC, and those studies reported the CELSR2 gene as a candidate gene based on positional proximity [39]. However, further analysis showed that the SNP is an eQTL of SORT1 [40]. Moreover, our approach using eQTL SNPs provides the possibility of understanding the internal mechanism that underlies the link from SNP to genes to phenotype. For example, in our study, SNP rs1535 was an eQTL of NXF1, which is regulated by the blood glucose levels. This result led us to speculate on the following internal link: the modulation of NXF1 expression by blood glucose levels may be modified by eQTL SNPs.

In conclusion, the relationships between functional eQTL SNPs and 10 biological traits may be helpful for understanding the underlying mechanism that connects genotype and phenotype. Because eQTL SNPs promote the understanding of gene expression and regulation, this approach might help identify biomarkers of metabolic traits.

Supporting Information

S1 Figure.

Population stratification of Ansung and Ansan. The PCA analysis by using the Affymetrix 5.0 SNP array were conducted by EIGENSTAT.

https://doi.org/10.1371/journal.pone.0114128.s001

(PPTX)

S1 Table.

The all associated eQTLs in both Ansan and Ansung.

https://doi.org/10.1371/journal.pone.0114128.s002

(XLSX)

Author Contributions

Conceived and designed the experiments: KWH MGC SBC. Performed the experiments: KWH SWJ. Analyzed the data: KWH SWJ. Contributed reagents/materials/analysis tools: KWH SWJ. Wrote the paper: KWH SWJ.

References

  1. 1. Welter D, MacArthur J, Morales J, Burdett T, Hall P, et al. (2014) The NHGRI GWAS Catalog, a curated resource of SNP-trait associations. Nucleic Acids Research 22:D1001–D1006.
  2. 2. Rueedi R, Ledda M, Nicholls AW, Salek RM, Marques-Vidal P, et al. (2014) Genome-wide association study reveals novel gene-metabolite-disease links. PLOS Genetics 10:e1004132.
  3. 3. Kim YJ, Go MJ, Hu C, Hong CB, Kim YK, et al. (2011) Large-scale genome-wide association studies in East Asians identify new genetic loci influencing metabolic traits. Nat Genet 43:990–995.
  4. 4. Kang TW, Kim HJ, Ju H, Kim JH, Jeon YJ, et al. (2010) Genome-wide association of serum bilirubin levels in Korean population. Hum Mol Genet 19:3672–3678.
  5. 5. Williams SM, Haines JL. (2011) Correcting away the hidden heritability. Ann Hum Genet 75:348–350.
  6. 6. Fransen K, Visschedijk MC, van Sommeren S, Fu JY, Franke L, et al. (2010) Analysis of SNPs with an effect on gene expression identifies UBE2L3 and BCL3 as potential new risk genes for Crohn’s disease. Hum Mol Genet 19:3482–3488.
  7. 7. Hong KW, Jin HS, Lim JE, Cho YS, Go MJ, et al. (2010) Non-synonymous single-nucleotide polymorphisms associated with blood pressure and hypertension. J Hum Hypertens 24:763–774.
  8. 8. Hong KW, Lim JE, Oh B (2011) A regulatory SNP in AKAP13 is associated with blood pressure in Koreans. J Hum Genet 56:205–210.
  9. 9. Zeller T, Wild P, Szymczak S, Rotival M, Schillert A, et al. (2010) Genetics and beyond – the transcriptome of human monocytes and disease susceptibility. PLoS One 5:e10693.
  10. 10. Gibbs JR, van der Brug MP, Hernandez DG, Traynor BJ, Nalls MA, et al. (2010) Abundant quantitative trait loci exist for DNA methylation and gene expression in human brain. PLoS Genet 6:e1000952.
  11. 11. Veyrieras JB, Kudaravalli S, Kim SY, Dermitzakis ET, Gilad Y, et al. (2008) High-resolution mapping of expression-QTLs yields insight into human gene regulation. PLoS Genet 4:e1000214.
  12. 12. Montgomery SB, Sammeth M, Gutierrez-Arcelus M, Lach RP, Ingle C, et al. (2010) Transcriptome genetics using second generation sequencing in a Caucasian population. Nature 464:773–777.
  13. 13. Schadt EE, Molony C, Chudin E, Hao K, Yang X, et al. (2008) Mapping the genetic architecture of gene expression in human liver. PLoS Biol 6:e107.
  14. 14. Liang L, Morar N, Dixon AL, Lathrop GM, Abecasis GR, et al. (2013) A cross-platform analysis of 14,177 expression quantitative trait loci derived from lymphoblastoid cell lines. Genome Res 23:716–726.
  15. 15. Wang X, Tang H, Teng M, Li Z, Li J, et al. (2014) Mapping of hepatic expression quantitative trait loci (eQTL) in a Han Chinese population. J Med Genet 51:319–326.
  16. 16. Cho YS, Go MJ, Kim YJ, Heo JY, Oh JH, et al. (2009) A large-scale genome-wide association study of Asian populations uncovers genetic factors influencing eight quantitative traits. Nat Genet 41:527–534.
  17. 17. Ko KP, Min H, Ahn Y, Park SJ, Kim CS, et al. (2011) A prospective study investigating the association between environmental tobacco smoke exposure and the incidence of type 2 diabetes in never smokers. Ann Epidemiol 21:42–47.
  18. 18. Marchini J, Howie B, Myers S, McVean G, Donnelly P (2007) A new multipoint method for genome-wide association studies by imputation of genotypes. Nat Genet 39:906–913.
  19. 19. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, et al. (2006) Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet 38:904–909.
  20. 20. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, et al. (2007) PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81:559–575.
  21. 21. Boyle AP, Hong EL, Hariharan M, Cheng Y, Schaub MA, et al. (2012) Annotation of functional variation in personal genomes using RegulomeDB. Genome Res 22:1790–1797.
  22. 22. Yang J, Lee SH, Goddard ME, Visscher PM (2011) GCTA: a tool for genome-wide complex trait analysis. Am J Hum Genet 88:76–82.
  23. 23. Yang J, Benyamin B, McEvoy BP, Gordon S, Henders AK, et al. (2010) Common SNPs explain a large proportion of heritability for human height. Nat Genet 42:565–569.
  24. 24. Ioannidis JP, Patsopoulos NA, Evangelou E (2007) Heterogeneity in meta-analyses of genome-wide association investigations. PLoS ONE 2:e841.
  25. 25. Lemaitre RN, Tanaka T, Tang W, Manichaikul A, Foy M, et al. (2011) Genetic loci associated with plasma phospholipid n-3 fatty acids: a meta-analysis of genome-wide association studies from the CHARGE Consortium. PLoS Genet 7:e1002193.
  26. 26. Zabaneh D, Balding DJ (2010) A genome-wide association study of the metabolic syndrome in Indian Asian men. PLoS One 5:e11961.
  27. 27. Kröger J, Schulze MB (2012) Recent insights into the relation of Δ5 desatuase and Δ6 desaturase activity to the development of type 2 diabetes. Curr Opin Lipidol 23:4–10.
  28. 28. Schrimpe-Rutledge AC, Fontes G, Gritsenko MA, Norbeck AD, Anderson DJ, et al. (2012) Discovery of novel glucose regulated proteins in isolated human pancreatic islets using LC-MS/MS-based proteomics. J proteome Res 11:3520–3532.
  29. 29. Maglione V, Marchi P, DI Pardo A, Lingrell S, Horkey M, et al. (2010) Impaired ganglioside metabolism in Hungtington’s disease and neuroprotective role of GM1. J Neurosci 30:4072–4080.
  30. 30. Gillard BK, Thomas JW, Nell LJ, Marcus DM (1989) Antibodies against ganglioside GT3 in the sera of patients with type I Diabetes mellitus. J immunol 142:3826–3832.
  31. 31. Tagami S, Inokuchi Ji J, Kabayama K, Yoshimura H, Kitamura F, et al. (2002) Ganglioside GM3 participates in the pathological conditions of insulin resistance. J Biol Chem 277:3085–3092.
  32. 32. Van Meurs JB, Pare G, Schwartz SM, Hazra A, Tanaka T, et al. (2013) Common genetic loci influencing plasma homocystein concentrations and their effect on risk of coronary artery disease. Am J Clin Nutr 98:668–676.
  33. 33. Lippi G, Salvagno GL, Targher G, Montagnana M, Guidi GC (2008) Plasma gamma-glutamyl transferase activity predicts homocystein concentration in a large cohort of unselected outpatients. Intern Med 47:705–707.
  34. 34. Chasman DI, Pare G, Mora S, Hopewell JC, Peloso G, et al. (2009) Forty three loci associated with plasma lipoprotein size, concentration, and cholesterol content in genome-wide analysis. PLoS Genet 5:e1000730.
  35. 35. Wung SF, Kulkarni MV, Pullinger CR, Malloy MJ, Kane JP, et al. (2006) The lipoprotein lipase gene in combined hyperlipidemia: evidence of a protective allele depletion. Lipids Health Dis 5:19.
  36. 36. Hu Y, Ren Y, Luo RZ, Mao X, Li X, et al. (2007) Novel mutations of the lipoprotein lipase gene associated with hypertriglyceridemia in members of type 2 diabetic pedigrees. J Lipid Res 48:1681–1688.
  37. 37. The ENCODE Project Consortium (2012) An intergrated encyclopedia of DNA elements in the human genome. Nature 489:57–74.
  38. 38. The 1000 Genome Project Consortium (2012) An integrated map of genetic variation from 1092 human genomes. 491:56–65.
  39. 39. Kathiresan S, Willer CJ, Peloso GM, Demissie S, Musunuru K, et al. (2009) Common variants at 30 loci contribute to polygenic dyslipidemia. Nat Genet 41:56–65.
  40. 40. Musunuru K, Strong A, Frank-Kamenetsky M, Lee NE, Ahfeldt T, et al. (2010) From noncoding variant to phenotype via SORT1 at the 1p13 cholesterol locus. Nature 466:714–719.
  41. 41. Sung J, Lee K, Song YM (2009) Heritabilities of the metabolic syndrome phenotypes and related factors in Korean twins. J Clin Endocrinol Metab 94:4946–4952.
  42. 42. Jee SH, Yun JE, Nam CM, Suh I (2005) Heritability and segregation analysis of the level of LDL-Cholesterol. Korean Circ J 35:233–239.
  43. 43. Jee SH, Suh I, Won SY, Kim MY (2002) Familial correlation and heritability for cardiovascular risk factors. Yonsei Med J 43:160–164.
  44. 44. Cheng HH, Yang SH, Chen C, Chiang MS (1999) Correlation of serum lipids, uric acid, and albumin among mothers, offspring, and siblings in Taipei, Taiwan. Acta Paediatr Taiwan 40:225–232.
  45. 45. van Beek JH, de Moor MH, de Geus EJ, Lubke GH, Vink JM, et al. (2013) The genetic architecture of liver enzyme levels: GGT, ALT and AST. Behav Genet 43:329–339.