Genome-wide copy number variation analysis identified deletions in SFMBT1 associated with fasting plasma glucose in a Han Chinese population
© The Author(s). 2017
Received: 3 November 2016
Accepted: 31 July 2017
Published: 8 August 2017
Fasting glucose and fasting insulin are glycemic traits closely related to diabetes, and understanding the role of genetic factors in these traits can help reveal the etiology of type 2 diabetes. Although single nucleotide polymorphisms (SNPs) in several candidate genes have been found to be associated with fasting glucose and fasting insulin, copy number variations (CNVs), which have been reported to be associated with several complex traits, have not been reported for association with these two traits. We aimed to identify CNVs associated with fasting glucose and fasting insulin.
We conducted a genome-wide CNV association analysis for fasting plasma glucose (FPG) and fasting plasma insulin (FPI) using a family-based genome-wide association study sample from a Han Chinese population in Taiwan. A family-based CNV association test was developed in this study to identify common CNVs (i.e., CNVs with frequencies ≥ 5%), and a generalized estimating equation approach was used to test the associations between the traits and counts of global rare CNVs (i.e., CNVs with frequencies <5%). We found a significant genome-wide association for common deletions with a frequency of 5.2% in the Scm-like with four mbt domains 1 (SFMBT1) gene with FPG (association p-value = 2×10−4 and an adjusted p-value = 0.0478 for multiple testing). No significant association was observed between global rare CNVs and FPG or FPI. The deletions in 20 individuals with DNA samples available were successfully validated using PCR-based amplification. The association of the deletions in SFMBT1 with FPG was further evaluated using an independent population-based replication sample obtained from the Taiwan Biobank. An association p-value of 0.065, which was close to the significance level of 0.05, for FPG was obtained by testing 9 individuals with CNVs in the SFMBT1 gene region and 11,692 individuals with normal copies in the replication cohort.
Previous studies have found that SNPs in SFMBT1 are associated with blood pressure and serum urate concentration, suggesting that SFMBT1 may have functional implications in some metabolic-related traits.
Fasting glucose and fasting insulin are glycemic traits closely related to diabetes. Understanding the genetic factors associated with these traits can help identify pathways causing pathological glucose levels and type 2 diabetes [1, 2]. Heritability of fasting glucose and fasting insulin was estimated as 0.52 and 0.47, respectively, in families with hypertension , suggesting that genetic factors are responsible for a large proportion of phenotypic variation in the traits. Single nucleotide polymorphisms (SNPs) in several candidate genes have been identified to be associated with fasting glucose and fasting insulin [4–6]. However, the effect sizes for the SNPs are generally modest, and these SNPs explained only a small portion of heritability . Therefore, more causal genetic variants for fasting glucose and fasting insulin remain to be found.
Common and rare copy number variations (CNVs) have been shown to be associated with many complex traits [8–12], including some metabolic-related traits [13–16]. However, to our knowledge, associations between CNVs and fasting glucose and fasting insulin have not been reported in the literature. Several sophisticated CNV calling algorithms, such as PennCNV  and Birdsuite , based on SNP arrays have been developed to infer CNV states (i.e., deletion and duplication) with high accuracy. Therefore, genome-wide association study (GWAS) data that are mainly used to identify SNP associations have been used to infer CNVs, and associations of CNVs with complex diseases such as autism and schizophrenia have been discovered [19, 20] using GWAS.
To investigate the role of CNVs in fasting glucose and fasting insulin, in this study, we performed a genome-wide CNV association study for fasting plasma glucose (FPG) and fasting plasma insulin (FPI) based on a GWAS dataset from the Stanford Asia-Pacific Program for Hypertension and Insulin Resistance (SAPPHIRe) family study . A family-based CNV association test was developed to identify common CNVs (i.e., CNVs with frequencies ≥ 5%) associated with these traits. We also conducted simulation studies to evaluate the type I error rates and power for the family-based CNV association test in the present study. Furthermore, we performed a genome-wide burden test to investigate the associations of counts of global rare CNVs (i.e., CNVs with frequencies <5%) with FPG and FPI. The CNVs with genome-wide significant p-values were validated using PCR-based amplification. Moreover, we performed a replication analysis for the significant CNVs using another independent population-based cohort obtained from the Taiwan Biobank (https://www.twbiobank.org.tw).
The samples were collected from the SAPPHIRe family study. Individuals were recruited from five sites in Taiwan, Hawaii, and San Francisco. The sample consisted of both concordant sib pairs (both with hypertension) and discordant sib pairs (one with and one without hypertension) from the Chinese and Japanese populations. Subjects were recruited as probands if their age at onset for hypertension was between 35 and 60 years or if their age was >60 years but they had records of hypertension before 60 years. Subjects with pre-existing malignancies or major chronic diseases (such as type 2 diabetes or chronic liver, renal, and heart diseases) were excluded from the study. More details of the ascertainment criteria can be found in Wu et al. .
The samples were genotyped using the Affymetrix Genome-Wide Human SNP Array 6.0, which contains more than 1,878,000 probes. The samples were assigned randomly to batches of 96 samples for genotyping following the Affymetrix protocol. Genotypes were called using Affymetrix Power Tools (APT), which implements the Birdseed algorithm  for genotype calling. The Birdseed algorithm produces conventional genotype calls (i.e., three genotypes AA, AB, and BB), which were used in quality control (QC) procedures such as sex checks and Hardy-Weinberg Equilibrium (HWE) tests.
Studies have found that different CNV calling algorithms have advantages and disadvantages for different types of analyses [23, 24]. Therefore, we applied two commonly used CNV calling algorithms, Birdsuite and PennCNV, to generate CNV calls based on the signal intensity data from the SNP arrays. Then the consensus calls from the two algorithms were used in the following analyses. In Birdsuite, the samples were processed as batches of 96 samples to eliminate batch effects. The CNV segments reported by the Birdseye program, which is based on a Hidden Markov Model (HMM), in Birdsuite were used. PennCNV also detects CNVs based on HMM. All samples were processed together in PennCNV, as suggested in the user manual of PennCNV. The CNV calls generated by Birdsuite and PennCNV were classified into 3 states, which are deletion, normal, and duplication.
We applied a two-stage QC procedure. In stage 1, PLINK  was used to perform the QC based on the genotype calls generated by APT. SNPs with call rates <90%, minor allele frequencies <5%, or HWE test p-values <10−4 were excluded. The PLINK PI_HAT statistic, which is the proportion of loci that are identity-by-descent between a pair of individuals, was used to examine the relatedness among samples based on the SNP genotypes that passed QC. Samples that were reported as sib pairs but with PI_HAT <0.05 were removed. We also removed an individual if the median of PI_HAT of the individual with others was greater than 0.05. In stage 2, we followed the suggestions in the PennCNV manual to perform QC based on the CNV calls generated by Birdseye and PennCNV. Adjacent CNVs that were classified into the same state were merged into the same CNV if the length of the gaps (measured based on the number of probes) between them was less than 20% of the length of either one of the adjacent CNVs. CNVs containing less than 10 SNPs or that were smaller than 10 kb were removed. Spurious CNV calls in regions such as immunoglobulin, centromeric and telomeric regions were also removed. Samples with a standard deviation for the log ratio of observed probe intensity to expected intensity larger than 0.35 were removed, as suggested in the PennCNV manual. Samples with more than 100 CNV calls generated by PennCNV were removed. Because Birdsuite generated many more CNV calls than PennCNV, samples with more than 200 CNV calls generated by Birdseye were removed. After the QC steps were applied to the CNV calls generated by PennCNV and Birdseye separately, consensus calls were generated from the two sets of calls. A consensus call was defined as the intersection of CNV calls with the same state from the two algorithms.
The clinical measurements of the participants were taken at 8 am after an 8–10 h overnight fast. The glucose oxidase method on a Beckman Glucose Analyzer II (Beckman Instruments, Fullerton, CA, USA) was used to determine plasma glucose concentrations, and plasma insulin was measured using a commercial immunoradiometric kit (BioSource Europe, Nivelles, Belgium). The intra-assay and inter-assay coefficients of variation for glucose were 0.6% and 1.5%, respectively. The intra-assay and inter-assay coefficients of variation for insulin were 2.2 and 6.5%, respectively. Subjects diagnosed with diabetes were excluded from the study. Moreover, subjects with FPG levels >126 mg/dl were defined as having diabetes and were excluded.
Phenotypes were first adjusted for covariates such as age, sex, body mass index (BMI), ethnicity, and site. As samples were recruited based on the hypertension status, phenotypes were also adjusted for hypertension status as an additional covariate. Moreover, as a large cohort study suggested that genetic variants associated with BMI may also have associations with metabolic traits such as fasting glucose , adjusting for BMI may eliminate the effects of CNVs with pleiotropic effects on BMI and the two traits we studied. Therefore, the phenotypes were also adjusted for only age, sex, ethnicity, and site. A linear regression model using generalized estimating equations (GEEs) was fit for the trait and covariates with the “exchangeable” within cluster correlation structure to account for correlations among sibs. Ethnicity was considered as a binary variable with values of Chinese and Japanese ethnicities. Site was considered as a categorical variable consisting of nominal values for the five recruiting sites in SAPPHIRe. Residuals from the linear model were used as the adjusted phenotype values for subsequent analyses.
We developed a family-based association test to evaluate the associations between CNV calls and the phenotypes. The test statistic was the difference in the mean phenotypic value between an abnormal CNV state (i.e., deletion or duplication) and the normal state calculated based on the phenotypic values for siblings in all families. To evaluate the significance of the test statistic, we randomly permuted the phenotypic values for siblings within each family, and the permuted statistics were calculated over a large number of permutations (e.g., 5000). The p-value for the test was the proportion of the permuted statistics that were equal to or more extreme than the original statistic. A two-sided test was performed. The null hypothesis was that the CNV state is not associated with the phenotype. Because subjects can have CNVs with different lengths in the same region, we performed the test based on the locations of SNPs. The CNV state of a SNP for an individual was defined as the CNV state for the region where the SNP was located. To account for multiple testing, the permuted statistics were also used to calculate the permutation adjusted p-values and false discovery rate (FDR)  based on the formulas in Wang et al. . Note that there were correlations among SNPs if they were in the same CNV region. These correlations were properly considered when we calculated the permutation adjusted p-values and FDR because the correlation structures were maintained in the permuted statistics. Based on our power calculations shown in the Results section, the test maintained reasonable power for CNVs with frequencies ≥ 5% given the sample size of the study dataset. Therefore, we focused on testing CNVs with frequencies ≥ 5%.
As some studies have suggested that genome-wide rare CNVs are associated with complex traits, we performed a global burden analysis for CNVs with a frequency < 5%. PLINK was used to extract the CNVs with a frequency < 5% and calculate the number of CNVs across the genome for each individual. A regression analysis based on the GEEs was used to test the association between the trait and the CNV count, while family correlation was considered using the “exchangeable” within-cluster correlation structure in the GEE.
We performed a replication analysis using a population-based cohort from the Taiwan Biobank (TWB) for the CNVs passing the multiple testing threshold. The TWB has recruited more than 80,000 population-based samples with survey data such as basic demographic variables, lifestyle, and family history of common diseases, body measurements such as weight, height, and blood pressure, and blood and urine measurements such as fasting glucose and urinary microalbumin . A portion of the TWB samples were genotyped using customized Affymetrix Axiom chips for Han Chinese (referred to as the TWB chips), which consisted of 648,290 probes. The same QC procedures in stage 1 as described in the Quality control section were applied to the TWB sample. Because Birdsuite was not applicable to the customized chip data, only PennCNV was used to generate CNV calls. PennCNV was performed with the same procedures as in Kendall et al. , who generated CNV calls also based on customized Affymetrix Axiom chips for the UK Biobank data with PennCNV. More detailed descriptions of the procedures for generating CNV calls are provided in Additional file 1. A permutation test was also used to evaluate the significance of the CNVs with the trait. Phenotypes were first adjusted for covariates including age, sex, BMI, batches, and hypertension based on a linear regression model and the residuals were used for the association analysis. Similar to the family-based association test, the difference in the mean phenotypic value between an abnormal CNV state and the normal state was calculated as the test statistic. The trait values across all samples were randomly permuted, and the permuted statistics and the association p-value were calculated.
Summary statistics for the traits and covariates
91.36 ± 16.93 (444a)
7.78 ± 5.27 (442)
48.27 ± 8.46
Proportion of males
25.33 ± 3.42
13.32%, 14.67%, 36.79%, 34.76%, 0.46%
Chinese: 96.38%; Japanese: 3.62%
Association test results for the two traits
CNV association results with p-values <0.01 for FPG and FPI
Validation of the deletions in SFMBT1
Type I error and power study
We evaluated the type I error rate and power of the family-based CNV test for detecting the association of the deletions in SFMBT1 with FPG. To calculate the type I error rate, we randomly generated deletions in the family samples with a frequency of 5.2%, which was the same as the deletion frequency observed in the gene, while the trait values for the family samples remained the same. A total of 5000 replicates of the simulated family samples were generated to calculate the type I error rate. The estimated type I error rate was 0.052 with a 95% CI of (0.046, 0.058) at the significance level of 0.05, whereas the type I error rate was 0.008 with a 95% CI of (0.005, 0.011) at the significance level of 0.01. These results suggest that the type I error rates were properly maintained by the test with a CNV frequency of 5.2%. We then used a bootstrap procedure  to calculate the power. For each bootstrap, the same number of families as that of original samples was generated by sampling the original families with replacement, and the CNV test was applied to the bootstrapped samples. A total of 1000 bootstraps were performed, and the power was calculated as the proportion of test p-values less than the specified significance level in the 1000 tests. The power was estimated as 88.3% and 79.8% at the 0.05 and 0.01 significance levels, respectively. Therefore, given the trait values and sample size, this study had sufficient power to detect a CNV with frequency of 5.2% associated with the trait.
Summary statistics for the trait and covariates in the TWB sample
91.92 ± 7.58
47.47 ± 10.72
Proportion of males
24.01 ± 3.51
10.80%, 12.34%, 14.14%, 10.79%, 13.24%, 12.11%, 10.35%, 16.23%
Proportion of hypertension
Our analysis identified a candidate region of deletions in SFMBT1 (chr3:53,003,415–53,013,826) significantly lowered FPG level in the SAPPHIRe sample, with a genome-wide significant p-value of 2 × 10−4. Interestingly, the same trend was also observed in the replication cohort (i.e., the TWB cohort) that samples with deletions had lower mean FPG level than the mean FPG level in samples with normal copies. Due to the restriction of the genotyping platform in TWB, only 9 individuals with larger CNVs covering the candidate SFMBT1 region were identified. However, the association p-value of 0.065 was close to the 0.05 significance level, supporting that the deletions in SFMBT1 have effects on FPG.
The SFMBT1 gene encodes a protein containing four malignant brain tumor repeat domains. Interestingly, SNPs in SFMBT1 have been reported to be associated with mean systolic and diastolic blood pressure, and significantly differential expression was observed for the gene between hypertensive cases and normal controls in another Han Chinese study in Taiwan . A large GWAS based on >140,000 samples with European ancestry identified that SNPs in the gene are significantly associated with serum urate concentrations . Another study found that uric acid levels are positively associated with FPG , and some candidate genes for uric acid have been found to be associated with FPG in a Chinese population . Hence, SFMBT1 may have functional implications in some metabolic related traits. As shown in Fig. 2, deletions in SFMBT1 were also found in Database of Genomic Variants (DGV)  and the CNV Discovery Project, which aimed to identify common CNVs , suggesting that deletions are common in this gene.
No duplications in the candidate SFMBT1 region were observed in the SAPPHIRe sample, and only one duplication was observed in the TWB sample. A total of 30 CNVs in the gene were found in the DGV based on results from various studies, where Caucasian samples were mainly analyzed. Only two of the 30 CNVs were duplications, while the others were deletions. Therefore, duplications in SFMBT1 could also be rare in the Han Chinese population. Further studies to evaluate whether duplications in SFMBT1 elevate fasting glucose levels in the Han Chinese population will be important. However, a large sample size with dense probes will be required to achieve the goal.
Although rare CNVs have been found to be associated with several complex traits, our burden analysis did not identify any significant associations between global rare CNVs and the two traits. This may be due to the limited size of our sample, where many rare CNVs were not observed. Again, a large sample size will be required to further evaluate the role of global rare CNVs in FPG and FPI.
We identified deletions in SFMBT1 that were significantly associated with FPG in the SAPPHIRe sample, and the deletions also showed marginal significance in the TWB sample. The deletions in the SAPPHIRe sample were validated using PCR-based amplification. Based on previous findings and our results, SFMBT1 may have functional implications in FPG and other metabolic traits. Our power study suggest that the proposed family-based CNV test had sufficient power to identify the deletions associated with FPG given the sample size. Further studies should be conducted to evaluate the role of duplications in the SFMBT1 gene and FPG.
We thank the participants in the SAPPHIRe study.
This study was supported by grants BS-090-PP-01, BS-091-PP-01, BS-092-PP-01, BS-093-PP-01, and BS-094-PP-01 from the National Health Research Institutes in Taiwan.
Availability of data and materials
The data that support the findings of this study are available on request from the corresponding author CAH. The data are not publicly available due to them containing information that could compromise research participant privacy/consent.
RHC, YFC, and CAH formulated the research goal. RHC and YFC designed the methods and performed the analyses. YJH, WJL, KDW, MWL, YDIC, TQ, and CAH contributed materials and analysis tools. HLC performed the CNV validation experiment. RHC, YFC, and CAH wrote the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Written informed consent for participation in the SAPPHIRe study was obtained from all participants. Research in the SAPPHIRe study was approved by the Institutional Review Board of the National Health Research Institutes in Taiwan.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Wheeler E, Barroso I. Genome-wide association studies and type 2 diabetes. Brief Funct Genomics. 2011;10(2):52–60.View ArticlePubMedGoogle Scholar
- Scott RA, Lagou V, Welch RP, Wheeler E, Montasser ME, Luan J, Magi R, Strawbridge RJ, Rehnberg E, Gustafsson S, et al. Large-scale association analyses identify new loci influencing glycemic traits and provide insight into the underlying biological pathways. Nat Genet. 2012;44(9):991–1005.View ArticlePubMedPubMed CentralGoogle Scholar
- Freedman BI, Rich SS, Sale MM, Heiss G, Djousse L, Pankow JS, Province MA, Rao DC, Lewis CE, Chen YD, et al. Genome-wide scans for heritability of fasting serum insulin and glucose concentrations in hypertensive families. Diabetologia. 2005;48(4):661–8.View ArticlePubMedGoogle Scholar
- Dupuis J, Langenberg C, Prokopenko I, Saxena R, Soranzo N, Jackson AU, Wheeler E, Glazer NL, Bouatia-Naji N, Gloyn AL, et al. New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat Genet. 2010;42(2):105–16.View ArticlePubMedPubMed CentralGoogle Scholar
- Fesinmeyer MD, Meigs JB, North KE, Schumacher FR, Buzkova P, Franceschini N, Haessler J, Goodloe R, Spencer KL, Voruganti VS, et al. Genetic variants associated with fasting glucose and insulin concentrations in an ethnically diverse population: results from the Population Architecture using Genomics and Epidemiology (PAGE) study. BMC Med Genet. 2013;14:98.View ArticlePubMedPubMed CentralGoogle Scholar
- Prokopenko I, Langenberg C, Florez JC, Saxena R, Soranzo N, Thorleifsson G, Loos RJ, Manning AK, Jackson AU, Aulchenko Y, et al. Variants in MTNR1B influence fasting glucose levels. Nat Genet. 2009;41(1):77–81.View ArticlePubMedGoogle Scholar
- Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Hunter DJ, McCarthy MI, Ramos EM, Cardon LR, Chakravarti A, et al. Finding the missing heritability of complex diseases. Nature. 2009;461(7265):747–53.View ArticlePubMedPubMed CentralGoogle Scholar
- Pinto D, Pagnamenta AT, Klei L, Anney R, Merico D, Regan R, Conroy J, Magalhaes TR, Correia C, Abrahams BS, et al. Functional impact of global rare copy number variation in autism spectrum disorders. Nature. 2010;466(7304):368–72.View ArticlePubMedPubMed CentralGoogle Scholar
- Soemedi R, Wilson IJ, Bentham J, Darlay R, Topf A, Zelenika D, Cosgrove C, Setchfield K, Thornborough C, Granados-Riveron J, et al. Contribution of global rare copy-number variants to the risk of sporadic congenital heart disease. Am J Hum Genet. 2012;91(3):489–501.View ArticlePubMedPubMed CentralGoogle Scholar
- Dauber A, Yu Y, Turchin MC, Chiang CW, Meng YA, Demerath EW, Patel SR, Rich SS, Rotter JI, Schreiner PJ, et al. Genome-wide association of copy-number variation reveals an association between short stature and the presence of low-frequency genomic deletions. Am J Hum Genet. 2011;89(6):751–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Lanktree MB, Rajakumar C, Brunt JH, Koschinsky ML, Connelly PW, Hegele RA. Determination of lipoprotein(a) kringle repeat number from genomic DNA: copy number variation genotyping using qPCR. J Lipid Res. 2009;50(4):768–72.View ArticlePubMedPubMed CentralGoogle Scholar
- Craddock N, Hurles ME, Cardin N, Pearson RD, Plagnol V, Robson S, Vukcevic D, Barnes C, Conrad DF, Giannoulatou E, et al. Genome-wide association study of CNVs in 16,000 cases of eight common diseases and 3,000 shared controls. Nature. 2010;464(7289):713–20.View ArticlePubMedGoogle Scholar
- Speliotes EK, Willer CJ, Berndt SI, Monda KL, Thorleifsson G, Jackson AU, Lango Allen H, Lindgren CM, Luan J, Magi R, et al. Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat Genet. 2010;42(11):937–48.View ArticlePubMedPubMed CentralGoogle Scholar
- Jeon JP, Shim SM, Nam HY, Ryu GM, Hong EJ, Kim HL, Han BG. Copy number variation at leptin receptor gene locus associated with metabolic traits and the risk of type 2 diabetes mellitus. BMC Genomics. 2010;11:426.View ArticlePubMedPubMed CentralGoogle Scholar
- Lacaria M, Saha P, Potocki L, Bi W, Yan J, Girirajan S, Burns B, Elsea S, Walz K, Chan L, et al. A duplication CNV that conveys traits reciprocal to metabolic syndrome and protects against diet-induced obesity in mice and men. PLoS Genet. 2012;8(5):e1002713.View ArticlePubMedPubMed CentralGoogle Scholar
- Irvin MR, Wineinger NE, Rice TK, Pajewski NM, Kabagambe EK, Gu CC, Pankow J, North KE, Wilk JB, Freedman BI, et al. Genome-wide detection of allele specific copy number variation associated with insulin resistance in African Americans from the HyperGEN study. PLoS One. 2011;6(8):e24052.View ArticlePubMedPubMed CentralGoogle Scholar
- Wang K, Li M, Hadley D, Liu R, Glessner J, Grant SF, Hakonarson H, Bucan M. PennCNV: an integrated hidden Markov model designed for high-resolution copy number variation detection in whole-genome SNP genotyping data. Genome Res. 2007;17(11):1665–74.View ArticlePubMedPubMed CentralGoogle Scholar
- Korn JM, Kuruvilla FG, McCarroll SA, Wysoker A, Nemesh J, Cawley S, Hubbell E, Veitch J, Collins PJ, Darvishi K, et al. Integrated genotype calling and association analysis of SNPs, common copy number polymorphisms and rare CNVs. Nat Genet. 2008;40(10):1253–60.View ArticlePubMedPubMed CentralGoogle Scholar
- Weiss LA, Arking DE. Gene Discovery Project of Johns H, the Autism C, Daly MJ, Chakravarti A: A genome-wide linkage and association scan reveals novel loci for autism. Nature. 2009;461(7265):802–8.View ArticlePubMedPubMed CentralGoogle Scholar
- International Schizophrenia C, Purcell SM, Wray NR, Stone JL, Visscher PM, O'Donovan MC, Sullivan PF, Sklar P. Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature. 2009;460(7256):748–52.Google Scholar
- Ranade K, Hsuing AC, Wu KD, Chang MS, Chen YT, Hebert J, Chen YI, Olshen R, Curb D, Dzau V, et al. Lack of evidence for an association between alpha-adducin and blood pressure regulation in Asian populations. Am J Hypertens. 2000;13(6 Pt 1):704–9.View ArticlePubMedGoogle Scholar
- Wu KD, Hsiao CF, Ho LT, Sheu WH, Pei D, Chuang LM, Curb D, Chen YD, Tsai HJ, Dzau VJ, et al. Clustering and heritability of insulin resistance in Chinese and Japanese hypertensive families: a Stanford-Asian Pacific Program in Hypertension and Insulin Resistance sibling study. Hypertens Res. 2002;25(4):529–36.View ArticlePubMedGoogle Scholar
- Zhang X, Du R, Li S, Zhang F, Jin L, Wang H. Evaluation of copy number variation detection for a SNP array platform. BMC Bioinf. 2014;15:50.View ArticleGoogle Scholar
- Zhang D, Qian Y, Akula N, Alliey-Rodriguez N, Tang J, Bipolar Genome S, Gershon ES, Liu C. Accuracy of CNV Detection from GWAS Data. PLoS One. 2011;6(1):e14511.View ArticlePubMedPubMed CentralGoogle Scholar
- Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Maller J, Sklar P, de Bakker PI, Daly MJ, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81(3):559–75.View ArticlePubMedPubMed CentralGoogle Scholar
- van Vliet-Ostaptchouk JV, den Hoed M, Luan J, Zhao JH, Ong KK, van der Most PJ, Wong A, Hardy R, Kuh D, van der Klauw MM, et al. Pleiotropic effects of obesity-susceptibility loci on metabolic traits: a meta-analysis of up to 37,874 individuals. Diabetologia. 2013;56(10):2134–46.View ArticlePubMedGoogle Scholar
- Reiner A, Yekutieli D, Benjamini Y. Identifying differentially expressed genes using false discovery rate controlling procedures. Bioinformatics. 2003;19(3):368–75.View ArticlePubMedGoogle Scholar
- Wang K, Li M, Bucan M. Pathway-based approaches for analysis of genomewide association studies. Am J Hum Genet. 2007;81(6):1278–83.View ArticlePubMedPubMed CentralGoogle Scholar
- Fan CT, Lin JC, Lee CH. Taiwan Biobank: a project aiming to aid Taiwan's transition into a biomedical island. Pharmacogenomics. 2008;9(2):235–46.View ArticlePubMedGoogle Scholar
- Kendall KM, Rees E, Escott-Price V, Einon M, Thomas R, Hewitt J, O'Donovan MC, Owen MJ, Walters JT, Kirov G. Cognitive Performance Among Carriers of Pathogenic Copy Number Variants: Analysis of 152,000 UK Biobank Subjects. Biol Psychiatry. 2016;Google Scholar
- Westfall PH, Young SS. Resampling-Based Multiple Testing. New York: John Wiley & Sons; 1993.Google Scholar
- Yang HC, Liang YJ, Chen JW, Chiang KM, Chung CM, Ho HY, Ting CT, Lin TH, Sheu SH, Tsai WC, et al. Identification of IGF1, SLC4A4, WWOX, and SFMBT1 as hypertension susceptibility genes in Han Chinese with a genome-wide gene-based association study. PLoS One. 2012;7(3):e32907.View ArticlePubMedPubMed CentralGoogle Scholar
- Kottgen A, Albrecht E, Teumer A, Vitart V, Krumsiek J, Hundertmark C, Pistis G, Ruggiero D, O'Seaghdha CM, Haller T, et al. Genome-wide association analyses identify 18 new loci associated with serum urate concentrations. Nat Genet. 2013;45(2):145–54.View ArticlePubMedGoogle Scholar
- Modan M, Halkin H, Karasik A, Lusky A. Elevated serum uric acid--a facet of hyperinsulinaemia. Diabetologia. 1987;30(9):713–8.View ArticlePubMedGoogle Scholar
- Sun X, Zhang R, Jiang F, Tang S, Chen M, Peng D, Yan J, Wang T, Wang S, Bao Y, et al. Common variants related to serum uric acid concentrations are associated with glucose metabolism and insulin secretion in a Chinese population. PLoS One. 2015;10(1):e0116714.View ArticlePubMedPubMed CentralGoogle Scholar
- MacDonald JR, Ziman R, Yuen RK, Feuk L, Scherer SW. The Database of Genomic Variants: a curated collection of structural variation in the human genome. Nucleic Acids Res. 2014;42(Database issue):D986–92.View ArticlePubMedGoogle Scholar
- Conrad DF, Pinto D, Redon R, Feuk L, Gokcumen O, Zhang Y, Aerts J, Andrews TD, Barnes C, Campbell P, et al. Origins and functional impact of copy number variation in the human genome. Nature. 2010;464(7289):704–12.View ArticlePubMedGoogle Scholar