Rare genetic variant analysis on blood pressure in related samples
© Chen et al.; licensee BioMed Central Ltd. 2014
Published: 17 June 2014
The genetic variants associated with blood pressure identified so far explain only a small proportion of the total heritability of this trait. With recent advances in sequencing technology and statistical methodology, it becomes feasible to study the association between blood pressure and rare genetic variants. Using real baseline phenotype data and imputed dosage data from Genetic Analysis Workshop 18, we performed a candidate gene association analysis. We focused on 8 genes shown to be associated with either systolic or diastolic blood pressure to identify the association with both common and rare genetic variants, and then did a genome-wide rare-variant analysis on blood pressure. We performed association analysis for rare coding and splicing variants within each gene region and all rare variants in each sliding window, using either burden tests or sequence kernel association tests accounting for familial correlation. With a sample size of only 747, we failed to find any novel associated genetic loci. Consequently, we performed analyses on simulated data, with knowledge of the underlying simulating model, to evaluate the type I error rate and power for the methods used in real data analysis.
Despite the tremendous success of genome-wide association studies (GWAS) to uncover genetic variants influencing complex traits and diseases, only a fraction of the total heritability of these traits is explained by the loci identified so far. Because GWAS focuses on common variants, a possible source of the missing heritability might be rare variants that were not included in the earlier genotyping platforms. The next logical step is to investigate rare variants, an endeavor that is now possible because of the ever-decreasing cost of sequencing.
Whole genome sequencing has the ability to uncover rare variants, but brings its own challenges. Despite a low error rate, the sheer number of base pairs sequenced makes it hard to distinguish very rare mutations from sequencing errors. Moreover, detecting association with rare variants requires very large sample sizes. Several methods to jointly analyze rare variants within a genomic region have been developed, however, and these methods have the potential to pinpoint additional variants contributing to the overall heritability of traits.
Blood pressure (BP) and hypertension are prime examples of the limitations of GWAS. Meta-analysis of GWAS from a large number of cohorts has identified multiple genetic loci over the genome that affect systolic blood pressure (SBP), diastolic blood pressure (DBP), hypertension, or a combination of these traits [1, 2]. However, the loci identified to date explain only a small portion of the total heritability in BP.
In this article, we investigate the association of rare variants in genomic regions that have been previously implicated by GWAS to identify the source of the original GWAS signal and to discover additional genetic loci influencing BP using either burden tests adjusting for familial correlation (famBT) or sequence kernel association tests (SKAT)  for family samples (famSKAT) [4, 5]. We also analyze rare variants genome-wide to uncover additional genomic regions harboring susceptibility variants. Finally, we use the simulated data sets with knowledge of the answer to evaluate type I error and power for famBT and famSKAT in family samples.
Characteristics of variables of interest at the baseline
For the BP candidate gene study, we performed both common and rare-variant analysis. Common variants were defined as any variants with minor allele frequency (MAF) >5% in our subset of participants, and rare variants were variants with MAF between 0% and 5%. We performed common variant analysis as single-marker association tests using linear mixed-effect models  to account for familial correlation and reported the most significant SNP in each region. We performed rare variant analysis for all rare variants within each gene region with famBT and famSKAT [4, 5], using Wu weights, which is a beta distribution probability density function of the MAF with parameters 1 and 25 . Both rare-variant approaches are described below.
Burden tests adjusting for familial correlation (famBT)
where β is the effect size for the combined genotype score, γ is the random effect vector for familial correlation, and ε is the normally distributed error. We assume , , where and are variance component parameters, and Φ is twice the kinship matrix. The model can be fitted as a linear mixed-effect model and the genotype effect can be tested as H0 : β = 0 versus H1 : β ≠ 0 in this framework.
SKAT for family (famSKAT)
where W is a diagonal matrix of weights w j , and we assume β ~N (0, τI q ). The genotype effects can be tested as H0 : τ= 0 versus H1 : τ> 0.
For the genome-wide rare-variant analysis on real data, we performed famBT and famSKAT, using both a gene-based coding and splicing variants analysis (GB) and sliding-window analysis (SW). GB was performed for each gene, using only nonsynonymous rare variants and rare variants at the splicing sites. SW was performed for all rare variants in each genomic region of 4000 base pairs (bp) length, with 2000 bp each overlapping with the previous and subsequent windows, regardless of the gene annotation.
In addition to real data analysis, we also performed rare-variant analysis on simulated data sets, with knowledge of the underlying simulating model. To be consistent with the real data analyses, we adjusted for sex, age, and smoking in all analyses, even though simulated smoking is not associated with simulated SBP or DBP. Because we did not have missing data in the simulated data sets, we took the first exam as the baseline and excluded individuals taking antihypertensive medication at baseline. Therefore, the sample size varies slightly in different simulation replicates. We used both famBT and famSKAT for GB and SW, but we analyzed only chromosome 3 because of limited computing resources. We evaluated the type I errors of these approaches using quantitative trait Q1, which was a simulated trait not associated with any genetic variants. We also calculated the empirical power for MAP4 on DBP and rSBP.
Candidate gene analysis
Candidate gene analysis results
Previous GWAS results
2.1 × 10−7
2.0 × 10−13
7.8 × 10−7
5.8 × 10−7
2.8 × 10−7
1.0 × 10−23
8.0 × 10−7
1.0 × 10−8
5.0 × 10−9
Genome-wide rare-variant analysis
Genome-wide rare-variant analysis top findings
1.8 × 10−5
1.1 × 10−4
1.3 × 10−4
1.6 × 10−4
4.2 × 10−4
2.1 × 10−4
6.5 × 10−5
8.8 × 10−5
7.3 × 10−5
6.9 × 10−4
3.5 × 10−4
1.1 × 10−3
1.6 × 10−6
3.8 × 10−6
1.9 × 10−6
7.1 × 10−6
3.6 × 10−6
1.1 × 10−5
1.6 × 10−6
3.0 × 10−6
1.6 × 10−6
1.9 × 10−5
2.6 × 10−6
1.6 × 10−5
Empirical type I errors from simulation data sets
Empirical power from simulation data sets
2.5 × 10−6
3.3 × 10−8
2.5 × 10−6
3.3 × 10−8
MAP4 encodes microtubule-associated protein 4. This gene is located in chromosome 3p21. The SNPs within the gene region have previously shown a genome-wide association with mean arterial pressure. The top-ranking SNP (rs319690) yields a p value of 2.69 × 10−8 . MAP4 microtubule decoration restricts with beta-adrenergic receptor recycling, which might explain beta-adrenergic receptor downregulation in heart failure .
It is not surprising that in gene-based analysis, famBT is slightly more powerful than famSKAT because, among the 6 rare coding variants in MAP4--3_47894286, 3_47913455, 3_47957741, 3_47957996, 3_48040283, and 3_48040284--5 were simulated to be negatively associated with both SBP and DBP. The last SNP, 3_47894286, has perfect linkage disequilibrium (r = 1) with 3_47913455. In such a simulation setting, with SNPs all having the same direction of effect, the burden test should outperform most statistical approaches.
In sliding-window analysis, however, even though MAP4 is the gene most significantly associated with both SBP and DBP, some rare regulatory variants were simulated to be positively associated with the traits. As a result, famBT has almost no power to detect the association in this gene region. In contrast, famSKAT performs very well because SKAT allows effects to be in different directions. After adjusting for multiple testing, famSKAT still attains good power even at low α levels.
The SW method is more computationally intensive than GB because more tests are performed. However, by using SW we can generally test all possible rare variants associated with the trait, no matter where they are located. In many scenarios, intergenic variants, especially those within regulatory regions, also may be associated with quantitative traits. Thus, for rare-variant analysis on real data, unless we have strong a priori knowledge that the associated variants are nonsynonymous, we would recommend running a sliding-window analysis. By using famSKAT, we can perform rare-variant analysis on family data and have much better power than simple burden tests when there are variants with both positive and negative effects.
This research was conducted using the Linux Clusters for Genetic Analysis (LinGA) computing resources at Boston University Medical Campus. The GAW18 whole genome sequence data were provided by the T2D-GENES Consortium, which is supported by NIH grants U01 DK085524, U01 DK085584, U01 DK085501, U01 DK085526, and U01 DK085545. The other genetic and phenotypic data for GAW18 were provided by the San Antonio Family Heart Study and San Antonio Family Diabetes/Gallbladder Study, which are supported by NIH grants P01 HL045222, R01 DK047482, and R01 DK053889. The Genetic Analysis Workshop is supported by NIH grant R01 GM031575.
This article has been published as part of BMC Proceedings Volume 8 Supplement 1, 2014: Genetic Analysis Workshop 18. The full contents of the supplement are available online at http://www.biomedcentral.com/bmcproc/supplements/8/S1. Publication charges for this supplement were funded by the Texas Biomedical Research Institute.
- Newton-Cheh C, Johnson T, Gateva V, Tobin MD, Bochud M, Coin L, Najjar SS, Zhao JH, Heath SC, Eyheramendy S, et al: Genome-wide association study identifies eight loci associated with blood pressure. Nat Genet. 2009, 41: 666-676. 10.1038/ng.361.PubMed CentralView ArticlePubMedGoogle Scholar
- Levy D, Ehret GB, Rice K, Verwoert GC, Launer LJ, Dehghan A, Glazer NL, Morrison AC, Johnson AD, Aspelund T, et al: Genome-wide association study of blood pressure and hypertension. Nat Genet. 2009, 41: 677-687. 10.1038/ng.384.PubMed CentralView ArticlePubMedGoogle Scholar
- Wu MC, Lee S, Cai T, Li Y, Boehnke M, Lin X: Rare-variant association testing for sequencing data with the sequence kernel association test. Am J Hum Genet. 2011, 89: 82-93. 10.1016/j.ajhg.2011.05.029.PubMed CentralView ArticlePubMedGoogle Scholar
- Schifano ED, Epstein MP, Bielak LF, Jhun MA, Kardia SL, Peyser PA, Lin X: SNP set association analysis for familial data. Genet Epidemiol. 2012, 36: 797-810.PubMed CentralPubMedGoogle Scholar
- Chen H, Meigs JB, Dupuis J: Sequence kernel association test for quantitative traits in family samples. Genet Epidemiol. 2013, 37: 196-204. 10.1002/gepi.21703.PubMed CentralView ArticlePubMedGoogle Scholar
- Pankratz VS, de Andrade M, Therneau TM: Random-effects Cox proportional hazard model: general variance components methods for time-to-event data. Genet Epidemiol. 2005, 28: 97-109. 10.1002/gepi.20043.View ArticlePubMedGoogle Scholar
- Wain LV, Verwoert GC, O'Reilly PF, Shi G, Johnson T, Johnson AD, Bochud M, Rice KM, Henneman P, Smith AV, et al: Genome-wide association study identifies six new loci influencing pulse pressure and mean arterial pressure. Nat Genet. 2011, 43: 1005-1011. 10.1038/ng.922.PubMed CentralView ArticlePubMedGoogle Scholar
- Cheng G, Qiao F, Gallien TN, Kuppuswamy D, Cooper G IV: Inhibition of β-adrenergic receptor trafficking in adult cardiocytes by MAP4 decoration of microtubules. Am J Physiol Heart Circ Physiol. 2005, 288: H1193-H1202.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 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.