Interpretation of simultaneous linkage and family-based association tests in genome screens

Sequencing Analysis at 8p23 Identifies Multiple Rare Variants in DLC1 Associated with Sleep-Related Oxyhemoglobin Saturation Level

Average arterial oxyhemoglobin saturation during sleep (AvSpO2S) is a clinically relevant measure of physiological stress associated with sleep-disordered breathing, and this measure predicts incident cardiovascular disease and mortality. Using high-depth whole-genome sequencing data from the National Heart, Lung, and Blood Institute (NHLBI) Trans-Omics for Precision Medicine (TOPMed) project and focusing on genes with linkage evidence on chromosome 8p23,1,2 we observed that six coding and 51 noncoding variants in a gene that encodes the GTPase-activating protein (DLC1) are significantly associated with AvSpO2S and replicated in independent subjects. The combined DLC1 association evidence of discovery and replication cohorts reaches genome-wide significance in European Americans (p = 7.9 × 10-7). A risk score for these variants, built on an independent dataset, explains 0.97% of the AvSpO2S variation and contributes to the linkage evidence. The 51 noncoding variants are enriched in regulatory features in a human lung fibroblast cell line and contribute to DLC1 expression variation. Mendelian randomization analysis using these variants indicates a significant causal effect of DLC1 expression in fibroblasts on AvSpO2S. Multiple sources of information, including genetic variants, gene expression, and methylation, consistently suggest that DLC1 is a gene associated with AvSpO2S.

A Major Locus for Quantitatively Measured Shank Skin Color Traits in Korean Native Chicken

Shank skin color of Korean native chicken (KNC) shows large color variations. It varies from white, yellow, green, bluish or grey to black, whilst in the majority of European breeds the shanks are typically yellow-colored. Three shank skin color-related traits (i.e., lightness [L*], redness [a*], and yellowness [b*]) were measured by a spectrophotometer in 585 progeny from 68 nuclear families in the KNC resource population. We performed genome scan linkage analysis to identify loci that affect quantitatively measured shank skin color traits in KNC. All these birds were genotyped with 167 DNA markers located throughout the 26 autosomes. The SOLAR program was used to conduct multipoint variance-component quantitative trait locus (QTL) analyses. We detected a major QTL that affects b* value (logarithm of odds [LOD] = 47.5, p = 1.60×10⁻⁴⁹) on GGA24 (GGA for Gallus gallus). At the same location, we also detected a QTL that influences a* value (LOD = 14.2, p = 6.14×10⁻¹⁶). Additionally, beta-carotene dioxygenase 2 (BCDO2), the obvious positional candidate gene under the linkage peaks on GGA24, was investigated by the two association tests: i.e., measured genotype association (MGA) and quantitative transmission disequilibrium test (QTDT). Significant associations were detected between BCDO2 g.9367 A>C and a* (P_MGA = 1.69×10⁻²⁸; P_QTDT = 2.40×10⁻²⁵). The strongest associations were between BCDO2 g.9367 A>C and b* (P_MGA = 3.56×10⁻⁶⁶; P_QTDT = 1.68×10⁻⁶⁵). However, linkage analyses conditional on the single nucleotide polymorphism indicated that other functional variants should exist. Taken together, we demonstrate for the first time the linkage and association between the BCDO2 locus on GGA24 and quantitatively measured shank skin color traits in KNC.

FamPipe: An Automatic Analysis Pipeline for Analyzing Sequencing Data in Families for Disease Studies

In disease studies, family-based designs have become an attractive approach to analyzing next-generation sequencing (NGS) data for the identification of rare mutations enriched in families. Substantial research effort has been devoted to developing pipelines for automating sequence alignment, variant calling, and annotation. However, fewer pipelines have been designed specifically for disease studies. Most of the current analysis pipelines for family-based disease studies using NGS data focus on a specific function, such as identifying variants with Mendelian inheritance or identifying shared chromosomal regions among affected family members. Consequently, some other useful family-based analysis tools, such as imputation, linkage, and association tools, have yet to be integrated and automated. We developed FamPipe, a comprehensive analysis pipeline, which includes several family-specific analysis modules, including the identification of shared chromosomal regions among affected family members, prioritizing variants assuming a disease model, imputation of untyped variants, and linkage and association tests. We used simulation studies to compare properties of some modules implemented in FamPipe, and based on the results, we provided suggestions for the selection of modules to achieve an optimal analysis strategy. The pipeline is under the GNU GPL License and can be downloaded for free at http://fampipe.sourceforge.net.

Gene-smoking interactions in multiple Rho-GTPase pathway genes in an early-onset coronary artery disease cohort

We performed a gene-smoking interaction analysis using families from an early-onset coronary artery disease cohort (GENECARD). This analysis was focused on validating and expanding results from previous studies implicating single nucleotide polymorphisms (SNPs) on chromosome 3 in smoking-mediated coronary artery disease. We analyzed 430 SNPs on chromosome 3 and identified 16 SNPs that showed a gene-smoking interaction at P < 0.05 using association in the presence of linkage--ordered subset analysis, a method that uses permutations of the data to empirically estimate the strength of the association signal. Seven of the 16 SNPs were in the Rho-GTPase pathway indicating a 1.87-fold enrichment for this pathway. A meta-analysis of gene-smoking interactions in three independent studies revealed that rs9289231 in KALRN had a Fisher's combined P value of 0.0017 for the interaction with smoking. In a gene-based meta-analysis KALRN had a P value of 0.026. Finally, a pathway-based analysis of the association results using WebGestalt revealed several enriched pathways including the regulation of the actin cytoskeleton pathway as defined by the Kyoto Encyclopedia of Genes and Genomes.

A novel method, the Variant Impact On Linkage Effect Test (VIOLET), leads to improved identification of causal variants in linkage regions

The Human Genome Project was expected to individualize medicine by rapidly advancing knowledge of common complex disease through discovery of disease-causing genetic variants. However, this has proved challenging. Although linkage analysis has identified replicated chromosomal regions, subsequent detection of causal variants for complex traits has been limited. One explanation for this difficulty is that utilization of association to follow up linkage is problematic given that linkage and association are not required to co-occur. Indeed, co-occurrence is likely to occur only in special circumstances, such as Mendelian inheritance, but cannot be universally expected. To overcome this problem, we propose a novel method, the Variant Impact On Linkage Effect Test (VIOLET), which differs from other quantitative methods in that it is designed to follow up linkage by identifying variants that influence the variance explained by a quantitative trait locus. VIOLET's performance was compared with measured genotype and combined linkage association in two data sets with quantitative traits. Using simulated data, VIOLET had high power to detect the causal variant and reduced false positives compared with standard methods. Using real data, VIOLET identified a single variant, which explained 24% of linkage; this variant exhibited only nominal association (P=0.04) using measured genotype and was not identified by combined linkage association. These results demonstrate that VIOLET is highly specific while retaining low false-negative results. In summary, VIOLET overcomes a barrier to gene discovery and thus may be broadly applicable to identify underlying genetic etiology for traits exhibiting linkage.

Accounting for a quantitative trait locus for plasma triglyceride levels: utilization of variants in multiple genes

Background

For decades, research efforts have tried to uncover the underlying genetic basis of human susceptibility to a variety of diseases. Linkage studies have resulted in highly replicated findings and helped identify quantitative trait loci (QTL) for many complex traits; however identification of specific alleles accounting for linkage remains elusive. The purpose of this study was to determine whether with a sufficient number of variants a linkage signal can be fully explained.

Method

We used comprehensive fine-mapping using a dense set of single nucleotide polymorphisms (SNPs) across the entire quantitative trait locus (QTL) on human chromosome 7q36 linked to plasma triglyceride levels. Analyses included measured genotype and combined linkage association analyses.

Results

Screening this linkage region, we found an over representation of nominally significant associations in five genes (MLL3, DPP6, PAXIP1, HTR5A, INSIG1). However, no single genetic variant was sufficient to account for the linkage. On the other hand, multiple variants capturing the variation in these five genes did account for the linkage at this locus. Permutation analyses suggested that this reduction in LOD score was unlikely to have occurred by chance (p = 0.008).

Discussion

With recent findings, it has become clear that most complex traits are influenced by a large number of genetic variants each contributing only a small percentage to the overall phenotype. We found that with a sufficient number of variants, the linkage can be fully explained. The results from this analysis suggest that perhaps the failure to identify causal variants for linkage peaks may be due to multiple variants under the linkage peak with small individual effect, rather than a single variant of large effect.

Fine mapping of a linkage peak with integration of lipid traits identifies novel coronary artery disease genes on chromosome 5

Background

Coronary artery disease (CAD), and one of its intermediate risk factors, dyslipidemia, possess a demonstrable genetic component, although the genetic architecture is incompletely defined. We previously reported a linkage peak on chromosome 5q31-33 for early-onset CAD where the strength of evidence for linkage was increased in families with higher mean low density lipoprotein-cholesterol (LDL-C). Therefore, we sought to fine-map the peak using association mapping of LDL-C as an intermediate disease-related trait to further define the etiology of this linkage peak. The study populations consisted of 1908 individuals from the CATHGEN biorepository of patients undergoing cardiac catheterization; 254 families (N = 827 individuals) from the GENECARD familial study of early-onset CAD; and 162 aorta samples harvested from deceased donors. Linkage disequilibrium-tagged SNPs were selected with an average of one SNP per 20 kb for 126.6-160.2 MB (region of highest linkage) and less dense spacing (one SNP per 50 kb) for the flanking regions (117.7-126.6 and 160.2-167.5 MB) and genotyped on all samples using a custom Illumina array. Association analysis of each SNP with LDL-C was performed using multivariable linear regression (CATHGEN) and the quantitative trait transmission disequilibrium test (QTDT; GENECARD). SNPs associated with the intermediate quantitative trait, LDL-C, were then assessed for association with CAD (i.e., a qualitative phenotype) using linkage and association in the presence of linkage (APL; GENECARD) and logistic regression (CATHGEN and aortas).

Results

We identified four genes with SNPs that showed the strongest and most consistent associations with LDL-C and CAD: EBF1, PPP2R2B, SPOCK1, and PRELID2. The most significant results for association of SNPs with LDL-C were: EBF1, rs6865969, p = 0.01; PPP2R2B, rs2125443, p = 0.005; SPOCK1, rs17600115, p = 0.003; and PRELID2, rs10074645, p = 0.0002). The most significant results for CAD were EBF1, rs6865969, p = 0.007; PPP2R2B, rs7736604, p = 0.0003; SPOCK1, rs17170899, p = 0.004; and PRELID2, rs7713855, p = 0.003.

Conclusion

Using an intermediate disease-related quantitative trait of LDL-C we have identified four novel CAD genes, EBF1, PRELID2, SPOCK1, and PPP2R2B. These four genes should be further examined in future functional studies as candidate susceptibility loci for cardiovascular disease mediated through LDL-cholesterol pathways.

Polymorphic variants in tenascin-C (TNC) are associated with atherosclerosis and coronary artery disease

Tenascin-C (TNC) is an extracellular matrix protein implicated in biological processes important for atherosclerotic plaque development and progression, including smooth muscle cell migration and proliferation. Previously, we observed differential expression of TNC in atherosclerotic aortas compared with healthy aortas. The goal of this study was to investigate whether common genetic variation within TNC is associated with risk of atherosclerosis and coronary artery disease (CAD) in three independent datasets. We genotyped 35 single nucleotide polymorphisms (SNPs), including 21 haplotype tagging SNPs, in two of these datasets: human aorta tissue samples (n = 205) and the CATHGEN cardiovascular study (n = 1,325). Eleven of these 35 SNPs were then genotyped in a third dataset, the GENECARD family study of early-onset CAD (n = 879 families). Three SNPs representing a block of linkage disequilibrium, rs3789875, rs12347433, and rs4552883, were significantly associated with atherosclerosis in multiple datasets and demonstrated consistent, but suggestive, genetic effects in all analyses. In combined analysis rs3789875 and rs12347433 were statistically significant after Bonferroni correction for 35 comparisons, p = 2 × 10(-6) and 5 × 10(-6), respectively. The SNP rs12347433 is a synonymous coding SNP and may be biologically relevant to the mechanism by which tenascin-C influences the pathophysiology of CAD and atherosclerosis. This is the first report of genetic association between polymorphisms in TNC and atherosclerosis or CAD.

A data-driven weighting scheme for family-based genome-wide association studies

Recently, Steen et al proposed a novel two-stage approach for family-based genome-wide association studies. In the first stage, a test based on between-family information is used to rank SNPs according to their P-values or conditional power of the test. In the second stage, the R most promising SNPs are tested using a family-based association test. We call this two-stage approach top R method. Ionita-Laza et al proposed an exponential weighting method within a two-stage framework. In the second stage of this approach, instead of testing top R SNPs, it tests all SNPs and weights the P-values of association test according to the information of the first stage. However, both of the top R and exponential weighting methods only use the information from the first stage to rank SNPs. It seems that the two methods do not use information from the first stage efficiently. Furthermore, it may be unreasonable for the exponential weighting method to use the same weight for all SNPs within a group when only one or a few SNPs are related with a disease. In this article, we propose a data-driven weighting scheme within a two-stage framework. In this method, we use the information from the first stage to determine a SNP-specific weight for each SNP. We use simulation studies to evaluate the performance of our method. The simulation results showed that our proposed method is consistently more powerful than the top R method and the exponential weighting method, regardless of the LD structure, population structure, and family structure.

Ordered-subset analysis (OSA) for family-based association mapping of complex traits

Association analysis provides a powerful tool for complex disease gene mapping. However, in the presence of genetic heterogeneity, the power for association analysis can be low since only a fraction of the collected families may carry a specific disease susceptibility allele. Ordered-subset analysis (OSA) is a linkage test that can be powerful in the presence of genetic heterogeneity. OSA uses trait-related covariates to identify a subset of families that provide the most evidence for linkage. A similar strategy applied to genetic association analysis would likely result in increased power to detect association. Association in the presence of linkage (APL) is a family-based association test (FBAT) for nuclear families with multiple affected siblings that properly infers missing parental genotypes when linkage is present. We propose here APL-OSA, which applies the OSA method to the APL statistic to identify a subset of families that provide the most evidence for association. A permutation procedure is used to approximate the distribution of the APL-OSA statistic under the null hypothesis that there is no relationship between the family-specific covariate and the family-specific evidence for allelic association. We performed a comprehensive simulation study to verify that APL-OSA has the correct type I error rate under the null hypothesis. This simulation study also showed that APL-OSA can increase power relative to other commonly used association tests (APL, FBAT and FBAT with covariate adjustment) in the presence of genetic heterogeneity. Finally, we applied APL-OSA to a family study of age-related macular degeneration, where cigarette smoking was used as a covariate.

Family-based methods for linkage and association analysis

Traditional epidemiological study concepts such as case-control or cohort designs can be used in the design of genetic association studies, giving them a prominent role in genetic association analysis. A different class of designs based on related individuals, typically families, uses the concept of Mendelian transmission to achieve design-independent randomization, which permits the testing of linkage and association. Family-based designs require specialized analytic methods but they have distinct advantages: They are robust to confounding and variance inflation, which can arise in standard designs in the presence of population substructure; they test for both linkage and association; and they offer a natural solution to the multiple comparison problem. This chapter focuses on family-based designs. We describe some basic study designs as well as general approaches to analysis for qualitative, quantitative, and complex traits. Finally, we review available software.

Comparison of GIST and LAMP on the GAW15 simulated data

After genetic linkage has been identified for a complex disease, the next step is often fine-mapping by association analysis, using single-nucleotide polymorphisms (SNPs) within a linkage region. If a SNP shows evidence of association, it is useful to know whether the linkage result can be explained in part or in full by the candidate SNP. The genotype identity-by-descent sharing test (GIST) and linkage and association modeling in pedigrees (LAMP) are two methods that were specifically proposed to address this question. GIST determines whether there is significant correlation between family-specific weights, defined by the presence of a tentatively associated allele in affected siblings, and family-specific nonparametric linkage scores. LAMP constructs a pedigree likelihood function of the marker data conditional on the trait data, and implements three likelihood ratio tests to characterize the relationship between the candidate SNP and the disease locus. The goal of our study was to compare the two approaches and evaluate their ability to identify disease-associated SNPs in the Genetic Analysis Workshop 15 (GAW15) simulated data. Our results can be summarized as follows: 1) GIST is simple and fast but, as a test of association, did not perform well in the GAW15 data, especially with adjustment for multiple testing; 2) as a test of association, the LAMP-LE test performs best when the linkage evidence is strong, or when there is at least moderate linkage disequilibrium between the candidate SNP and the trait locus. We conclude that LAMP is more flexible and reliable to use in practice.

Two-stage study designs for analyzing disease-associated covariates: linkage thresholds and case-selection strategies

The incorporation of disease-associated covariates into studies aiming to identify susceptibility genes for complex human traits is a challenging problem. Accounting for such covariates in genetic linkage and association analyses may help reduce the genetic heterogeneity inherent in these complex phenotypes. For Genetic Analysis Workshop 15 (GAW15) Problem 3 simulated data, our goal was to compare the power of several two-stage study designs to identify rheumatoid arthritis-related genes on chromosome 9 (disease severity), 11 (IgM), and 18 (anti-cyclic citrinullated protein), with knowledge of the answers. Five study designs incorporating an initial linkage step, followed by a case-selection scheme and case-control association analysis by logistic regression, were considered. The linkage step was either qualitative-trait linkage analysis as implemented in MERLIN-nonparametric linkage (NPL), or quantitative-trait locus analysis as implemented in MERLIN-REGRESS. A set of cases representing either one case from each available family, one case per linked family (NPL >/= 0), or one case from each family identified by ordered-subset analysis was chosen for comparison with the full set of 2000 simulated controls. As expected, the performance of these study designs depended on the disease model used to generate the data, especially the simulated allele frequency difference between cases and controls. The quantitative trait loci analysis performed well in identifying these loci, and the power to identify disease-associated alleles was increased by using ordered-subset analysis as a case selection tool.

Increased efficiency of case-control association analysis by using allele-sharing and covariate information

OBJECTIVE

We compared the efficiency of case selection strategies for following up a genome-wide linkage screen of multiplex families. We simulated datasets under three models by which continuous environmental or clinical covariates may contribute to disease risk or linkage heterogeneity: (i) a quantitative trait locus (QTL) underlying a continuous disease risk factor, (ii) a gene-environment interaction model, (iii) a heterogeneity model defined by distinct covariate distributions in linked and unlinked families.

METHODS

Marker genotypes and covariate values were generated for affected sibling pair (ASP) families, according to the three models above. We evaluated two case selection strategies relative to a reference design, which compared all family probands to a sample of unrelated controls ('all'). The first strategy ignored covariates and selected probands from families with NPL scores > or =0 ('linked best'). The second strategy selected probands from families identified by an ordered subset analysis (OSA), which utilizes family-specific linkage and covariate information.

RESULTS

The 'linked best' design provided power very similar to the 'all' design under all three models. Under some QTL and heterogeneity models, the OSA design was both most powerful and most efficient.

CONCLUSIONS

Incorporating allele sharing and covariate information from ASP families into a case-control study design can increase power and reduce genotyping cost.