Multiple subsampling of dense SNP data localizes disease genes with increased precision

Model-Free Linkage Analysis of a Binary Trait

Genetic linkage analysis aims to detect chromosomal regions containing genetic variants that influence risk of specific inherited diseases. The presence of linkage is indicated when a disease or trait cosegregates through the families with genetic markers at a particular region of the genome. Two main types of genetic linkage analysis are in common use, namely model-based linkage analysis and model-free linkage analysis. In this chapter, we focus solely on the latter type and specifically on binary traits or phenotypes, such as the presence or absence of a specific disease. Model-free linkage analysis is based on allele-sharing, where patterns of genetic similarity among affected relatives are compared to chance expectations. Because the model-free methods do not require the specification of the inheritance parameters of a genetic model, they are preferred by many researchers at early stages in the study of a complex disease. We introduce the history of model-free linkage analysis in Subheading 1. Table 1 describes a standard model-free linkage analysis workflow. We describe three popular model-free linkage analysis methods, the nonparametric linkage (NPL) statistic, the affected sib-pair (ASP) likelihood ratio test, and a likelihood approach for pedigrees. The theory behind each linkage test is described in this section together with a simple example of the relevant calculations. Table 4 provides a summary of popular genetic analysis software packages that implement model-free linkage models. In Subheading 2, we work through the methods on a rich example providing sample software code and output. Subheading 3 contains notes with additional details on various topics that may need further consideration during analysis.

A powerful test of independent assortment that determines genome-wide significance quickly and accurately

In the analysis of DNA sequences on related individuals, most methods strive to incorporate as much information as possible, with little or no attention paid to the issue of statistical significance. For example, a modern workstation can easily handle the computations needed to perform a large-scale genome-wide inheritance-by-descent (IBD) scan, but accurate assessment of the significance of that scan is often hindered by inaccurate approximations and computationally intensive simulation. To address these issues, we developed gLOD—a test of co-segregation that, for large samples, models chromosome-specific IBD statistics as a collection of stationary Gaussian processes. With this simple model, the parametric bootstrap yields an accurate and rapid assessment of significance—the genome-wide corrected P-value. Furthermore, we show that (i) under the null hypothesis, the limiting distribution of the gLOD is the standard Gumbel distribution; (ii) our parametric bootstrap simulator is approximately 40 000 times faster than gene-dropping methods, and it is more powerful than methods that approximate the adjusted P-value; and, (iii) the gLOD has the same statistical power as the widely used maximum Kong and Cox LOD. Thus, our approach gives researchers the ability to determine quickly and accurately the significance of most large-scale IBD scans, which may contain multiple traits, thousands of families and tens of thousands of DNA sequences.

Next-generation linkage and association methods applied to hypertension: a multifaceted approach to the analysis of sequence data

To realize the full potential of next-generation sequencing, it is important to consider multiple sources of genetic information, including inheritance, association, and bioinformatics. To illustrate the promise of such an approach, we applied our next-generation linkage and association (NGLA) methods to the sequence data of a large 57-member Mexican American family with hypertension. Our results show that OSBPL10--a disease susceptibility gene for dyslipidemia--may also influence systolic blood pressure (SBP). In particular, our NGLA dense single-nucleotide polymorphism (SNP) analysis identified a 2.5-megabase (Mb) region that strongly cosegregates with low SBP (maximum posterior probability of linkage [PPL] = 68%). Furthermore, using the posterior probability of linkage disequilibrium (PPLD), we fine-mapped this region and identified 12 SBP-associated variants (PPLD ranging between 4% and 14%) that comprise a rare, 4-site haplotype. This haplotype extends into the candidate gene, OSBPL10 (oxysterol-binding protein-like 10). In contrast to our NGLA methods, a commonly used filter-based approach identified 23 variants with little evidence for spatial clustering around any particular gene or region of interest.

Increasing the power of association studies with affected families, unrelated cases and controls

When studying the genetics of inherited diseases, researchers often collect data on affected families, unrelated cases, and healthy controls. However, the joint analysis of such heterogeneous data is difficult, and the simpler analysis of homogeneous subsets is often suboptimal. For example, while case-control tests of association are sensitive to allele frequency differences, the preferential transmission of risk alleles from heterozygous parents to their affected offspring is typically ignored. Similarly, the transmission disequilibrium test (TDT) fails to incorporate the difference in allele frequencies when testing for association. To boost the power of modern genetic studies, we propose POPFAM – a fast and efficient test of association that can accommodate large affected families, unrelated cases, and controls. We use simulations to assess the type I error and power of POPFAM across different genetic models, and minor allele frequencies. For comparison, we examine the power of competing methods: the trend test, a Wald test (equivalent to the TDT), and SCOUT. Our results show that POPFAM maintains the correct type I error, and that it is more powerful than the trend test or the TDT. It performs as well as, or better than the likelihood ratio test SCOUT, which was developed specifically for case-parent/case-control data. Furthermore, when applied to the human leukocyte antigen genotypes of 401 type 1 diabetic families, POPFAM confirmed the previously reported association between DRB1^*03:01 and microvascular complications (p = 0.04). In general, we expect our proposed test to facilitate the identification of clinically important genomic regions, and to better inform the design of follow-up sequencing efforts.

How should we be searching for genes for common epilepsy? A critique and a prescription

Despite enormous data collection and analysis efforts, the genetic influences on common epilepsies remain mostly unknown. We propose that reasons for the lack of progress can be traced to three factors: (1) A reluctance to consider fine-grained phenotype definitions based on extensive and carefully collected clinical data; (2) the pursuit of genetic analysis methods that are popular but poorly conceived and are inadequate to the task of resolving the problems inherent in common disease studies; (3) preconceived ideas about the genetic mechanisms that cause epilepsy (which we have discussed elsewhere). We propose a paradigm for finding epilepsy-related loci and alleles that has proven successful in other common diseases.

Obtaining accurate p values from a dense SNP linkage scan

A major concern of resequencing studies is that the pathogenicity of most mutations is difficult to predict. To address this concern, linkage (i.e. co-segregation) analysis is often used to exclude neutral mutations and to better predict pathogenicity among the candidate mutations that remain. However, when linkage disequilibrium (LD) is present in the population but ignored in the analysis, unlinked regions with high LD can inflate the type 1 error and thousands of neutral mutations may be mistakenly included in a follow-up resequencing study, which could dramatically reduce the power to identify causal variants. To illustrate the need for concern, we simulated data on a sparsely spaced panel of single nucleotide polymorphisms (average spacing 1.27 cM) using an LD pattern estimated from real data. In our simulations, we find that the type 1 error of the maximum LOD can be as high as 14%. Therefore, to control the type 1 error of linkage tests we created Haplodrop - a fast and flexible simulation program that generates the haplotypes of founders with LD and then 'drops' these haplotypes with recombination to all non-founders in the pedigree. Haplodrop can be used to control the type 1 error of any linkage test, agrees well with existing software, accommodates arbitrary pedigree structures, and scales easily to the whole genome. Moreover, by correctly excluding mutations that lie in unlinked regions with high LD, Haplodrop should aid significantly in reducing the multiple testing burden of follow-up resequencing studies.

A confidence set inference method for identifying SNPs that regulate quantitative phenotypes

AIMS

We introduce a family-based confidence set inference (CSI) method that can be used in preliminary genome-wide association studies to obtain confidence sets of SNPs that contribute a specific percentage to the additive genetic variance of quantitative traits.

METHODS

Developed in the framework of generalized linear mixed models, the method utilizes data from outbred families of arbitrary size and structure. Through our own simulation study and analysis of the Genetics Analysis Workshop 16 simulated data, we study the properties of our method and compare its performance to that of the family association method described by Chen and Abecasis [Am J Hum Genet 2007;81:913-926]. We also analyze the Framingham Heart Study data to identify SNPs regulating high-density lipoprotein levels.

RESULTS

The simulation studies demonstrated that CSI yields confidence sets with correct coverage and that it can outperform the method introduced by Chen and Abecasis [Am J Hum Genet 2007;81:913-926]. Furthermore, we identified five SNPs that potentially regulate high-density lipoprotein levels: rs9989419, rs11586238, rs1754415, rs9355648, and rs9356560.

CONCLUSION

The CSI method provides confidence sets of SNPs that contribute to the genetic variance of quantitative traits and is a competitive alternative to currently used family association methods. The approach is particularly useful in genome-wide association studies as it significantly reduces the number of SNPs investigated in follow-up studies.

Model-free linkage analysis of a binary trait

Genetic linkage analysis aims to detect chromosomal regions containing genes that influence risk of specific inherited diseases. The presence of linkage is indicated when a disease or trait cosegregates through the families with genetic markers at a particular region of the genome. Two main types of genetic linkage analysis are in common use, namely model-based linkage analysis and model-free linkage analysis. In this chapter, we focus solely on the latter type and specifically on binary traits or phenotypes, such as the presence or absence of a specific disease. Model-free linkage analysis is based on allele-sharing, where patterns of genetic similarity among affected relatives are compared to chance expectations. Because the model-free methods do not require the specification of the inheritance parameters of a genetic model, they are preferred by many researchers at early stages in the study of a complex disease. We introduce the history of model-free linkage analysis in Subheading 1. Table 1 describes a standard model-free linkage analysis workflow. We describe three popular model-free linkage analysis methods, the nonparametric linkage (NPL) statistic, the affected sib-pair (ASP) likelihood ratio test, and a likelihood approach for pedigrees. The theory behind each linkage test is described in this section, together with a simple example of the relevant calculations. Table 4 provides a summary of popular genetic analysis software packages that implement model-free linkage models. In Subheading 2, we work through the methods on a rich example providing sample software code and output. Subheading 3 contains notes with additional details on various topics that may need further consideration during analysis.

Finding disease genes: a fast and flexible approach for analyzing high-throughput data

Linkage disequilibrium (LD) is the non-random distribution of alleles across the genome, and it can create serious problems for modern linkage studies. In particular, computational feasibility is often obtained at the expense of power, precision, and/or accuracy. In our new approach, we combine linkage results over multiple marker subsets to provide fast, efficient, and robust analyses, without compromising power, precision, or accuracy. Allele frequencies and LD in the densely spaced markers are used to construct subsamples that are highly informative for linkage. We have tested our approach extensively, and implemented it in the software package EAGLET (Efficient Analysis of Genetic Linkage: Estimation and Testing). Relative to several commonly used methods we show that EAGLET has increased power to detect disease genes across a range of trait models, LD patterns, and family structures using both simulated and real data. In particular, when the underlying LD pattern is derived from real data, we find that EAGLET outperforms several commonly used linkage methods. In-depth analysis of family data, simulated with linkage and under the real-data derived LD pattern, showed that EAGLET had 78.1% power to detect a dominant disease with incomplete penetrance, whereas the method that uses one marker per cM had 69.7% power, and the cluster-based approach implemented in MERLIN had 76.7% power. In this same setting, EAGLET was three times faster than MERLIN, and it narrowed the MERLIN-based confidence interval for trait location by 29%. Overall, EAGLET gives researchers a fast, accurate, and powerful new tool for analyzing high-throughput linkage data, and large extended families are easily accommodated.