Robust multipoint simultaneous identical-by-descent mapping for two linked loci

A challenging issue in genetic mapping of complex human diseases is localizing disease susceptibility genes when the genetic effects are small to moderate. There are greater complexities when multiple loci are linked to a chromosomal region. Liang et al. [Hum Hered 2001;51:64-78] proposed a robust multipoint method that can simultaneously estimate both the position of a trait locus and its effect on disease status by using affected sib pairs (ASPs). Based on the framework of generalized estimating equations (GEEs), the estimate and standard error of the position of a trait locus are robust to different genetic models. To utilize other relative pairs collected in pedigree data, Schaid et al. [Am J Hum Genet 2005;76:128-138] extended Liang's method to various types of affected relative pairs (ARPs) by two approaches: unconstrained and constrained methods. However, the above methods are limited to situations in which only one trait locus exists on the chromosome of interest. The mean functions are no longer correctly specified when there are multiple causative loci linked to a chromosomal region. To overcome this, Biernacka et al. [Genet Epidemiol 2005;28:33-47] considered the multipoint methods for ASPs to allow for two linked disease genes. We further generalize the approach to cover other types of ARPs. To reflect realistic situations for complex human diseases, we set modest sizes of genetic effects in our simulation. Our results suggest that several hundred independent pedigrees are needed, and markers with high information, to provide reliable estimates of trait locus positions and their confidence intervals. Bootstrap resampling can correct the downward bias of the robust variance for location estimates. These methods are applied to a prostate cancer linkage study on chromosome 20 and compared with the results for the one-locus model [Am J Hum Genet 2005;76:128-138]. We have implemented the multipoint IBD mapping for one and two linked loci in our software GEEARP, which allows analyses for five general types of ARPs.     

Comparison of 'model-free' and 'model-based' linkage statistics in the presence of locus heterogeneity: single data set and multiple data set applications

Earlier work [Knapp et al.: Hum Hered 1994;44:44-51] focusing on affected sib pair (ASP) data established the equivalence between the mean test and a test based on a simple recessive lod score, as well as equivalences between certain forms of the maximum likelihood score (MLS) statistic [Risch: Am J Hum Genet 1990;46:242-253] and particular forms of the lod score. Here we extend the results of Knapp et al. [1994] by reconsidering these equivalences for ASP data, but in the presence of locus heterogeneity. We show that Risch's MLS statistic under the possible triangle constraints [Holmans: Am J Hum Genet 1993;52:362-374] is locally equivalent to the ordinary heterogeneity lod score assuming a simple recessive model (HLOD/R); while the one-parameter MLS assuming no dominance variance is locally equivalent to the (homogeneity) recessive lod. The companion paper (this issue, pp 199-208) showed that when considering multiple data sets in the presence of locus heterogeneity, the HLOD can suffer appreciable losses in power. We show here that in ASP data, these equivalences ensure that this same loss in power is incurred by both forms of the MLS statistic as well. The companion paper also introduced an adaptation of the lod, the compound lod score (HLOD/C). We confirm that the HLOD/C maintains higher power than these 'model-free' methods when applied to multiple heterogeneous data sets, even when it is calculated assuming the wrong genetic model.     

The linkage information content value of polymorphism genetic markers in model-free linkage analysis

Guo and Elston [Hum Hered 1999;49:112-118] developed a linkage information content (LIC) value to measure the informativeness of a marker for identity-by-descent (IBD) sharing status of relative pairs. LIC values were derived for five types of relative pairs: full sib, half sib, grandparent-grandchild, first cousin and avuncular. In this paper, we give corrected LIC values for full sib, grandparent-grandchild, first cousin and avuncular pairs, and indicate the availability of a computer program to calculate them.     

X-linked extension of the revised Haseman-Elston algorithm for linkage analysis in sib pairs

Haseman and Elston (H-E) proposed a regression-based robust test of linkage between a marker and an autosomal quantitative trait locus, using the squared sib pair trait difference as a dependent variable and the proportion of alleles shared identical by descent by the sib pair as an independent variable. Several authors have proposed improvement of the original H-E's seminal work by using an optimal linear combination of squared sum and squared difference as the dependent variable. In this paper, we extend Haseman and Elston's sib pair method to an X-linked locus. We give a general formulation of the complete regression model and details of the regression coefficients in terms of variance components. Simulation results are presented to describe the power of this technique for a theoretical best case scenario.     

Efficiency robust tests for mapping quantitative trait loci using extremely discordant sib pairs

In 1972, Haseman and Elston proposed a pioneering regression method for mapping quantitative trait loci using randomly selected sib pairs. Recently, the statistical power of their method was shown to be increased when extremely discordant sib pairs are ascertained. While the precise genetic model may not be known, prior information that constrains IBD probabilities is often available. We investigate properties of tests that are robust against model uncertainty and show that the power gain from further constraining IBD probabilities is marginal. The additional linkage information contained in the trait values can be incorporated by combining the Haseman-Elston regression method and a robust allele sharing test.     

Equivalence between Haseman-Elston and variance-components linkage analyses for sib pairs

The Haseman-Elston regression method offers a simpler alternative to variance-components (VC) models, for the linkage analysis of quantitative traits. However, even the "revisited" method, which uses the cross-product--rather than the squared difference--in sib trait values, is, in general, less powerful than VC models. In this report, we clarify the relative efficiencies of existing Haseman-Elston methods and show how a new Haseman-Elston method can be constructed to have power equivalent to that of VC models. This method uses as the dependent variable a linear combination of squared sums and squared differences, in which the weights are determined by the overall trait correlation between sibs in a population. We show how this method can be used for both the selection of maximally informative sib pairs for genotyping and the subsequent analysis of such selected samples.     

A method for evaluating the results of Bayesian model selection: application to linkage analyses of attributes determined by two or more genes

OBJECTIVES: We apply and evaluate the intrinsic Bayes factor (IBF) of Berger and Pericchi [J Am Stat Assoc 1996;91:109-122; Bayesian Statistics, Oxford University Press, vol 5, 1996] to linkage analyses done using the stochastic search variable selection (SSVS) method of George and McCulloch [J Am Stat Assoc 1993;88:881-889] as proposed by Suh et al. [Genet Epidemiol 2001;21(suppl 1):S706-S711]. METHODS: We consider 20 simulations of linkage data obtained under two different generating models. The SSVS is applied to a multiple regression extension [Genet Epidemiol 2001;21(suppl 1): S706-S711] of the Haseman-Elston [Behav Genet 1972;2:3-19; Genet Epidemiol 2000;19:1-17] methods. Four prior distributions are considered. We apply the IBF criterion to those samples where different prior distributions result in different top models. RESULTS: In those samples where three different models were obtained using the four priors, application of the IBFs eliminated one of the two wrong models in 4 out of 5 situations. Further elimination using the IBF criterion for situations with two different subsets did not serve as well. CONCLUSIONS: When different priors result in three or more different subsets of markers, one can use the IBF to get this number down to two for consideration. When two subsets result we recommend that both be considered.     

A unified Haseman-Elston method for testing linkage with quantitative traits

The Haseman and Elston (H-E) method uses a simple linear regression to model the squared trait difference of sib pairs with the shared allele identical by descent (IBD) at marker locus for linkage testing. Under this setting, the squared mean-corrected trait sum is also linearly related to the IBD sharing. However, the resulting slope estimate for either model is not efficient. In this report, we propose a simple linkage test that optimally uses information from the estimates of both models. We also demonstrate that the new test is more powerful than both the traditional one and the recently revisited H-E methods.     

A multivariate method for detecting genetic linkage, with application to a pedigree with an adverse lipoprotein phenotype.

The robust or model-free method for detecting linkage developed by Haseman and Elston for data from sib pairs is extended to incorporate observations of multiple traits on each individual. A method is proposed that estimates the linear function that results in the strongest correlation between the squared pair differences in the trait measurements and identity by descent at a marker locus. The method is illustrated by the study of apolipoprotein and cholesterol levels in individuals from a large family that had many members diagnosed with coronary heart disease.     

Effects of misspecification of allele frequencies on the type I error rate of model-free linkage analysis

Linkage analyses of simulated quantitative trait data were performed using the Haseman-Elston (H-E) sib pair regression test to investigate the effects of inaccurate allele frequency estimates on the type I error rates of this test. Computer simulations generating a quantitative trait in nuclear families were performed using GASP [1]. Assuming no linkage, several data sets were simulated; they differed in marker allele numbers and frequencies, number of sib pairs and number of sibships. Each set of simulated data was analyzed using (1) all parental marker data, (2) half of the parental marker data, and (3) no parental marker data, using both correct and incorrect allele frequencies in the latter 2 cases. The H-E sib pair linkage method was found to be robust to misspecification of marker allele frequencies regardless of the number of alleles.     

Haseman-Elston weighted by marker informativity

In the Haseman-Elston approach the squared phenotypic difference is regressed on the proportion of alleles shared identical by descent (IBD) to map a quantitative trait to a genetic marker. In applications the IBD distribution is estimated and usually cannot be determined uniquely owing to incomplete marker information. At Genetic Analysis Workshop (GAW) 13, Jacobs et al. [BMC Genet 2003, 4(Suppl 1):S82] proposed to improve the power of the Haseman-Elston algorithm by weighting for information available from marker genotypes. The authors did not show, however, the validity of the employed asymptotic distribution. In this paper, we use the simulated data provided for GAW 14 and show that weighting Haseman-Elston by marker information results in increased type I error rates. Specifically, we demonstrate that the number of significant findings throughout the chromosome is significantly increased with weighting schemes. Furthermore, we show that the classical Haseman-Elston method keeps its nominal significance level when applied to the same data. We therefore recommend to use Haseman-Elston with marker informativity weights only in conjunction with empirical p-values. Whether this approach in fact yields an increase in power needs to be investigated further.     

Adding further power to the Haseman and Elston method for detecting linkage in larger sibships: weighting sums and differences

Haseman and Elston (H-E) proposed a robust test to detect linkage between a quantitative trait and a genetic marker. In their method the squared sib-pair trait difference is regressed on the estimated proportion of alleles at a locus shared identical by descent by sib pairs. This method has recently been improved by changing the dependent variable from the squared difference to the mean-corrected product of the sib-pair trait values, a significantly positive regression indicating linkage. Because situations arise in which the original test is more powerful, a further improvement of the H-E method occurs when the dependent variable is changed to a weighted average of the squared sib-pair trait difference and the squared sib-pair mean-corrected trait sum. Here we propose an optimal method of performing this weighting for larger sibships, allowing for the correlation between pairs within a sibship. The optimal weights are inversely proportional to the residual variances obtained from the two different regressions based on the squared sib-pair trait differences and the squared sib-pair mean-corrected trait sums, respectively, allowing for correlations among sib pairs. The proposed method is compared with the existing extension of the H-E approach for larger sibships. Control of the type I error probabilities for sibships of any size can be improved by using a generalized estimating equation approach and the robust sandwich estimate of the variance, or a Monte-Carlo permutation test.     

Heterogeneity in IBD allele sharing among covariate-defined subgroups: issues and findings for affected relatives

OBJECTIVES: Modelling of variation in identical-by-descent (IBD) allele sharing using covariates can increase power to detect linkage, identify covariate-defined subgroups linked to particular marker regions, and improve the design of subsequent studies to localize genes and characterize their effects. In this report, we highlight issues that arise in studies of families with affected relatives. METHODS: Mirea et al. [Genet Epidemiol 2003, in press] extended linear and exponential linkage likelihood models [Kong and Cox, Am J Hum Genet 1997;61: 1179-1188] to model variation in NPL scores among covariate-defined groups of families, and proposed likelihood ratio (LR) and t statistics to detect differences in allele sharing between groups defined by a binary covariate. Here we evaluate factors affecting the power of these tests analytically and by example, as well as effects of constraints, nuisance parameters, and incomplete data on test validity by simulation of locus heterogeneity in families with affected siblings or affected cousins. RESULTS: Provided constraints on the parameters are avoided, these tests are particularly useful when one subgroup has less than expected IBD sharing. The distribution of the LR statistic depends on the extent of linkage, particularly in the presence of constraints. The t statistic may be biased by group differences in information content. CONCLUSIONS: We recommend that constraints be applied cautiously, and covariate effects in IBD allele sharing models interpreted with care.     

Effect of Box-Cox transformation on power of Haseman-Elston and maximum-likelihood variance components tests to detect quantitative trait Loci

Non-normality of the phenotypic distribution can affect power to detect quantitative trait loci in sib pair studies. Previously, we observed that Winsorizing the sib pair phenotypes increased the power of quantitative trait locus (QTL) detection for both Haseman-Elston (HE) least-squares tests [Hum Hered 2002;53:59-67] and maximum likelihood-based variance components (MLVC) analysis [Behav Genet (in press)]. Winsorizing the phenotypes led to a slight increase in type 1 error in H-E tests and a slight decrease in type I error for MLVC analysis. Herein, we considered transforming the sib pair phenotypes using the Box-Cox family of transformations. Data were simulated for normal and non-normal (skewed and kurtic) distributions. Phenotypic values were replaced by Box-Cox transformed values. Twenty thousand replications were performed for three H-E tests of linkage and the likelihood ratio test (LRT), the Wald test and other robust versions based on the MLVC method. We calculated the relative nominal inflation rate as the ratio of observed empirical type 1 error divided by the set alpha level (5, 1 and 0.1% alpha levels). MLVC tests applied to non-normal data had inflated type I errors (rate ratio greater than 1.0), which were controlled best by Box-Cox transformation and to a lesser degree by Winsorizing. For example, for non-transformed, skewed phenotypes (derived from a chi2 distribution with 2 degrees of freedom), the rates of empirical type 1 error with respect to set alpha level=0.01 were 0.80, 4.35 and 7.33 for the original H-E test, LRT and Wald test, respectively. For the same alpha level=0.01, these rates were 1.12, 3.095 and 4.088 after Winsorizing and 0.723, 1.195 and 1.905 after Box-Cox transformation. Winsorizing reduced inflated error rates for the leptokurtic distribution (derived from a Laplace distribution with mean 0 and variance 8). Further, power (adjusted for empirical type 1 error) at the 0.01 alpha level ranged from 4.7 to 17.3% across all tests using the non-transformed, skewed phenotypes, from 7.5 to 20.1% after Winsorizing and from 12.6 to 33.2% after Box-Cox transformation. Likewise, power (adjusted for empirical type 1 error) using leptokurtic phenotypes at the 0.01 alpha level ranged from 4.4 to 12.5% across all tests with no transformation, from 7 to 19.2% after Winsorizing and from 4.5 to 13.8% after Box-Cox transformation. Thus the Box-Cox transformation apparently provided the best type 1 error control and maximal power among the procedures we considered for analyzing a non-normal, skewed distribution (chi2) while Winzorizing worked best for the non-normal, kurtic distribution (Laplace). We repeated the same simulations using a larger sample size (200 sib pairs) and found similar results.     

Transformation of sib-pair values for the Haseman-Elston method

The squared sib-pair phenotype difference (SQD) has been used as a dependent variable in the Haseman-Elston (H-E) regression quantitative-trait locus (QTL) linkage method, but it has been shown that the SQD does not make full use of linkage information. In this study, we examine the efficiency of SQD in H-E regression compared to other proposed functions of the sib-pair phenotypes. A new function of sib-pair phenotypes, the product of pair values corrected with family mean (PCF), is shown to have desirable properties in many realistic situations. Consistent results were obtained using a combination of large-sample analytic approximations, simulation, and analyses of quantitative-trait data from Genetic Analysis Workshop 10. The advantages of PCF are further improved in the presence of family-specific effects arising from environmental factors or when additional QTLs influence the trait. All of the phenotype functions are incorporated in our new, freely available linkage-mapping program MULTIGENE 1.0 for the PC environment.     

Using parametric multipoint lods and mods for linkage analysis requires a shift in statistical thinking

Multipoint (MP) linkage analysis represents a valuable tool for whole-genome studies but suffers from the disadvantage that its probability distribution is unknown and varies as a function of marker information and density, genetic model, number and structure of pedigrees, and the affection status distribution [Xing and Elston: Genet Epidemiol 2006;30:447-458; Hodge et al.: Genet Epidemiol 2008;32:800-815]. This implies that the MP significance criterion can differ for each marker and each dataset, and this fact makes planning and evaluation of MP linkage studies difficult. One way to circumvent this difficulty is to use simulations or permutation testing. Another approach is to use an alternative statistical paradigm to assess the statistical evidence for linkage, one that does not require computation of a p value. Here we show how to use the evidential statistical paradigm for planning, conducting, and interpreting MP linkage studies when the disease model is known (lod analysis) or unknown (mod analysis). As a key feature, the evidential paradigm decouples uncertainty (i.e. error probabilities) from statistical evidence. In the planning stage, the user calculates error probabilities, as functions of one's design choices (sample size, choice of alternative hypothesis, choice of likelihood ratio (LR) criterion k) in order to ensure a reliable study design. In the data analysis stage one no longer pays attention to those error probabilities. In this stage, one calculates the LR for two simple hypotheses (i.e. trait locus is unlinked vs. trait locus is located at a particular position) as a function of the parameter of interest (position). The LR directly measures the strength of evidence for linkage in a given data set and remains completely divorced from the error probabilities calculated in the planning stage. An important consequence of this procedure is that one can use the same criterion k for all analyses. This contrasts with the situation described above, in which the value one uses to conclude significance may differ for each marker and each dataset in order to accommodate a fixed test size, α. In this study we accomplish two goals that lead to a general algorithm for conducting evidential MP linkage studies. (1) We provide two theoretical results that translate into guidelines for investigators conducting evidential MP linkage: (a) Comparing mods to lods, error rates (including probabilities of weak evidence) are generally higher for mods when the null hypothesis is true, but lower for mods in the presence of true linkage. Royall [J Am Stat Assoc 2000;95:760-780] has shown that errors based on lods are bounded and generally small. Therefore when the true disease model is unknown and one chooses to use mods, one needs to control misleading evidence rates only under the null hypothesis; (b) for any given pair of contiguous marker loci, error rates under the null are greatest at the midpoint between the markers spaced furthest apart, which provides an obvious simple alternative hypothesis to specify for planning MP linkage studies. (2) We demonstrate through extensive simulation that this evidential approach can yield low error rates under the null and alternative hypotheses for both lods and mods, despite the fact that mod scores are not true LRs. Using these results we provide a coherent approach to implement a MP linkage study using the evidential paradigm.     

Improving the power of sib pair quantitative trait loci detection by phenotype winsorization

OBJECTIVES: In sib pair studies, quantitative trait loci (QTL) identification may be adversely affected by non-normality in the phenotypic distribution, particularly when subjects falling in the tails of the distribution bias the trait mean or variance. We evaluated the robustness and power of reducing the influence of subjects with extreme phenotypic values by Winsorizing non-normal distributions in three versions of Haseman-Elston regression-based methods of QTL linkage analysis. METHODS: Data were simulated for normal and non-normal distributions. Phenotypic values that correspond to cutoff points at the omega and 1 - omega percentiles of the distribution were identified, and phenotypic values falling outside the boundaries of the omega and 1 - omega cutoff points were replaced by the omega and 1 - omega values, respectively. One million replications were performed for the three tests of linkage for Winsorized and non-Winsorized data. RESULTS: Winsorization reduced conservatism in the tails of the empirical type I error rate for the vast majority of the tests of linkage, increased the power of QTL detection in non-normal data and created a slight negative bias in symmetrical phenotypic distributions. CONCLUSIONS: Winsorizing can improve the power of QTL detection with certain non-normal distributions but can also introduce bias into the estimate of the QTL effect.     

Comments on the entropy-based transmission/disequilibrium test

It has recently been claimed in this journal (Zhao et al. in Hum Genet 121:357–367, 2007) that a so-called “entropy-based” TDT test has improved power over the standard TDT test of Spielman et al. (Am J Hum Genet 52:506–516, 1993). We show that this claim is contradicted by standard statistical theory as well as by our simulation results. We show that the incorrect claim arises because of inappropriate assumptions, and also show that the entropy-based statistic has various undesirable properties.     

Brief communication: How much larger is the relative volume of area 10 of the prefrontal cortex in humans?

It has long been thought that the prefrontal cerebral cortex has been greatly expanded in the human brain. Semendeferi et al. ([2001] Am. J. Phys. Anthropol. 114:224-241) showed that Brodmann's area 10 is relatively larger in the human compared to pongid brains. The question is: how much larger relatively is it? Using their data, it can be shown that the relative increase for human prefrontal area 10 is only 6% larger. Looking at the data base of neural structures provided by Stephan et al. ([1981] Folia Primatol. (Basel) 35:1-29), it is apparent that 6% is a relatively low residual value from a predicted value based on allometric considerations between total brain weight and any given neural structure. When this small increase is combined with their earlier findings on area 13 of prefrontal cortex (Semendeferi et al. [1997] J. Hum. Evol. 32:375-388), it appears that the prefrontal cortex in humans is not some 200% larger as claimed by some researchers (Deacon [1997] Symbolic Species, New York: W.W. Norton; cf. Holloway [1998] Am Sci 86:184-186), and that the findings of Semendeferi et al. ([2001] Am. J. Phys. Anthropol. 114:224-241) are in agreement with the earlier work (Semendeferi and Damasio [2000] J. Hum. Evol. 38:317-332; Semendeferi et al. [1997] J. Hum. Evol. 32:375-388), showing that the human frontal lobe volume is what would be expected for a primate of its brain size. While the prefrontal cortex may have increased relatively in Homo sapiens, the increase is likely to have been far less than currently believed.     

Nonparametric trend statistic incorporating dispersion differences in sib pair linkage for quantitative traits

The Haseman-Elston (HE) regression method and its extensions are widely used in genetic studies for detecting linkage to quantitative trait loci (QTL) using sib pairs. The principle underlying the simple HE regression method is that the similarity in phenotypes between two siblings increases as they share an increasing number of alleles identical by descent (IBD) from their parents at a particular marker locus. In such a procedure, similarity was identified with the locations, that is, means of groups of sib pairs sharing 0, 1, and 2 alleles IBD. A more powerful, rank-based nonparametric test to detect increasing similarity in sib pairs is presented by combining univariate trend statistics not only of locations, but also of dispersions of the squared phenotypic differences of two siblings for three groups. This trend test does not rely on distributional assumptions, and is applicable to the skewed or leptokurtic phenotypic distributions, in addition to normal or near normal phenotypic distributions. The performances of nonparametric trend statistics, including nonparametric regression slope, are compared with the HE regression methods as genetic linkage strategies.