Significance levels in genome-wide interaction analysis (GWIA).

Interaction between genetic variants is hypothesized to be one of several putative explanations for the 'case of missing heritability.' Therefore, Genome-Wide Interaction Analysis (GWIA) has recently gained substantial interest. GWIA is computationally challenging and respective power type I error studies are particularly difficult. Therefore, an accepted significance level for GWIA studies does not currently exist. It has been shown that for a GWAS single-marker analysis with n SNPs a correction for multiple testing with 1/2 · n is appropriate for populations of European ancestry. We speculated that for GWIA, correction by 1/4 · m should be appropriate, where m = n · (n- 1)/2 is the number of SNP pairs. We tried to verify this hypothesis using the INTERSNP program that implements interaction analysis and genome-wide Monte-Carlo (MC) simulation. Using a type I error study based on Illumina(®) HumanHap 550 data, we were able to reproduce the published result for single-marker analysis. For GWIA using a test for allelic interaction, we show that correction with roughly 0.4 · m is appropriate, a number that is somewhat larger than that of our hypothesis. In summary, it can be stated that for an Illumina(®) -type marker panel with 500,000 SNPs, an uncorrected P-value of 1.0 × 10⁻¹² is needed to establish genome-wide significance at the 0.05 level.

Feasible and successful: genome-wide interaction analysis involving all 1.9 x 10(11) pair-wise interaction tests.

The Genome-Wide Association Study (GWAS) is the study design of choice for detecting common genetic risk factors for multifactorial diseases. The performance of full Genome-Wide Interaction Analyses (GWIA) has always been considered computationally challenging. Two-stage strategies to reduce the amount of numerical analysis require the detection of single marker effects or prior pathophysiological hypotheses before the analysis of interaction. This prevents the detection of pure epistatic effects. Our case-control study in idiopathic generalized epilepsy demonstrates that a full GWIA is feasible through use of data compression, specific data representation, interleaved data organization, and parallelization of the analysis on a multi-processor system. Following extensive quality control of the genotypes, our final list of top interaction hits contains only pairs of interacting SNPs with negligible marginal effects. The TOP HIT interaction was between a SNP-pair intragenic to gene DNER (chr 2) and gene CTNNA3 (chr 10). Both of these genes are functionally involved in neuronal migration, synaptogenesis, and the formation of neuronal circuits. Our results therefore indicate a possible interaction between these two genes in epileptogenesis. Results from GWAS are beginning to reveal a 'missing heritability' in complex traits and diseases. Systematic, hypothesis-free analysis of epistatic interaction (GWIA) may help to close this increasingly recognized gap in heritability.

INTERSNP: genome-wide interaction analysis guided by a priori information.

SUMMARY: Genome-wide association studies (GWAS) have lead to the identification of hundreds of genomic regions associated with complex diseases. Nevertheless, a large fraction of their heritability remains unexplained. Interaction between genetic variants is one of several putative explanations for the 'case of missing heritability' and, therefore, a compelling next analysis step. However, genome-wide interaction analysis (GWIA) of all pairs of SNPs from a standard marker panel is computationally unfeasible without massive parallelization. Furthermore, GWIA of all SNP triples is utopian. In order to overcome these computational constraints, we present a GWIA approach that selects combinations of SNPs for interaction analysis based on a priori information. Sources of information are statistical evidence (single marker association at a moderate level), genetic relevance (genomic location) and biologic relevance (SNP function class and pathway information). We introduce the software package INTERSNP that implements a logistic regression framework as well as log-linear models for joint analysis of multiple SNPs. Automatic handling of SNP annotation and pathways from the KEGG database is provided. In addition, Monte Carlo simulations to judge genome-wide significance are implemented. We introduce various meaningful GWIA strategies that can be conducted using INTERSNP. Typical examples are, for instance, the analysis of all pairs of non-synonymous SNPs, or, the analysis of all combinations of three SNPs that lie in a common pathway and that are among the top 50,000 single-marker results. We demonstrate the feasibility of these and other GWIA strategies by application to a GWAS dataset and discuss promising results. AVAILABILITY: The software is available at http://intersnp.meb.uni-bonn.de CONTACT: herold@imbie.meb.uni-bonn.de; becker@imbie.meb.uni-bonn.de.

Linkage analysis of alcohol dependence using MOD scores.

Alcohol dependence is a typical example of a complex trait that is governed by several genes and for which the mode of inheritance is unknown. We analyzed the microsatellite markers and the Affymetrix single-nucleotide polymorphisms (SNPs) for a subset of the Collaborative Study on the Genetics of Alcoholism family sample, 93 pedigrees of Caucasian ancestry comprising 919 persons, 390 of whom are affected according to DSM III-R and Feighner criteria. In particular, we performed parametric single-marker linkage analysis using MLINK of the LINKAGE package (for the microsatellite data), as well as multipoint MOD-score analysis with GENEHUNTER-MODSCORE (for the microsatellite and SNP data). By use of two liability classes, different penetrances were assigned to males and females. In order to investigate parent-of-origin effects, we calculated MOD scores under trait models with and without imprinting. In addition, for the microsatellite data, the MOD-score analysis was performed with sex-averaged as well as sex-specific maps. The highest linkage peaks were obtained on chromosomes 1, 2, 7, 10, 12, 13, 15, and 21. There was evidence for paternal imprinting at the loci on chromosomes 2, 10, 12, 13, 15, and 21. A tendency to maternal imprinting was observed at two loci on chromosome 7. Our findings underscore the fact that an adequate modeling of the genotype-phenotype relation is crucial for the genetic mapping of a complex trait.

Quantitative trait linkage analysis of longitudinal change in body weight.

One of the great strengths of the Framingham Heart Study data, provided for the Genetic Analysis Workshop 13, is the long-term survey of phenotypic data. We used this unique data to create new phenotypes representing the pattern of longitudinal change of the provided phenotypes, especially systolic blood pressure and body weight. We performed a linear regression of body weight and systolic blood pressure on age and took the slopes as new phenotypes for quantitative trait linkage analysis using the SOLAR package. There was no evidence for heritability of systolic blood pressure change. Heritability was estimated as 0.15 for adult life "body weight change", measured as the regression slope, and "body weight gain" (including only individuals with a positive regression slope), and as 0.22 for body weight "change up to 50" (regression slope of weight on age up to an age of 50). With multipoint analysis, two regions on the long arm of chromosome 8 showed the highest LOD scores of 1.6 at 152 cM for "body weight change" and of >1.9 around location 102 cM for "body weight gain" and "change up to 50". The latter two LOD scores almost reach the threshold for suggestive linkage. We conclude that the chromosome 8 region may harbor a gene acting on long-term body weight regulation, thereby contributing to the development of the metabolic syndrome.

An unusual case of identification by DNA analysis of siblings.

A badly decomposed body required identification by means of DNA analysis. A brother and sister of the deceased were available as reference subjects. Although investigation of Y-chromosomal markers established an exclusion condition, autosomal markers suggested a positive identification. In order to increase the reliability of the tests, X-chromosomal markers were also investigated. This analysis showed the body to have an XXY genotype (Klinefelter's syndrome). A number of hypotheses were assessed using biostatistical methods, ultimately resulting in a definite identification. The special aspect of Klinefelter's syndrome proved highly useful for biostatistical analysis.

Spatial assessment of Argentinean genetic admixture with geographical information systems.

In recent years there has been much attention to Argentinean population stratification. We were interested in assessing population stratification from a geographical perspective and summarizing it in form of maps. We mapped the genetic admixture of the extant male population in central and northern Argentina on the basis of forensic Y-chromosomal haplotypes. We addressed the question which group of genetically similar individuals is predominant in this area. Haplotypes containing seven Y-chromosomal short tandem repeat polymorphisms (Y-STRs), also known as microsatellites - DYS19, DYS389I, DYS389II, DYS390, DYS391, DYS392, DYS393 - were constructed for 145 individuals, recruited in 10 provinces. 97 distinct haplotypes were clustered into four clusters according to molecular distances. A genetic geostatistical analysis was conducted with the open-source geographical information system GRASS GIS. For each haplotype cluster, the according frequency was spatially interpolated over the total study area. Juxtaposing the interpolation surfaces, we screened point-wisely the maximal frequency as well as the label of the respective cluster. The screening results were combined in one summary map. We repeated this procedure for the second maximal frequencies. The resulting maps subdivide the study area into continuous regions comprising one predominant group of similar haplotypes. The first summary map divides the study area into three regions and the second summary map divides the area into four regions. The results of our analysis indicate that two groups of similar European haplotypes alternatively dominate the largest extension of the Argentinean territory. A third group, including South-American haplotypes, dominates the indigenous northwestern Argentinean area. The last group, including worldwide dispersed haplotypes, preponderates in frequency in second place in central Argentina. Our findings confirm a widespread European paternal ancestry, a substantial Amerindian contribution in the northwest, as well as a considerable proportion of diverse paternal lineages. In this work, we further discuss these findings in reference to ethno-historical, genetic, and demographic information.

Geostatistical inference of main Y-STR-haplotype groups in Europe.

We examined the multifarious genetic heterogeneity of Europe and neighboring regions from a geographical perspective. We created composite maps outlining the estimated geographical distribution of major groups of genetically similar individuals on the basis of forensic Y-chromosomal markers. We analyzed Y-chromosomal haplotypes composed of 7 highly polymorphic STR loci, genotyped for 33,010 samples, collected at 249 sites in Europe, Western Asia and North Africa, deposited in the YHRD database (www.yhrd.org). The data set comprised 4176 different haplotypes, which we grouped into 20 clusters. For each cluster, the frequency per site was calculated. All geostatistical analysis was performed with the geographic information system GRASS-GIS. We interpolated frequency values across the study area separately for each cluster. Juxtaposing all 20 interpolated surfaces, we point-wisely screened for the highest cluster frequencies and stored it in parallel with the respective cluster label. We combined these two types of data in a composite map. We repeated this procedure for the second highest frequencies in Europe. Major groups were assigned to Northern, Western and Eastern Europe. North Africa built a separate region, Southeastern Europe, Turkey and Near East were divided into several regions. The spatial distribution of the groups accounting for the second highest frequencies in Europe overlapped with the territories of the largest countries. The genetic structure presented in the composite maps fits major historical geopolitical regions and is in agreement with previous studies of genetic frequencies, validating our approach. Our genetic geostatistical approach provides, on the basis of two composite maps, detailed evidence of the geographical distribution and relative frequencies of the most predominant groups of the extant male European population, examined on the basis of forensic Y-STR haplotypes. The existence of considerable genetic differences among geographic subgroups in Europe has important consequences for the statistical inference in forensic Y-STR haplotype analyses.

Inferential testing for linkage with GENEHUNTER-MODSCORE: the impact of the pedigree structure on the null distribution of multipoint MOD scores.

The asymptotic distribution of [MOD] scores under the null hypothesis of no linkage is only known for affected sib pairs and other types of affected relative pairs. We have extended the GENEHUNTER-MODSCORE program to allow for simulations under the null hypothesis of no linkage to determine the empirical significance of MOD-score results in general situations. We performed simulations with families of different size (one million replicates of 500 families per simulation setting) to thoroughly investigate the impact of the pedigree size on the null distribution of multipoint MOD scores. It is shown that the distribution is dependent on the size and structure of the pedigrees under study. By performing simulations in the context of MOD-score analysis, our new tool efficiently explores the linkage data in a comprehensive way and also provides a valid method to inferentially test for linkage.

Susceptibility variants for male-pattern baldness on chromosome 20p11.

We carried out a genome-wide association study in 296 individuals with male-pattern baldness (androgenetic alopecia) and 347 controls. We then investigated the 30 best SNPs in an independent replication sample and found highly significant association for five SNPs on chromosome 20p11 (rs2180439 combined P = 2.7 x 10(-15)). No interaction was detected with the X-chromosomal androgen receptor locus, suggesting that the 20p11 locus has a role in a yet-to-be-identified androgen-independent pathway.

ISFG: Recommendations on biostatistics in paternity testing.

The Paternity Testing Commission (PTC) of the International Society for Forensic Genetics has taken up the task of establishing the biostatistical recommendations in accordance with the ISO 17025 standards and a previous set of ISFG recommendations specific to the genetic investigations in paternity cases. In the initial set, the PTC recommended that biostatistical evaluations of paternity are based on a likelihood ratio principle - yielding the paternity index, PI. Here, we have made five supplementary biostatistical recommendations. The first recommendation clarifies and defines basic concepts of genetic hypotheses and calculation concerns needed to produce valid PIs. The second and third recommendations address issues associated with population genetics (allele probabilities, Y-chromosome markers, mtDNA, and population substructuring) and special circumstances (deficiency/reconstruction and immigration cases), respectively. The fourth recommendation considers strategies regarding genetic evidence against paternity. The fifth recommendation covers necessary documentation, reporting details and assumptions underlying calculations. The PTC strongly suggests that these recommendations should be adopted by all laboratories involved in paternity testing as the basis for their biostatistical analysis.

Detection of parent-of-origin effects in nuclear families using haplotype analysis.

Despite the potential pitfalls of stratification, population-based association studies nowadays are being conducted more often than family-based association studies. However, the mechanism of genomic imprinting has lately been implicated in the etiology of genetic complex diseases and can be detected using statistics only in family-based designs. Powerful tests for association and imprinting have been proposed previously for case-parent trios and single markers. Since the power of association studies can be improved if multiple affected children and haplotypes are considered, we extended the parental asymmetry test (PAT) for imprinting to a test that is suited for both general nuclear families and haplotypes, called HAP-PAT. Significance of the HAP-PAT is determined via a Monte-Carlo simulation procedure. In addition to the HAP-PAT, we modified a haplotype-based association test, proposed by us before, in such a way that either only paternal or maternal transmissions contribute to the test statistic. The approaches were implemented in FAMHAP and we evaluated their performance under a variety of disease models. We were able to demonstrate the usefulness of our haplotype-based approaches to detect parent-of-origin effects. Furthermore, we showed that also in the presence of imprinting it is more reasonable to consider all affected children of a nuclear family, than to randomly select one affected child from each family and to conduct a trio study using the selected individuals.

SNP-based analysis of genetic substructure in the German population.

OBJECTIVE: To evaluate the relevance and necessity to account for the effects of population substructure on association studies under a case-control design in central Europe, we analysed three samples drawn from different geographic areas of Germany. Two of the three samples, POPGEN (n = 720) and SHIP (n = 709), are from north and north-east Germany, respectively, and one sample, KORA (n = 730), is from southern Germany. METHODS: Population genetic differentiation was measured by classical F-statistics for different marker sets, either consisting of genome-wide selected coding SNPs located in functional genes, or consisting of selectively neutral SNPs from 'genomic deserts'. Quantitative estimates of the degree of stratification were performed comparing the genomic control approach [Devlin B, Roeder K: Biometrics 1999;55:997-1004], structured association [Pritchard JK, Stephens M, Donnelly P: Genetics 2000;155:945-959] and sophisticated methods like random forests [Breiman L: Machine Learning 2001;45:5-32]. RESULTS: F-statistics showed that there exists a low genetic differentiation between the samples along a north-south gradient within Germany (F(ST)(KORA/POPGEN): 1.7 . 10(-4); F(ST)(KORA/SHIP): 5.4 . 10(-4); F(ST)(POPGEN/SHIP): -1.3 . 10(-5)). CONCLUSION: Although the F(ST )-values are very small, indicating a minor degree of population structure, and are too low to be detectable from methods without using prior information of subpopulation membership, such as STRUCTURE [Pritchard JK, Stephens M, Donnelly P: Genetics 2000;155:945-959], they may be a possible source for confounding due to population stratification.

Parent-of-origin, imprinting, mitochondrial, and X-linked effects in traits related to alcohol dependence: presentation Group 18 of Genetic Analysis Workshop 14.

The participants of Presentation Group 18 of Genetic Analysis Workshop 14 analyzed the Collaborative Study on the Genetics of Alcoholism data set to investigate sex-specific effects for phenotypes related to alcohol dependence. In particular, the participants looked at imprinting (which is also known as parent-of-origin effect), differences between recombination fractions for the two sexes, and mitochondrial and X-chromosomal effects. Five of the seven groups employed newly developed or existing methods that take imprinting into account when testing for linkage, or test for imprinting itself. Single-marker and multipoint analyses were performed for microsatellite as well as single-nucleotide polymorphism markers, and several groups used a sex-specific genetic map in addition to a sex-averaged map. Evidence for paternal imprinting (i.e., maternal expression) was consistently obtained by at least two groups at genetic regions on chromosomes 10, 12, and 21 that possibly harbor genes responsible for alcoholism. Evidence for maternal imprinting (which is equivalent to paternal expression) was consistently found at a locus on chromosome 11. Two groups applied extensions of variance components analysis that model a mitochondrial or X-chromosomal effect to latent class variables and electrophysiological traits employed in the diagnosis of alcoholism. The analysis, without using genetic markers, revealed mitochondrial or X-chromosomal effects for several of these traits. Accounting for sex-specific environmental variances appeared to be crucial for the identification of an X-chromosomal factor. In linkage analysis using marker data, modeling a mitochondrial variance component increased the linkage signals obtained for autosomal loci.

Haplotype interaction analysis of unlinked regions.

Genetically complex diseases are caused by interacting environmental factors and genes. As a consequence, statistical methods that consider multiple unlinked genomic regions simultaneously are desirable. Such consideration, however, may lead to a vast number of different high-dimensional tests whose appropriate analysis pose a problem. Here, we present a method to analyze case-control studies with multiple SNP data without phase information that considers gene-gene interaction effects while correcting appropriately for multiple testing. In particular, we allow for interactions of haplotypes that belong to different unlinked regions, as haplotype analysis often proves to be more powerful than single marker analysis. In addition, we consider different marker combinations at each unlinked region. The multiple testing issue is settled via the minP approach; the P value of the "best" marker/region configuration is corrected via Monte-Carlo simulations. Thus, we do not explicitly test for a specific pre-defined interaction model, but test for the global hypothesis that none of the considered haplotype interactions shows association with the disease. We carry out a simulation study for case-control data that confirms the validity of our approach. When simulating two-locus disease models, our test proves to be more powerful than association methods that analyze each linked region separately. In addition, when one of the tested regions is not involved in the etiology of the disease, only a small amount of power is lost with interaction analysis as compared to analysis without interaction. We successfully applied our method to a real case-control data set with markers from two genes controlling a common pathway. While classical analysis failed to reach significance, we obtained a significant result even after correction for multiple testing with our proposed haplotype interaction analysis. The method described here has been implemented in FAMHAP.

A recoding scheme for X-linked and pseudoautosomal loci to be used with computer programs for autosomal LOD-score analysis.

We present a recoding scheme that allows for a parametric multipoint X-chromosomal linkage analysis of dichotomous traits in the context of a computer program for autosomes that can use trait models with imprinting. Furthermore, with this scheme, it is possible to perform a joint multipoint analysis of X-linked and pseudoautosomal loci. It is required that (1) the marker genotypes of all female nonfounders are available and that (2) there are no male nonfounders who have daughters in the pedigree. The second requirement does not apply if the trait locus is pseudoautosomal. The X-linked marker loci are recorded by adding a dummy allele to the males' hemizygous genotypes. For modelling an X-linked trait locus, five different liability classes are defined, in conjunction with a paternal imprinting model for male nonfounders. The formulation aims at the mapping of a diallelic trait locus relative to an arbitrary number of codominant markers with known genetic distances, in cases where a program for a genuine X-chromosomal analysis is not available.

How to model a complex trait. 2. Analysis with two disease loci.

Complex traits are often governed by more than one trait locus. The first step towards an adequate model for such diseases is a linkage analysis with two trait loci. Such an analysis can be expected to have higher power to detect linkage than a standard single-trait-locus linkage analysis. However, it is crucial to accurately specify the parameters of the two-locus model. Here, we recapitulate the general two-locus model with and without genomic imprinting. We relate heterogeneity, multiplicative, and additive two-locus models to biological or pathophysiological mechanisms, and give the corresponding averaged ("best-fitting") single-trait-locus models for each of the two loci. Furthermore, we derive the two-locus penetrances from the averaged single-locus models, under the assumption of one of the three model classes mentioned above. Using these formulae, if the best-fitting single-locus models are available, investigators may perform a two-trait-locus linkage analysis under a realistic model. This procedure will maximize the power to detect linkage for traits which are governed by two or more loci, and lead to more accurate estimates of the disease-locus positions.

How to model a complex trait. 1. General considerations and suggestions.

Usually, when complex traits are at issue, not only are the loci of the responsible genes a priori unknown; the same also holds for the mode of inheritance of the trait, and sometimes even for the phenotype definition. The term mode of inheritance relates to both the genetic mechanism, i.e., the number of loci implicated in the etiology of the disease, and the genotype-phenotype relation, which describes the influence of these loci on the trait. Having an idea of the genetic model can crucially facilitate the mapping process. This holds especially in the context of linkage analysis, where an appropriate parametric model or a suitable nonparametric allele sharing statistic may accordingly be selected. Here, we review the difficulties with parametric and nonparametric linkage analysis when applied to multifactorial diseases. We address the question why it is necessary to adequately model a genetically complex trait in a linkage study, and elucidate the steps to do so. Furthermore, we discuss the value of including unaffected individuals into the analysis, as well as of looking at larger pedigrees, both with parametric and nonparametric methods. Our considerations and suggestions aim at guiding researchers to genotyping individuals at a trait locus as accurately as possible.

Genetic analysis of phenotypes derived from longitudinal data: Presentation Group 1 of Genetic Analysis Workshop 13.

The participants of Presentation Group 1 used the GAW13 data to derive new phenotypes, which were then analyzed for linkage and, in one case, for association to the genetic markers. Since the trait measurements ranged over longer time periods, the participants looked at the time dependence of particular traits in addition to the trait itself. The phenotypes analyzed with the Framingham data can be roughly divided into 1) body weight-related traits, which also include a type 2 diabetes progression trait, and 2) traits related to systolic blood pressure. Both trait classes are associated with metabolic syndrome. For traits related to body weight, linkage was consistently identified by at least two participating groups to genetic regions on chromosomes 4, 8, 11, and 18. For systolic blood pressure, or its derivatives, at least two groups obtained linkage for regions on chromosomes 4, 6, 8, 11, 14, 16, and 19. Five of the 13 participating groups focused on the simulated data. Due to the rather sparse grid of microsatellite markers, an association analysis for several traits was not successful. Linkage analysis of hypertension and body mass index using LODs and heterogeneity LODs (HLODs) had low power. For the glucose phenotype, a combination of random coefficient regression models and variance component linkage analysis turned out to be strikingly powerful in the identification of a trait locus simulated on chromosome 5. Haseman-Elston regression methods, applied to the same phenotype, had low power, but the above-mentioned chromosome 5 locus was not included in this analysis.

Mitochondrial DNA control region diversity in hairs and body fluids of monozygotic triplets.

Length heteroplasmy of the homopolymeric cytosine stretch in the hypervariable region II of the mitochondrial D-loop was investigated in blood, buccal cells and hair shafts of monozygotic triplets. The proportions of length heteroplasmy were determined by cloning and sequencing of multiple independent clones. Blood and buccal cells showed an accumulation of molecules with one and two insertions of cytosine residues in relation to the Cambridge Reference Sequence (CRS). The results did not show statistically significant differences between blood and buccal cells of one and the same individual and also not between the three monozygotic brothers. In the hair samples a loss of cytosine residues was established in all three monozygotic individuals compared to blood and buccal cells, suggesting that this must be a regular process. Furthermore, the hair shaft samples showed significant differences between the frequencies of 7, 8 or 9 Cs in the poly C region comparing the three individuals (p<0.008) and in addition there were highly significant differences (p<0.0001) when comparing the results for six different hairs of each individual separately. From these results it can be assumed that besides a common genetic bottleneck during embryonic development, a post-embryonic bottleneck seems to exist in each hair follicle.