Genomewide rapid association using mixed model and regression: a fast and simple method for genomewide pedigree-based quantitative trait loci association analysis

For pedigree-based quantitative trait loci (QTL) association analysis, a range of methods utilizing within-family variation such as transmission-disequilibrium test (TDT)-based methods have been developed. In scenarios where stratification is not a concern, methods exploiting between-family variation in addition to within-family variation, such as the measured genotype (MG) approach, have greater power. Application of MG methods can be computationally demanding (especially for large pedigrees), making genomewide scans practically infeasible. Here we suggest a novel approach for genomewide pedigree-based quantitative trait loci (QTL) association analysis: genomewide rapid association using mixed model and regression (GRAMMAR). The method first obtains residuals adjusted for family effects and subsequently analyzes the association between these residuals and genetic polymorphisms using rapid least-squares methods. At the final step, the selected polymorphisms may be followed up with the full measured genotype (MG) analysis. In a simulation study, we compared type 1 error, power, and operational characteristics of the proposed method with those of MG and TDT-based approaches. For moderately heritable (30%) traits in human pedigrees the power of the GRAMMAR and the MG approaches is similar and is much higher than that of TDT-based approaches. When using tabulated thresholds, the proposed method is less powerful than MG for very high heritabilities and pedigrees including large sibships like those observed in livestock pedigrees. However, there is little or no difference in empirical power of MG and the proposed method. In any scenario, GRAMMAR is much faster than MG and enables rapid analysis of hundreds of thousands of markers.     

A test for linkage and association in general pedigrees: the pedigree disequilibrium test

Family-based tests of linkage disequilibrium typically are based on nuclear-family data including affected individuals and their parents or their unaffected siblings. A limitation of such tests is that they generally are not valid tests of association when data from related nuclear families from larger pedigrees are used. Standard methods require selection of a single nuclear family from any extended pedigrees when testing for linkage disequilibrium. Often data are available for larger pedigrees, and it would be desirable to have a valid test of linkage disequilibrium that can use all potentially informative data. In this study, we present the pedigree disequilibrium test (PDT) for analysis of linkage disequilibrium in general pedigrees. The PDT can use data from related nuclear families from extended pedigrees and is valid even when there is population substructure. Using computer simulations, we demonstrated validity of the test when the asymptotic distribution is used to assess the significance, and examined statistical power. Power simulations demonstrate that, when extended pedigree data are available, substantial gains in power can be attained by use of the PDT rather than existing methods that use only a subset of the data. Furthermore, the PDT remains more powerful even when there is misclassification of unaffected individuals. Our simulations suggest that there may be advantages to using the PDT even if the data consist of independent families without extended family information. Thus, the PDT provides a general test of linkage disequilibrium that can be widely applied to different data structures.     

PRIMAL: Fast and Accurate Pedigree-based Imputation from Sequence Data in a Founder Population

Founder populations and large pedigrees offer many well-known advantages for genetic mapping studies, including cost-efficient study designs. Here, we describe PRIMAL (PedigRee IMputation ALgorithm), a fast and accurate pedigree-based phasing and imputation algorithm for founder populations. PRIMAL incorporates both existing and original ideas, such as a novel indexing strategy of Identity-By-Descent (IBD) segments based on clique graphs. We were able to impute the genomes of 1,317 South Dakota Hutterites, who had genome-wide genotypes for ~300,000 common single nucleotide variants (SNVs), from 98 whole genome sequences. Using a combination of pedigree-based and LD-based imputation, we were able to assign 87% of genotypes with >99% accuracy over the full range of allele frequencies. Using the IBD cliques we were also able to infer the parental origin of 83% of alleles, and genotypes of deceased recent ancestors for whom no genotype information was available. This imputed data set will enable us to better study the relative contribution of rare and common variants on human phenotypes, as well as parental origin effect of disease risk alleles in >1,000 individuals at minimal cost.     

Finding consistent gene transmission patterns on large and complex pedigrees

A heuristic algorithm for finding gene transmission patterns on large and complex pedigrees with partially observed genotype data is proposed. The method can be used to generate an initial point for a Markov chain Monte Carlo simulation or to check that the given pedigree and the genotype data are consistent. In small pedigrees, the algorithm is exact by exhaustively enumerating all possibilities, but, in large pedigrees, with a considerable amount of unknown data, only a subset of promising configurations can actually be checked. For that purpose, the configurations are ordered by combining the approximative conditional probability distribution of the unknown genotypes with the information on the relationships between individuals. We also introduce a way to divide the task into subparts, which has been shown to be useful in large pedigrees. The algorithm has been implemented in a program called APE (Allelic Path Explorer) and tested in three different settings with good results.     

A combined long-range phasing and long haplotype imputation method to impute phase for SNP genotypes

Background

Knowing the phase of marker genotype data can be useful in genome-wide association studies, because it makes it possible to use analysis frameworks that account for identity by descent or parent of origin of alleles and it can lead to a large increase in data quantities via genotype or sequence imputation. Long-range phasing and haplotype library imputation constitute a fast and accurate method to impute phase for SNP data.

Methods

A long-range phasing and haplotype library imputation algorithm was developed. It combines information from surrogate parents and long haplotypes to resolve phase in a manner that is not dependent on the family structure of a dataset or on the presence of pedigree information.

Results

The algorithm performed well in both simulated and real livestock and human datasets in terms of both phasing accuracy and computation efficiency. The percentage of alleles that could be phased in both simulated and real datasets of varying size generally exceeded 98% while the percentage of alleles incorrectly phased in simulated data was generally less than 0.5%. The accuracy of phasing was affected by dataset size, with lower accuracy for dataset sizes less than 1000, but was not affected by effective population size, family data structure, presence or absence of pedigree information, and SNP density. The method was computationally fast. In comparison to a commonly used statistical method (fastPHASE), the current method made about 8% less phasing mistakes and ran about 26 times faster for a small dataset. For larger datasets, the differences in computational time are expected to be even greater. A computer program implementing these methods has been made available.

Conclusions

The algorithm and software developed in this study make feasible the routine phasing of high-density SNP chips in large datasets.     

The impact of imputation on meta-analysis of genome-wide association studies

Genotype imputation is often used in the meta-analysis of genome-wide association studies (GWAS), for combining data from different studies and/or genotyping platforms, in order to improve the ability for detecting disease variants with small to moderate effects. However, how genotype imputation affects the performance of the meta-analysis of GWAS is largely unknown. In this study, we investigated the effects of genotype imputation on the performance of meta-analysis through simulations based on empirical data from the Framingham Heart Study. We found that when fix-effects models were used, considerable between-study heterogeneity was detected when causal variants were typed in only some but not all individual studies, resulting in up to ～25% reduction of detection power. For certain situations, the power of the meta-analysis can be even less than that of individual studies. Additional analyses showed that the detection power was slightly improved when between-study heterogeneity was partially controlled through the random-effects model, relative to that of the fixed-effects model. Our study may aid in the planning, data analysis, and interpretation of GWAS meta-analysis results when genotype imputation is necessary.     

Efficiency and power as a function of sequence coverage, SNP array density, and imputation

High coverage whole genome sequencing provides near complete information about genetic variation. However, other technologies can be more efficient in some settings by (a) reducing redundant coverage within samples and (b) exploiting patterns of genetic variation across samples. To characterize as many samples as possible, many genetic studies therefore employ lower coverage sequencing or SNP array genotyping coupled to statistical imputation. To compare these approaches individually and in conjunction, we developed a statistical framework to estimate genotypes jointly from sequence reads, array intensities, and imputation. In European samples, we find similar sensitivity (89%) and specificity (99.6%) from imputation with either 1× sequencing or 1 M SNP arrays. Sensitivity is increased, particularly for low-frequency polymorphisms (MAF < 5%), when low coverage sequence reads are added to dense genome-wide SNP arrays--the converse, however, is not true. At sites where sequence reads and array intensities produce different sample genotypes, joint analysis reduces genotype errors and identifies novel error modes. Our joint framework informs the use of next-generation sequencing in genome wide association studies and supports development of improved methods for genotype calling.     

Leveraging Identity-by-Descent for Accurate Genotype Inference in Family Sequencing Data

Sequencing family DNA samples provides an attractive alternative to population based designs to identify rare variants associated with human disease due to the enrichment of causal variants in pedigrees. Previous studies showed that genotype calling accuracy can be improved by modeling family relatedness compared to standard calling algorithms. Current family-based variant calling methods use sequencing data on single variants and ignore the identity-by-descent (IBD) sharing along the genome. In this study we describe a new computational framework to accurately estimate the IBD sharing from the sequencing data, and to utilize the inferred IBD among family members to jointly call genotypes in pedigrees. Through simulations and application to real data, we showed that IBD can be reliably estimated across the genome, even at very low coverage (e.g. 2X), and genotype accuracy can be dramatically improved. Moreover, the improvement is more pronounced for variants with low frequencies, especially at low to intermediate coverage (e.g. 10X to 20X), making our approach effective in studying rare variants in cost-effective whole genome sequencing in pedigrees. We hope that our tool is useful to the research community for identifying rare variants for human disease through family-based sequencing.     

Case-control association testing with related individuals: a more powerful quasi-likelihood score test

We consider the problem of genomewide association testing of a binary trait when some sampled individuals are related, with known relationships. This commonly arises when families sampled for a linkage study are included in an association study. Furthermore, power to detect association with complex traits can be increased when affected individuals with affected relatives are sampled, because they are more likely to carry disease alleles than are randomly sampled affected individuals. With related individuals, correlations among relatives must be taken into account, to ensure validity of the test, and consideration of these correlations can also improve power. We provide new insight into the use of pedigree-based weights to improve power, and we propose a novel test, the MQLS test, which, as we demonstrate, represents an overall, and in many cases, substantial, improvement in power over previous tests, while retaining a computational simplicity that makes it useful in genomewide association studies in arbitrary pedigrees. Other features of the MQLS are as follows: (1) it is applicable to completely general combinations of family and case-control designs, (2) it can incorporate both unaffected controls and controls of unknown phenotype into the same analysis, and (3) it can incorporate phenotype data about relatives with missing genotype data. The methods are applied to data from the Genetic Analysis Workshop 14 Collaborative Study of the Genetics of Alcoholism, where the MQLS detects genomewide significant association (after Bonferroni correction) with an alcoholism-related phenotype for four different single-nucleotide polymorphisms: tsc1177811 (P=5.9x10(-7)), tsc1750530 (P=4.0x10(-7)), tsc0046696 (P=4.7x10(-7)), and tsc0057290 (P=5.2x10(-7)) on chromosomes 1, 16, 18, and 18, respectively. Three of these four significant associations were not detected in previous studies analyzing these data.     

Haplotype association analyses in resources of mixed structure using Monte Carlo testing

Background  

Genomewide association studies have resulted in a great many genomic regions that are likely to harbor disease genes. Thorough interrogation of these specific regions is the logical next step, including regional haplotype studies to identify risk haplotypes upon which the underlying critical variants lie. Pedigrees ascertained for disease can be powerful for genetic analysis due to the cases being enriched for genetic disease. Here we present a Monte Carlo based method to perform haplotype association analysis. Our method, hapMC, allows for the analysis of full-length and sub-haplotypes, including imputation of missing data, in resources of nuclear families, general pedigrees, case-control data or mixtures thereof. Both traditional association statistics and transmission/disequilibrium statistics can be performed. The method includes a phasing algorithm that can be used in large pedigrees and optional use of pseudocontrols.     

Methods and results from the genome-wide association group at GAW20

Background

This paper summarizes the contributions from the Genome-wide Association Study group (GWAS group) of the GAW20. The GWAS group contributions focused on topics such as association tests, phenotype imputation, and application of empirical kinships. The goals of the GWAS group contributions were varied. A real or a simulated data set based on the Genetics of Lipid Lowering Drugs and Diet Network (GOLDN) study was employed by different methods. Different outcomes and covariates were considered, and quality control procedures varied throughout the contributions.

Results

The consideration of heritability and family structure played a major role in some contributions. The inclusion of family information and adaptive weights based on data were found to improve power in genome-wide association studies. It was proven that gene-level approaches are more powerful than single-marker analysis. Other contributions focused on the comparison between pedigree-based kinship and empirical kinship matrices, and investigated similar results in heritability estimation, association mapping, and genomic prediction. A new approach for linkage mapping of triglyceride levels was able to identify a novel linkage signal.

Conclusions

This summary paper reports on promising statistical approaches and findings of the members of the GWAS group applied on real and simulated data which encompass the current topics of epigenetic and pharmacogenomics.

     

Practical Issues in Imputation-Based Association Mapping

Imputation-based association methods provide a powerful framework for testing untyped variants for association with phenotypes and for combining results from multiple studies that use different genotyping platforms. Here, we consider several issues that arise when applying these methods in practice, including: (i) factors affecting imputation accuracy, including choice of reference panel; (ii) the effects of imputation accuracy on power to detect associations; (iii) the relative merits of Bayesian and frequentist approaches to testing imputed genotypes for association with phenotype; and (iv) how to quickly and accurately compute Bayes factors for testing imputed SNPs. We find that imputation-based methods can be robust to imputation accuracy and can improve power to detect associations, even when average imputation accuracy is poor. We explain how ranking SNPs for association by a standard likelihood ratio test gives the same results as a Bayesian procedure that uses an unnatural prior assumption—specifically, that difficult-to-impute SNPs tend to have larger effects—and assess the power gained from using a Bayesian approach that does not make this assumption. Within the Bayesian framework, we find that good approximations to a full analysis can be achieved by simply replacing unknown genotypes with a point estimate—their posterior mean. This approximation considerably reduces computational expense compared with published sampling-based approaches, and the methods we present are practical on a genome-wide scale with very modest computational resources (e.g., a single desktop computer). The approximation also facilitates combining information across studies, using only summary data for each SNP. Methods discussed here are implemented in the software package BIMBAM, which is available from http://stephenslab.uchicago.edu/software.html.     

Imputation of non-genotyped sheep from the genotypes of their mates and resulting progeny

Accurate genomic analyses are predicated on access to a large quantity of accurately genotyped and phenotyped animals. Because the cost of genotyping is often less than the cost of phenotyping, interest is increasing in generating genotypes for phenotyped animals. In some instances this may imply the requirement to genotype older animals with greater phenotypic information content. Biological material for these older informative animals may, however, no longer exist. The objective of the present study was to quantify the ability to impute 11 129 single nucleotide polymorphism (SNP) genotypes of non-genotyped animals (in this instance sires) from the genotypes of their progeny with or without including the genotypes of the progenys’ dams (i.e. mates of the sire to be imputed). The impact on the accuracy of genotype imputation by including more progeny (and their dams’) genotypes in the imputation reference population was also quantified. When genotypes of the dams were not available, genotypes of 41 sires with at least 15 genotyped progeny were used for the imputation; when genotypes of the dams were available, genotypes of 21 sires with at least 10 genotyped progeny were used for the imputation. Imputation was undertaken exploiting family and population level information. The mean and variability in the proportion of genotypes per individual that could not be imputed reduced as the number of progeny genotypes used per individual increased. Little improvement in the proportion of genotypes that could not be imputed was achieved once genotypes of seven progeny and their dams were used or genotypes of 11 progeny without their respective dam’s genotypes were used. Mean imputation accuracy per individual (depicted by both concordance rates and correlation between true and imputed) increased with increasing progeny group size. Moreover, the range in mean imputation accuracy per individual reduced as more progeny genotypes were used in the imputation. If the genotype of the mate of the sire was also used, high accuracy of imputation (mean genotype concordance rate per individual of 0.988), with little additional benefit thereafter, was achieved with seven genotyped progeny. In the absence of genotypes on the dam, similar imputation accuracy could not be achieved even using genotypes on up to 15 progeny. Results therefore suggest, at least for the SNP density used in the present study, that it is possible to accurately impute the genotypes of a non-genotyped parent from the genotypes of its progeny and there is a benefit of also including the genotype of the sire’s mate (i.e. dam of the progeny).     

Imputation of genotypes in Danish purebred and two-way crossbred pigs using low-density panels

Background

Genotype imputation is commonly used as an initial step in genomic selection since the accuracy of genomic selection does not decline if accurately imputed genotypes are used instead of actual genotypes but for a lower cost. Performance of imputation has rarely been investigated in crossbred animals and, in particular, in pigs. The extent and pattern of linkage disequilibrium differ in crossbred versus purebred animals, which may impact the performance of imputation. In this study, first we compared different scenarios of imputation from 5 K to 8 K single nucleotide polymorphisms (SNPs) in genotyped Danish Landrace and Yorkshire and crossbred Landrace-Yorkshire datasets and, second, we compared imputation from 8 K to 60 K SNPs in genotyped purebred and simulated crossbred datasets. All imputations were done using software Beagle version 3.3.2. Then, we investigated the reasons that could explain the differences observed.

Results

Genotype imputation performs as well in crossbred animals as in purebred animals when both parental breeds are included in the reference population. When the size of the reference population is very large, it is not necessary to use a reference population that combines the two breeds to impute the genotypes of purebred animals because a within-breed reference population can provide a very high level of imputation accuracy (correct rate ≥ 0.99, correlation ≥ 0.95). However, to ensure that similar imputation accuracies are obtained for crossbred animals, a reference population that combines both parental purebred animals is required. Imputation accuracies are higher when a larger proportion of haplotypes are shared between the reference population and the validation (imputed) populations.

Conclusions

The results from both real data and pedigree-based simulated data demonstrate that genotype imputation from low-density panels to medium-density panels is highly accurate in both purebred and crossbred pigs. In crossbred pigs, combining the parental purebred animals in the reference population is necessary to obtain high imputation accuracy.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-015-0134-4) contains supplementary material, which is available to authorized users.     

The mutation rate in the human mtDNA control region

The mutation rate of the mitochondrial control region has been widely used to calibrate human population history. However, estimates of the mutation rate in this region have spanned two orders of magnitude. To readdress this rate, we sequenced the mtDNA control region in 272 individuals, who were related by a total of 705 mtDNA transmission events, from 26 large Icelandic pedigrees. Three base substitutions were observed, and the mutation rate across the two hypervariable regions was estimated to be 3/705 =.0043 per generation (95% confidence interval [CI].00088-.013), or.32/site/1 million years (95% CI.065-.97). This study is substantially larger than others published, which have directly assessed mtDNA mutation rates on the basis of pedigrees, and the estimated mutation rate is intermediate among those derived from pedigree-based studies. Our estimated rate remains higher than those based on phylogenetic comparisons. We discuss possible reasons for-and consequences of-this discrepancy. The present study also provides information on rates of insertion/deletion mutations, rates of heteroplasmy, and the reliability of maternal links in the Icelandic genealogy database.     

Estimation of heritability from limited family data using genome-wide identity-by-descent sharing

Background

In classical pedigree-based analysis, additive genetic variance is estimated from between-family variation, which requires the existence of larger phenotyped and pedigreed populations involving numerous families (parents). However, estimation is often complicated by confounding of genetic and environmental family effects, with the latter typically occurring among full-sibs. For this reason, genetic variance is often inferred based on covariance among more distant relatives, which reduces the power of the analysis. This simulation study shows that genome-wide identity-by-descent sharing among close relatives can be used to quantify additive genetic variance solely from within-family variation using data on extremely small family samples.

Methods

Identity-by-descent relationships among full-sibs were simulated assuming a genome size similar to that of humans (effective number of loci ~80). Genetic variance was estimated from phenotypic data assuming that genomic identity-by-descent relationships could be accurately re-created using information from genome-wide markers. The results were compared with standard pedigree-based genetic analysis.

Results

For a polygenic trait and a given number of phenotypes, the most accurate estimates of genetic variance were based on data from a single large full-sib family only. Compared with classical pedigree-based analysis, the proposed method is more robust to selection among parents and for confounding of environmental and genetic effects. Furthermore, in some cases, satisfactory results can be achieved even with less ideal data structures, i.e., for selectively genotyped data and for traits for which the genetic variance is largely under the control of a few major genes.

Conclusions

Estimation of genetic variance using genomic identity-by-descent relationships is especially useful for studies aiming at estimating additive genetic variance of highly fecund species, using data from small populations with limited pedigree information and/or few available parents, i.e., parents originating from non-pedigreed or even wild populations.     

Comparison of genotype imputation strategies using a combined reference panel for chicken population

Using whole-genome sequence (WGS) data are supposed to be optimal for genome-wide association studies and genomic predictions. However, sequencing thousands of individuals of interest is expensive. Imputation from single nucleotide polymorphisms panels to WGS data is an attractive approach to obtain highly reliable WGS data at low cost. Here, we conducted a genotype imputation study with a combined reference panel in yellow-feather dwarf broiler population. The combined reference panel was assembled by sequencing 24 key individuals of a yellow-feather dwarf broiler population (internal reference panel) and WGS data from 311 chickens in public databases (external reference panel). Three scenarios were investigated to determine how different factors affect the accuracy of imputation from 600 K array data to WGS data, including: genotype imputation with internal, external and combined reference panels; the number of internal reference individuals in the combined reference panel; and different reference sizes and selection strategies of an external reference panel. Results showed that imputation accuracy from 600 K to WGS data were 0.834±0.012, 0.920±0.007 and 0.982±0.003 for the internal, external and combined reference panels, respectively. Increasing the reference size from 50 to 250 improved the accuracy of genotype imputation from 0.848 to 0.974 for the combined reference panel and from 0.647 to 0.917 for the external reference panel. The selection strategies for the external reference panel had no impact on the accuracy of imputation using the combined reference panel. However, if only an external reference panel with reference size >50 was used, the selection strategy of minimizing the average distance to the closest leaf had the greatest imputation accuracy compared with other methods. Generally, using a combined reference panel provided greater imputation accuracy, especially for low-frequency variants. In conclusion, the optimal imputation strategy with a combined reference panel should comprehensively consider genetic diversity of the study population, availability and properties of external reference panels, sequencing and computing costs, and frequency of imputed variants. This work sheds light on how to design and execute genotype imputation with a combined external reference panel in a livestock population.     

A Unified Approach to Genotype Imputation and Haplotype-Phase Inference for Large Data Sets of Trios and Unrelated Individuals

We present methods for imputing data for ungenotyped markers and for inferring haplotype phase in large data sets of unrelated individuals and parent-offspring trios. Our methods make use of known haplotype phase when it is available, and our methods are computationally efficient so that the full information in large reference panels with thousands of individuals is utilized. We demonstrate that substantial gains in imputation accuracy accrue with increasingly large reference panel sizes, particularly when imputing low-frequency variants, and that unphased reference panels can provide highly accurate genotype imputation. We place our methodology in a unified framework that enables the simultaneous use of unphased and phased data from trios and unrelated individuals in a single analysis. For unrelated individuals, our imputation methods produce well-calibrated posterior genotype probabilities and highly accurate allele-frequency estimates. For trios, our haplotype-inference method is four orders of magnitude faster than the gold-standard PHASE program and has excellent accuracy. Our methods enable genotype imputation to be performed with unphased trio or unrelated reference panels, thus accounting for haplotype-phase uncertainty in the reference panel. We present a useful measure of imputation accuracy, allelic R², and show that this measure can be estimated accurately from posterior genotype probabilities. Our methods are implemented in version 3.0 of the BEAGLE software package.     

Accuracy of genotype imputation in Labrador Retrievers

The dog is a valuable model species for the genetic analysis of complex traits, and the use of genotype imputation in dogs will be an important tool for future studies. It is of particular interest to analyse the effect of factors like single nucleotide polymorphism (SNP) density of genotyping arrays and relatedness between dogs on imputation accuracy due to the acknowledged genetic and pedigree structure of dog breeds. In this study, we simulated different genotyping strategies based on data from 1179 Labrador Retriever dogs. The study involved 5826 SNPs on chromosome 1 representing the high density (HighD) array; the low‐density (LowD) array was simulated by masking different proportions of SNPs on the HighD array. The correlations between true and imputed genotypes for a realistic masking level of 87.5% ranged from 0.92 to 0.97, depending on the scenario used. A correlation of 0.92 was found for a likely scenario (10% of dogs genotyped using HighD, 87.5% of HighD SNPs masked in the LowD array), which indicates that genotype imputation in Labrador Retrievers can be a valuable tool to reduce experimental costs while increasing sample size. Furthermore, we show that genotype imputation can be performed successfully even without pedigree information and with low relatedness between dogs in the reference and validation sets. Based on these results, the impact of genotype imputation was evaluated in a genome‐wide association analysis and genomic prediction in Labrador Retrievers.     

Statistical Inference for Hardy-Weinberg Proportions in the Presence of Missing Genotype Information

In genetic association studies, tests for Hardy-Weinberg proportions are often employed as a quality control checking procedure. Missing genotypes are typically discarded prior to testing. In this paper we show that inference for Hardy-Weinberg proportions can be biased when missing values are discarded. We propose to use multiple imputation of missing values in order to improve inference for Hardy-Weinberg proportions. For imputation we employ a multinomial logit model that uses information from allele intensities and/or neighbouring markers. Analysis of an empirical data set of single nucleotide polymorphisms possibly related to colon cancer reveals that missing genotypes are not missing completely at random. Deviation from Hardy-Weinberg proportions is mostly due to a lack of heterozygotes. Inbreeding coefficients estimated by multiple imputation of the missings are typically lowered with respect to inbreeding coefficients estimated by discarding the missings. Accounting for missings by multiple imputation qualitatively changed the results of 10 to 17% of the statistical tests performed. Estimates of inbreeding coefficients obtained by multiple imputation showed high correlation with estimates obtained by single imputation using an external reference panel. Our conclusion is that imputation of missing data leads to improved statistical inference for Hardy-Weinberg proportions.