HIBAG--HLA genotype imputation with attribute bagging.

Genotyping of classical human leukocyte antigen (HLA) alleles is an essential tool in the analysis of diseases and adverse drug reactions with associations mapping to the major histocompatibility complex (MHC). However, deriving high-resolution HLA types subsequent to whole-genome single-nucleotide polymorphism (SNP) typing or sequencing is often cost prohibitive for large samples. An alternative approach takes advantage of the extended haplotype structure within the MHC to predict HLA alleles using dense SNP genotypes, such as those available from genome-wide SNP panels. Current methods for HLA imputation are difficult to apply or may require the user to have access to large training data sets with SNP and HLA types. We propose HIBAG, HLA Imputation using attribute BAGging, that makes predictions by averaging HLA-type posterior probabilities over an ensemble of classifiers built on bootstrap samples. We assess the performance of HIBAG using our study data (n=2668 subjects of European ancestry) as a training set and HLA data from the British 1958 birth cohort study (n≈1000 subjects) as independent validation samples. Prediction accuracies for HLA-A, B, C, DRB1 and DQB1 range from 92.2% to 98.1% using a set of SNP markers common to the Illumina 1M Duo, OmniQuad, OmniExpress, 660K and 550K platforms. HIBAG performed well compared with the other two leading methods, HLA*IMP and BEAGLE. This method is implemented in a freely available HIBAG R package that includes pre-fit classifiers for European, Asian, Hispanic and African ancestries, providing a readily available imputation approach without the need to have access to large training data sets.

Genome-wide approaches to identify pharmacogenetic contributions to adverse drug reactions.

Adverse drug reactions (ADRs) have a major impact on patients, physicians, health care providers, regulatory agencies and pharmaceutical companies. Identifying the genetic contributions to ADR risk may lead to a better understanding of the underlying mechanisms, identification of patients at risk and a decrease in the number of events. Technological advances have made the routine monitoring and investigation of the genetic basis of ADRs during clinical trials possible. We demonstrate through simulation that genome-wide genotyping, coupled with the use of clinically matched or population controls, can yield sufficient statistical power to permit the identification of strong genetic predictors of ADR risk in a prospective manner with modest numbers of ADR cases. The results of a 500,000 single nucleotide polymorphism analysis of abacavir-associated hypersensitivity reaction suggest that the known HLA-B gene region could be identified with as few as 15 cases and 200 population controls in a sequential analysis.

Linkage disequilibrium mapping identifies a 390 kb region associated with CYP2D6 poor drug metabolising activity.

The cytochrome p450 enzyme, CYP2D6, metabolises approximately 20% of marketed drugs. CYP2D6 multiple variants are associated with altered enzyme activities. Genotyping 1018 Caucasians for CYP2D6 polymorphisms (G1846A, delT1707, delA2549 and A2935C), known to result in the recessive CYP2D6 poor drug metaboliser (PM) phenotype, identified 41 individuals with predicted PM phenotype. These 41 individuals were classified as 'cases'. Single nucleotide polymorphisms (SNPs) mapping within an 880 kb region flanking CYP2D6, were identified to evaluate potential association between genetic variation and the CYP2D6 PM phenotype. The 41 PM cases and 977 controls were genotyped and analysed for 27 SNPs. Associations were observed across a 390 kb region between 14 SNPs and the PM phenotype (P values from 6.20 x 10(-4) to 4.54 x 10(-35)). Haplotype analysis revealed more significant levels of association (P = 3.54 x 10(-56)). Strong (D' > 0.7) linkage disequilibrium (LD) between SNPs was observed across the same 390 kb region associated with the CYP2D6 phenotype. The observed phenotype:genotype association reached genome-wide levels of significance, and supports the strategy for potential application of LD mapping and whole genome association scans to pharmacogenetic studies.

Migraine with aura susceptibility locus on chromosome 19p13 is distinct from the familial hemiplegic migraine locus.

Migraine is a common neurological disease with a major genetic component. Recently, it has been proposed that a single locus on chromosome 19p13 contributes to the genetic susceptibility of both rare familial hemiplegic migraine (FHM) and more common types of migraine, migraine with aura and migraine without aura. We analyzed 16 families for co-segregation of migraine with aura and chromosome 19p13 markers. Using multipoint model-free linkage analysis, we obtained a lod score of 4.28 near D19S592. Using an affecteds-only model of linkage, we observed a lod score of 4.79 near D19S592. We were able to provide statistical evidence that this locus on chromosome 19p13 is most likely not the gene CACNA1A, mutations in which cause FHM. These data indicate that chromosome 19p13 contains a locus which contributes to the genetic susceptibility of migraine with aura that is distinct from the FHM locus.

Genome-wide scan of obesity in the Old Order Amish.

To identify the genetic determinants of typical obesity, we performed a genome-wide scan of obesity-related traits using data from the Amish. Multipoint linkage analysis was performed using a variance components procedure on body mass index (BMI), waist circumference, percentage of body fat, and serum leptin concentrations. All 672 individuals were genotyped for 357 markers in 22 autosomes. We observed modest evidence for linkage, with the maximum log odds (lod) scores for linkage for these traits occurring on chromosomes 3p (percentage of body fat: lod = 1.61, near the peroxisome proliferator-activated receptor-alpha gene), 14q (waist: lod = 1.80), and 16p (leptin: lod = 1.72; BMI: lod = 1.68). We also tested for linkage to BMI-adjusted leptin concentrations and observed suggestive evidence for linkage on chromosome 10p (lod = 2.73), approximately 10-20 cM telomeric from obesity loci previously reported in French and German Caucasians. Two additional linkage signals for this trait were observed on chromosomes 7q (lod = 1.77, approximately 20 cM from the leptin gene) and 14q (lod = 2.47). Follow-up studies may be warranted to pursue some of these linkage signals, especially those detected near known obesity candidate genes, and those in regions coinciding with linkage signals reported previously.

Identifying susceptibility genes using linkage and linkage disequilibrium analysis in large pedigrees.

Linkage and linkage disequilibrium tests are powerful tools for mapping complex disease genes. We investigated two approaches to identifying markers associated with disease. One method applied linkage analysis and then linkage disequilibrium tests to markers within linked regions. The other method looked for linkage disequilibrium with disease using all markers. Additionally, we investigated using Simes' test to combine p-values from linkage disequilibrium tests for nearby markers. We applied both approaches to all replicates of the Genetic Analysis Workshop 12 problem 2 isolated population data set. We reported results from the 25th replicate as if it were a real problem and assessed the power of our methods using all replicates. Using all replicates, we found that testing all markers for linkage disequilibrium with disease was more powerful than identifying markers that were in linkage with disease and then testing markers within those regions for linkage disequilibrium with the implementations that we chose. Using Simes' test to combine p-values for linkage disequilibrium tests on correlated markers seemed to be of marginal value.

Genomewide search for type 2 diabetes susceptibility genes in four American populations.

Type 2 diabetes is a serious, genetically influenced disease for which no fully effective treatments are available. Identification of biochemical or regulatory pathways involved in the disease syndrome could lead to innovative therapeutic interventions. One way to identify such pathways is the genetic analysis of families with multiple affected members where disease predisposing genes are likely to be segregating. We undertook a genomewide screen (389-395 microsatellite markers) in samples of 835 white, 591 Mexican American, 229 black, and 128 Japanese American individuals collected as part of the American Diabetes Association's GENNID study. Multipoint nonparametric linkage analyses were performed with diabetes, and diabetes or impaired glucose homeostasis (IH). Linkage to diabetes or IH was detected near markers D5S1404 (map position 77 cM, LOD = 2.80), D12S853 (map position 82 cM, LOD = 2.81) and GATA172D05 (X-chromosome map position 130 cM, LOD = 2.99) in whites, near marker D3S2432 (map position 51 cM, LOD = 3.91) in Mexican Americans, and near marker D10S1412 (map position 14 cM, LOD = 2.39) in African Americans mainly collected in phase 1 of the study. Further analyses showed evidence for interactions between the chromosome 5 locus and region on chromosome 12 containing the MODY 3 gene (map position 132 cM) and between the X-chromosome locus and region near D12S853 (map position 82 cM) in whites. Although these results were not replicated in samples collected in phase 2 of the GENNID study, the region on chromosome 12 was replicated in samples from whites described by Bektas et al. (1999).

Search for susceptibility genes, gene x gene interactions, and gene x environment interactions utilizing nonparametric linkage analysis.

There may be a different genetic basis for the two forms of the disease simulated in Problem 2 and the complex disease may be affected by environmental factors. Hence, we investigate the effects of two environmental factors. We selected 400 nuclear families from the data generated for Problem 2. Affection status was investigated in several ways. Individuals with severe and mild forms of the disease were both considered affected, individuals with only the severe form were considered affected while those with the mild form were considered unknown, and individuals with only the mild form were considered affected while those with the severe form were considered unknown. We found evidence of linkage between putative disease loci and markers in the D3G042-D3G049 and D5G031-D5G042 regions when we considered severely and mildly affected individuals as affected and also in the region D1G004-D1G013 when mildly affected individuals were considered unknown. We observed interactions between the first environmental factor and D1G043 among healthy sib pairs.

Detecting marker-disease association by testing for Hardy-Weinberg disequilibrium at a marker locus.

We review and extend a recent suggestion that fine-scale localization of a disease-susceptibility locus for a complex disease be done on the basis of deviations from Hardy-Weinberg equilibrium among affected individuals. This deviation is driven by linkage disequilibrium between disease and marker loci in the whole population and requires a heterogeneous genetic basis for the disease. A finding of marker-locus Hardy-Weinberg disequilibrium therefore implies disease heterogeneity and marker-disease linkage disequilibrium. Although a lack of departure of Hardy-Weinberg disequilibrium at marker loci implies that disease susceptibilityweighted linkage disequilibria are zero, given disease heterogeneity, it does not follow that the usual measures of linkage disequilibrium are zero. For disease-susceptibility loci with more than two alleles, therefore, care is needed in the drawing of inferences from marker Hardy-Weinberg disequilibria.

An autosomal genomic scan for loci linked to type II diabetes mellitus and body-mass index in Pima Indians.

Genetic factors influence the development of type II diabetes mellitus, but genetic loci for the most common forms of diabetes have not been identified. A genomic scan was conducted to identify loci linked to diabetes and body-mass index (BMI) in Pima Indians, a Native American population with a high prevalence of type II diabetes. Among 264 nuclear families containing 966 siblings, 516 autosomal markers with a median distance between adjacent markers of 6.4 cM were genotyped. Variance-components methods were used to test for linkage with an age-adjusted diabetes score and with BMI. In multipoint analyses, the strongest evidence for linkage with age-adjusted diabetes (LOD = 1.7) was on chromosome 11q, in the region that was also linked most strongly with BMI (LOD = 3.6). Bivariate linkage analyses strongly rejected both the null hypothesis of no linkage with either trait and the null hypothesis of no contribution of the locus to the covariation among the two traits. Sib-pair analyses suggest additional potential diabetes-susceptibility loci on chromosomes 1q and 7q.

An autosomal genomic scan for loci linked to prediabetic phenotypes in Pima Indians.

Type 2 diabetes mellitus is a common chronic disease that is thought to have a substantial genetic basis. Identification of the genes responsible has been hampered by the complex nature of the syndrome. Abnormalities in insulin secretion and insulin action predict the development of type 2 diabetes and are, themselves, highly heritable traits. Since fewer genes may contribute to these precursors of type 2 diabetes than to the overall syndrome, such genes may be easier to identify. We, therefore, undertook an autosomal genomic scan to identify loci linked to prediabetic traits in Pima Indians, a population with a high prevalence of type 2 diabetes. 363 nondiabetic Pima Indians were genotyped at 516 polymorphic microsatellite markers on all 22 autosomes. Linkage analyses were performed using three methods (single-marker, nonparametric multipoint [MAPMAKER/SIBS], and variance components multipoint). These analyses provided evidence for linkage at several chromosomal regions, including 3q21-24 linked to fasting plasma insulin concentration and in vivo insulin action, 4p15-q12 linked to fasting plasma insulin concentration, 9q21 linked to 2-h insulin concentration during oral glucose tolerance testing, and 22q12-13 linked to fasting plasma glucose concentration. These results suggest loci that may harbor genes contributing to type 2 diabetes in Pima Indians. None of the linkages exceeded a LOD score of 3.6 (a 5% probability of occurring in a genome-wide scan). These findings must, therefore, be considered tentative until extended in this population or replicated in others.

Autosomal genomic scan for loci linked to obesity and energy metabolism in Pima Indians.

An autosomal genomic scan to search for linkage to obesity and energy metabolism was completed in Pima Indians, a population prone to obesity. Obesity was assessed by percent body fat (by hydrodensitometry) and fat distribution (the ratio of waist circumference to thigh circumference). Energy metabolism was measured in a respiratory chamber as 24-h metabolic rate, sleeping metabolic rate, and 24-h respiratory quotient (24RQ), an indicator of the ratio of carbohydrate oxidation to fat oxidation. Five hundred sixteen microsatellite markers with a median spacing of 6.4 cM were analyzed, in 362 siblings who had measurements of body composition and in 220 siblings who had measurements of energy metabolism. These comprised 451 sib pairs in 127 nuclear families, for linkage analysis to obesity, and 236 sib pairs in 82 nuclear families, for linkage analysis to energy metabolism. Pointwise and multipoint methods for regression of sib-pair differences in identity by descent, as well as a sibling-based variance-components method, were used to detect linkage. LOD scores >=2 were found at 11q21-q22, for percent body fat (LOD=2.1; P=.001), at 11q23-q24, for 24-h energy expenditure (LOD=2.0; P=.001), and at 1p31-p21 (LOD=2.0) and 20q11.2 (LOD=3.0; P=.0001), for 24RQ, by pointwise and multipoint analyses. With the variance-components method, the highest LOD score (LOD=2.3 P=.0006) was found at 18q21, for percent body fat, and at 1p31-p21 (LOD=2.8; P=.0003), for 24RQ. Possible candidate genes include LEPR (leptin receptor), at 1p31, and ASIP (agouti-signaling protein), at 20q11.2.

Error detection for genetic data, using likelihood methods.

As genetic maps become denser, the effect of laboratory typing errors becomes more serious. We review a general method for detecting errors in pedigree genotyping data that is a variant of the likelihood-ratio test statistic. It pinpoints individuals and loci with relatively unlikely genotypes. Power and significance studies using Monte Carlo methods are shown by using simulated data with pedigree structures similar to the CEPH pedigrees and a larger experimental pedigree used in the study of idiopathic dilated cardiomyopathy (DCM). The studies show the index detects errors for small values of theta with high power and an acceptable false positive rate. The method was also used to check for errors in DCM laboratory pedigree data and to estimate the error rate in CEPH-chromosome 6 data. The errors flagged by our method in the DCM pedigree were confirmed by the laboratory. The results are consistent with estimated false-positive and false-negative rates obtained using simulation.