The hazards of genotype imputation when mapping disease susceptibility variants.

BACKGROUND: The cost-free increase in statistical power of using imputation to infer missing genotypes is undoubtedly appealing, but is it hazard-free? This case study of three type-2 diabetes (T2D) loci demonstrates that it is not; it sheds light on why this is so and raises concerns as to the shortcomings of imputation at disease loci, where haplotypes differ between cases and reference panel. RESULTS: T2D-associated variants were previously identified using targeted sequencing. We removed these significantly associated SNPs and used neighbouring SNPs to infer them by imputation. We compared imputed with observed genotypes, examined the altered pattern of T2D-SNP association, and investigated the cause of imputation errors by studying haplotype structure. Most T2D variants were incorrectly imputed with a low density of scaffold SNPs, but the majority failed to impute even at high density, despite obtaining high certainty scores. Missing and discordant imputation errors, which were observed disproportionately for the risk alleles, produced monomorphic genotype calls or false-negative associations. We show that haplotypes carrying risk alleles are considerably more common in the T2D cases than the reference panel, for all loci. CONCLUSIONS: Imputation is not a panacea for fine mapping, nor for meta-analysing multiple GWAS based on different arrays and different populations. A total of 80% of the SNPs we have tested are not included in array platforms, explaining why these and other such associated variants may previously have been missed. Regardless of the choice of software and reference haplotypes, imputation drives genotype inference towards the reference panel, introducing errors at disease loci.

The hazards of genotype imputation in chromosomal regions under selection: A case study using the Lactase gene region.

Although imputation of missing SNP results has been widely used in genetic studies, claims about the quality and usefulness of imputation have outnumbered the few studies that have questioned its limitations. But it is becoming clear that these limitations are real-for example, disease association signals can be missed in regions of LD breakdown. Here, as a case study, using the chromosomal region of the well-known lactase gene, LCT, we address the issue of imputation in the context of variants that have become frequent in a limited number of modern population groups only recently, due to selection. We study SNPs in a 500 bp region covering the enhancer of LCT, and compare imputed genotypes with directly genotyped data. We examine the haplotype pairs of all individuals with discrepant and missing genotypes. We highlight the nonrandom nature of the allelic errors and show that most incorrect imputations and missing data result from long haplotypes that are evolutionarily closely related to those carrying the derived alleles, while some relate to rare and recombinant haplotypes. We conclude that bias of incorrectly imputed and missing genotypes can decrease the accuracy of imputed results substantially.

New Insights Into Mitochondrial Dysfunction at Disease Susceptibility Loci in the Development of Type 2 Diabetes.

This study investigated the potential genetic mechanisms which underlie adipose tissue mitochondrial dysfunction in Type 2 diabetes (T2D), by systematically identifying nuclear-encoded mitochondrial genes (NEMGs) among the genes regulated by T2D-associated genetic loci. The target genes of these 'disease loci' were identified by mapping genetic loci associated with both disease and gene expression levels (expression quantitative trait loci, eQTL) using high resolution genetic maps, with independent estimates co-locating to within a small genetic distance. These co-locating signals were defined as T2D-eQTL and the target genes as T2D cis-genes. In total, 763 cis-genes were associated with T2D-eQTL, of which 50 were NEMGs. Independent gene expression datasets for T2D and insulin resistant cases and controls confirmed that the cis-genes and cis-NEMGs were enriched for differential expression in cases, providing independent validation that genetic maps can identify informative functional genes. Two additional results were consistent with a potential role of T2D-eQTL in regulating the 50 identified cis-NEMGs in the context of T2D risk: (1) the 50 cis-NEMGs showed greater differential expression compared to other NEMGs and (2) other NEMGs showed a trend towards significantly decreased expression if their expression levels correlated more highly with the subset of 50 cis-NEMGs. These 50 cis-NEMGs, which are differentially expressed and associated with mapped T2D disease loci, encode proteins acting within key mitochondrial pathways, including some of current therapeutic interest such as the metabolism of branched-chain amino acids, GABA and biotin.

World-wide distributions of lactase persistence alleles and the complex effects of recombination and selection.

The genetic trait of lactase persistence (LP) is associated with at least five independent functional single nucleotide variants in a regulatory region about 14 kb upstream of the lactase gene [-13910*T (rs4988235), -13907*G (rs41525747), -13915*G (rs41380347), -14009*G (rs869051967) and -14010*C (rs145946881)]. These alleles have been inferred to have spread recently and present-day frequencies have been attributed to positive selection for the ability of adult humans to digest lactose without risk of symptoms of lactose intolerance. One of the inferential approaches used to estimate the level of past selection has been to determine the extent of haplotype homozygosity (EHH) of the sequence surrounding the SNP of interest. We report here new data on the frequencies of the known LP alleles in the 'Old World' and their haplotype lineages. We examine and confirm EHH of each of the LP alleles in relation to their distinct lineages, but also show marked EHH for one of the older haplotypes that does not carry any of the five LP alleles. The region of EHH of this (B) haplotype exactly coincides with a region of suppressed recombination that is detectable in families as well as in population data, and the results show how such suppression may have exaggerated haplotype-based measures of past selection.

High-Resolution Genetic Maps Identify Multiple Type 2 Diabetes Loci at Regulatory Hotspots in African Americans and Europeans.

Interpretation of results from genome-wide association studies for T2D is challenging. Only very few loci have been replicated in African ancestry populations and the identification of the implicated functional genes remain largely undefined. We used genetic maps that capture detailed linkage disequilibrium information in European and African Americans and applied these to large T2D case-control samples in order to estimate locations for putative functional variants in both populations. Replicated T2D locations were tested for evidence of being regulatory hotspots using adipose expression. We validated a sample of our co-location intervals using next generation sequencing and functional annotation, including enhancers, transcription, and chromatin modifications. We identified 111 additional disease-susceptibility locations, 93 of which are cosmopolitan and 18 of which are European specific. We show that many previously known signals are also risk loci in African Americans. The majority of the disease locations appear to confer risk of T2D via the regulation of expression levels for a large number (266) of cis-regulated genes, the majority of which are not the nearest genes to the disease loci. Sequencing three cosmopolitan locations provided candidate functional variants that precisely co-locate with cell-specific chromatin domains and pancreatic islet enhancers. These variants have large effect sizes and are common across populations. Results show that disease-associated loci in different populations, gene expression, and cell-specific regulatory annotation can be effectively integrated by localizing these effects on high-resolution genetic maps. The cis-regulated genes provide insights into the complex molecular pathways involved and can be used as targets for sequencing and functional molecular studies.

The genomic and phenotypic diversity of Schizosaccharomyces pombe.

Natural variation within species reveals aspects of genome evolution and function. The fission yeast Schizosaccharomyces pombe is an important model for eukaryotic biology, but researchers typically use one standard laboratory strain. To extend the usefulness of this model, we surveyed the genomic and phenotypic variation in 161 natural isolates. We sequenced the genomes of all strains, finding moderate genetic diversity (π = 3 × 10(-3) substitutions/site) and weak global population structure. We estimate that dispersal of S. pombe began during human antiquity (∼340 BCE), and ancestors of these strains reached the Americas at ∼1623 CE. We quantified 74 traits, finding substantial heritable phenotypic diversity. We conducted 223 genome-wide association studies, with 89 traits showing at least one association. The most significant variant for each trait explained 22% of the phenotypic variance on average, with indels having larger effects than SNPs. This analysis represents a rich resource to examine genotype-phenotype relationships in a tractable model.

ABCC5 transporter is a novel type 2 diabetes susceptibility gene in European and African American populations.

Numerous functional studies have implicated PARL in relation to type 2 diabetes (T2D). We hypothesised that conflicting human association studies may be due to neighbouring causal variants being in linkage disequilibrium (LD) with PARL. We conducted a comprehensive candidate gene study of the extended LD genomic region that includes PARL and transporter ABCC5 using three data sets (two European and one African American), in relation to healthy glycaemic variation, visceral fat accumulation and T2D disease. We observed no evidence for previously reported T2D association with Val262Leu or PARL using array and fine-map genomic and expression data. By contrast, we observed strong evidence of T2D association with ABCC5 (intron 26) for European and African American samples (P = 3E-07) and with ABCC5 adipose expression in Europeans [odds ratio (OR) = 3.8, P = 2E-04]. The genomic location estimate for the ABCC5 functional variant, associated with all phenotypes and expression data (P = 1E-11), was identical for all samples (at Chr3q 185,136 kb B36), indicating that the risk variant is an expression quantitative trait locus (eQTL) with increased expression conferring risk of disease. That the association with T2D is observed in populations of disparate ancestry suggests the variant is a ubiquitous risk factor for T2D.

Refinement in localization and identification of gene regions associated with Crohn disease.

The risk of Crohn disease (CD) has a large genetic component. A recent meta-analysis of 6 genome-wide association studies reported 71 chromosomal intervals but does not account for all of the known genetic contribution. Here, we refine localization of the previously reported intervals and also identify additional CD susceptibility genes using a mapping approach that localizes causal variants based on genetic maps in linkage disequilibrium units (LDU maps). Using 2 of the 6 cohorts, 66 of the 71 previously reported loci are confirmed and more precise location estimates for these intervals are given. We identify 78 additional gene regions that pass genome-wide significance, providing strong evidence for 144 genes. Additionally, 56 nominally significant signals, but with more stringent and precise colocalization, are identified. In total, we provide evidence for 200 gene regions confirming that CD is truly multifactorial and complex in nature. Many identified genes have functions that are compatible with involvement in immune/inflammatory processes and seem to have a large effect in individuals with extra ileal as well as ileal inflammation. The precise locations and the evidence that some genes reflect phenotypic subgroups will help identify functional variants and will lead to greater insight of CD etiology.

A candidate-gene association study for berry colour and anthocyanin content in Vitis vinifera L.

Anthocyanin content is a trait of major interest in Vitis vinifera L. These compounds affect grape and wine quality, and have beneficial effects on human health. A candidate-gene approach was used to identify genetic variants associated with anthocyanin content in grape berries. A total of 445 polymorphisms were identified in 5 genes encoding transcription factors and 10 genes involved in either the biosynthetic pathway or transport of anthocyanins. A total of 124 SNPs were selected to examine association with a wide range of phenotypes based on RP-HPLC analysis and visual characterization. The phenotypes were total skin anthocyanin (TSA) concentration but also specific types of anthocyanins and relative abundance. The visual assessment was based on OIV (Organisation Internationale de la Vigne et du Vin) descriptors for berry and skin colour. The genes encoding the transcription factors MYB11, MYBCC and MYC(B) were significantly associated with TSA concentration. UFGT and MRP were associated with several different types of anthocyanins. Skin and pulp colour were associated with nine genes (MYB11, MYBCC, MYC(B), UFGT, MRP, DFR, LDOX, CHI and GST). Pulp colour was associated with a similar group of 11 genes (MYB11, MYBCC, MYC(B), MYC(A), UFGT, MRP, GST, DFR, LDOX, CHI and CHS(A)). Statistical interactions were observed between SNPs within the transcription factors MYB11, MYBCC and MYC(B). SNPs within LDOX interacted with MYB11 and MYC(B), while SNPs within CHI interacted with MYB11 only. Together, these findings suggest the involvement of these genes in anthocyanin content and on the regulation of anthocyanin biosynthesis. This work forms a benchmark for replication and functional studies.

Dissecting the genetics of complex inheritance: linkage disequilibrium mapping provides insight into Crohn disease.

Family studies for Crohn disease (CD) report extensive linkage on chromosome 16q and pinpoint NOD2 as a possible causative locus. However, linkage is also observed in families that do not bear the most frequent NOD2 causative mutations, but no other signals on 16q have been found so far in published genome-wide association studies. Our aim is to identify this missing genetic contribution. We apply a powerful genetic mapping approach to the Wellcome Trust Case-Control Consortium and the National Institute of Diabetes and Digestive and Kidney Diseases genome-wide association data on CD. This method takes into account the underlying structure of linkage disequilibrium (LD) by using genetic distances from LD maps and provides a location for the causal agent. We find genetic heterogeneity within the NOD2 locus and also show an independent and unsuspected involvement of the neighboring gene, CYLD. We find associations with the IRF8 region and the region containing CDH1 and CDH3, as well as substantial phenotypic and genetic heterogeneity for CD itself. The genes are known to be involved in inflammation and immune dysregulation. These findings provide insight into the genetics of CD and suggest promising directions for understanding disease heterogeneity. The application of this method thus paves the way for understanding complex inheritance in general, leading to the dissection of different pathways and ultimately, personalized treatment.

Identification and replication of three novel myopia common susceptibility gene loci on chromosome 3q26 using linkage and linkage disequilibrium mapping.

Refractive error is a highly heritable quantitative trait responsible for considerable morbidity. Following an initial genome-wide linkage study using microsatellite markers, we confirmed evidence for linkage to chromosome 3q26 and then conducted fine-scale association mapping using high-resolution linkage disequilibrium unit (LDU) maps. We used a preliminary discovery marker set across the 30-Mb region with an average SNP density of 1 SNP/15 kb (Map 1). Map 1 was divided into 51 LDU windows and additional SNPs were genotyped for six regions (Map 2) that showed preliminary evidence of multi-marker association using composite likelihood. A total of 575 cases and controls selected from the tails of the trait distribution were genotyped for the discovery sample. Malecot model estimates indicate three loci with putative common functional variants centred on MFN1 (180,566 kb; 95% confidence interval 180,505-180, 655 kb), approximately 156 kb upstream from alternate-splicing SOX2OT (182,595 kb; 95% CI 182,533-182,688 kb) and PSARL (184,386 kb; 95% CI 184,356-184,411 kb), with the loci showing modest to strong evidence of association for the Map 2 discovery samples (p<10(-7), p<10(-10), and p = 0.01, respectively). Using an unselected independent sample of 1,430 individuals, results replicated for the MFN1 (p = 0.006), SOX2OT (p = 0.0002), and PSARL (p = 0.0005) gene regions. MFN1 and PSARL both interact with OPA1 to regulate mitochondrial fusion and the inhibition of mitochondrial-led apoptosis, respectively. That two mitochondrial regulatory processes in the retina are implicated in the aetiology of myopia is surprising and is likely to provide novel insight into the molecular genetic basis of common myopia.

CHROMSCAN: genome-wide association using a linkage disequilibrium map.

CHROMSCAN implements a composite likelihood model for the analysis of association data. Disease-gene localisation is on a linkage disequilibrium unit (LDU) map, and locations and standard errors, for putatively causal polymorphisms, are determined by the programme. Distortions of the probability distribution created by auto-correlation are avoided by implementation of a permutation test. We evaluated the relative efficiency of the LDU map by simulating pseudo-phenotypes in real genotype samples. We observed that multi-locus mapping on an underlying LDU map reduces location error by approximately 46%. Furthermore, there is a small, but significant, increase in power of approximately 5%. Effective meta-analysis across multiple samples, increasingly important to combine evidence from genome-wide and other association data, is achieved through the weighted combination of location evidence provided by the programme.

LDMAP: the construction of high-resolution linkage disequilibrium maps of the human genome.

The precise characterization of the linkage disequilibrium (LD) landscape from high-density single-nucleotide polymorphism (SNP) data underpins the association mapping of diseases and other studies. We describe the algorithm and implementation of a powerful approach for constructing LD genetic maps with meaningful map distances. The computational problems posed by the enormous number of SNPs typed in the HapMap data are addressed by developing segmental map construction with the potential for parallelization, which we are developing. There is remarkably little loss of information (1-2%) through this approach, but the computation times are dramatically reduced (more than fourfold for sequential map assembly). These developments enable the construction of very high-density genome-wide LD maps using data from more than 3 million SNPs in HapMap. We anticipate that a whole-genome LD map will be useful for disease gene mapping, genomic research, and population genetics.

Association mapping of susceptibility loci for rheumatoid arthritis.

We analyzed a case-control data set for chromosome 18q from the Genetic Analysis Workshop 15 to detect susceptibility loci for rheumatoid arthritis (RA). A total number of 460 cases and 460 unaffected controls were genotyped on 2300 single-nucleotide polymorphisms (SNPs) by the North American Rheumatoid Arthritis Consortium. Using a multimarker approach for association mapping under the framework of the Malecot model and composite likelihood, we identified a region showing significant association with RA (p < 0.002) and the predicted disease locus was at a genomic location of 53,306 kb with a 95% confidence interval (CI) of 53,295-53,331 kb. A common haplotype in this region was protective against RA (p = 0.002). In another region showing nominal significant association (51,585 kb, 95% CI: 51,541-51,628 kb, p = 0.037), a haplotype was also protective (p = 0.002). We further demonstrated that reducing SNP density decreased power and accuracy of association mapping. SNP selection based on equal linkage disequilibrium (LD) distance generally produced higher accuracy than that based on equal kilobase distance or tagging.

Impact of marker density on the accuracy of association mapping.

We studied the impact of marker density on the accuracy of association mapping using Genetic Analysis Workshop 15 simulated dense single-nucleotide polymorphism (SNP) data on chromosome 6. A total of 1500 cases and 2000 unaffected controls genotyped for 17,820 SNPs were analyzed. We applied the approach that combines information from multiple SNPs under the framework of the Malecot model and composite likelihood to non-overlapping regions of the chromosome. We successfully detected the associations with disease Loci C and D and predicted their locations as small as zero distance to Locus C when it was "typed" and 112 kb from the untyped rare Locus D. Reducing marker density decreased the accuracy of location estimates. However, the predicted locations were robust to variations in the number of SNPs. Generally, the linkage disequilibrium (LD) map reflecting distances between markers in relation to LD produced higher accuracy than the physical map. We also demonstrated that SNP selection based on equal LD distance outperforms that based on equal physical distance or SNP tagging. Furthermore, ignoring rare SNPs diminished the ability to detect rare causal variants.

Exploiting large scale computing to construct high resolution linkage disequilibrium maps of the human genome.

UNLABELLED: Linkage disequilibrium (LD) maps increase power and precision in association mapping, define optimal marker spacing and identify recombination hot-spots and regions influenced by natural selection. Phase II of HapMap provides approximately 2.8-fold more single nucleotide polymorphisms (SNPs) than phase I for constructing higher resolution maps. LDMAP-cluster, is a parallel program for rapid map construction in a Linux environment used here to construct genome-wide LD maps with >8.2 million SNPs from the phase II data. AVAILABILITY: The LD maps, LDMAP-cluster and documentation are available from: http://www.som.soton.ac.uk/research/geneticsdiv/epidemiology/LDMAP. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Mapping genes for common diseases: the case for genetic (LD) maps.

We examine the current effort to develop a haplotype map of the human genome and suggest an alternative approach which represents linkage disequilibrium patterns in the form of a metric LD map. LD maps have some of the useful properties of genetic linkage maps but have a much higher resolution which is optimal for SNP-based association mapping of common diseases. The studies that have been undertaken to date suggest that LD and recombination maps show some close similarities because of abundant, narrow, recombination hot spots. These hot spots are co-localised in all populations but, unlike linkage maps, LD maps differ in scale for different populations because of differences in population history. The prospects for developing optimized panels of SNPs and the use of linkage disequilibrium maps in disease gene localisation are assessed in the light of recent evidence.