Imputation Reference Panel Research Articles

Performance and accuracy evaluation of reference panels for genotype imputation in sub-Saharan African populations

Based on evaluations of imputation performed on a genotype dataset consisting of about 11,000 sub-Saharan African (SSA) participants, we show Trans-Omics for Precision Medicine (TOPMed) and the African Genome Resource (AGR) to be currently the best panels for imputing SSA datasets. We report notable differences in the number of single-nucleotide polymorphisms (SNPs) that are imputed by different panels in datasets from East, West, and South Africa. Comparisons with a subset of 95 SSA high-coverage whole-genome sequences (WGSs) show that despite being about 20-fold smaller, the AGR imputed dataset has higher concordance with the WGSs. Moreover, the level of concordance between imputed and WGS datasets was strongly influenced by the extent of Khoe-San ancestry in a genome, highlighting the need for integration of not only geographically but also ancestrally diverse WGS data in reference panels for further improvement in imputation of SSA datasets. Approaches that integrate imputed data from different panels could also lead to better imputation.

Antithrombin, Protein C, and Protein S: Genome and Transcriptome-Wide Association Studies Identify 7 Novel Loci Regulating Plasma Levels.

Antithrombin, PC (protein C), and PS (protein S) are circulating natural anticoagulant proteins that regulate hemostasis and of which partial deficiencies are causes of venous thromboembolism. Previous genetic association studies involving antithrombin, PC, and PS were limited by modest sample sizes or by being restricted to candidate genes. In the setting of the Cohorts for Heart and Aging Research in Genomic Epidemiology consortium, we meta-analyzed across ancestries the results from 10 genome-wide association studies of plasma levels of antithrombin, PC, PS free, and PS total. Study participants were of European and African ancestries, and genotype data were imputed to TOPMed, a dense multiancestry reference panel. Each of the 10 studies conducted a genome-wide association studies for each phenotype and summary results were meta-analyzed, stratified by ancestry. Analysis of antithrombin included 25 243 European ancestry and 2688 African ancestry participants, PC analysis included 16 597 European ancestry and 2688 African ancestry participants, PSF and PST analysis included 4113 and 6409 European ancestry participants. We also conducted transcriptome-wide association analyses and multiphenotype analysis to discover additional associations. Novel genome-wide association studies and transcriptome-wide association analyses findings were validated by in vitro functional experiments. Mendelian randomization was performed to assess the causal relationship between these proteins and cardiovascular outcomes. Genome-wide association studies meta-analyses identified 4 newly associated loci: 3 with antithrombin levels (GCKR, BAZ1B, and HP-TXNL4B) and 1 with PS levels (ORM1-ORM2). transcriptome-wide association analyses identified 3 newly associated genes: 1 with antithrombin level (FCGRT), 1 with PC (GOLM2), and 1 with PS (MYL7). In addition, we replicated 7 independent loci reported in previous studies. Functional experiments provided evidence for the involvement of GCKR, SNX17, and HP genes in antithrombin regulation. The use of larger sample sizes, diverse populations, and a denser imputation reference panel allowed the detection of 7 novel genomic loci associated with plasma antithrombin, PC, and PS levels.

Abstract 6587: Method to increase performance of genome-wide association study using imputation for Asian ancestry

Abstract The emergence of the large-scale public genomic database facilitates genome-wide association study. An unprecedented number of samples emerged, which enable elucidate valid genetic inferences and contribute to finding novel associations for various phenotypes including cancer. However, most of the genome data is from European-based ancestry, which leads to the disparity in study for relatively minor ethnic groups including Asian. Meanwhile, among public genomic database, TOPMed consortium consists of ~180k participants, over 60% of the 180k samples are of predominantly non-European ancestry, and Asian ancestry consisted of 8 % in total. In addition, 1000 Genomes Project(1KG) contains 2,504 individuals from 26 populations and has sequenced 698 samples in addition. These projects serve as a reference to impute missing variants to improve association power for Genome-Wide Association Study (GWAS) subsequently. We compared the imputation results using those reference panels. Korean Genome Project (Korea1K), including 1094 whole genomes are public datasets consisting of only Korean (East Asian) origin. On GRCh38, Korea1K variants consisting of 59,463,566 variants in total, 99.6% with MAF <1% rare variant, only 35.7% of those rare variants passed from GATK Best Practice VQSR filter. With Korea1K data, we have downsampled the result into the exonic level, imputed with reference panels, and seen the consequences for which reference panel shows the best performances. Using 1KG as a reference, we could identify novel rare variants only in 0.1 to 0.38 fold of filtered rare variant in 1KG phase3 and 1KG phase3 30X respectively. On the other hand, TOPMed imputation resulted in rare variants over 1.65 fold increase. With well-imputed variants R2 over 0.8, all three reference database showed poor conservation which only 0.009 % of rare variant are rescued. The reference panel we commonly use are large but required to have more diversity in a Asian rare variant perspective. Despite increase of the Asian fraction in the public genomic database, our result suggests that the number of reference Asian samples are still low to perform association study. In conclusion, our finding highlights both the utility and limitation of the imputation reference panel for Asian ancestry. Novel variants filled by imputation are promising for adding power to various association studies. Citation Format: GangPyo Ryu, Youngil Koh, Sung-soo Yoon. Method to increase performance of genome-wide association study using imputation for Asian ancestry [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6587.

5613165 INVESTIGATING VARIANTS ASSOCIATED WITH QUANTITATIVE RED BLOOD CELL PHENOTYPES FROM TOPMED PREDICTIONS IN AFRICAN SICKLE CELL DISEASE POPULATIONS.

Background: Sickle cell disease (SCD) is caused by a single point mutation (c.20A>T, p.Glu6Val), which results in the formation of mutant haemoglobin S (HbS) instead of haemoglobin A (HbA). It is a life-long inherited recessive disorder that predominantly affects people of African, Indian, and Mediterranean descent. The polymerisation of HbS within red blood cells (RBCs) is the primary pathophysiological alteration in SCD, which results in the sickle-shape of RBCs. Sickle-shaped RBCs are more prone to haemolysis and blocking blood flow, resulting in anaemia and painful episodes. Haematological indices including haematocrit (HCT), haemoglobin concentration (Hb), mean corpuscular volume (MCV), mean corpuscular haemoglobin (MCH), mean corpuscular haemoglobin concentration (MCHC), red blood cell width (RDW) and RBC count indicate the size, development, and haemoglobin content of RBCs. Haematological indices are routinely measured as a means of monitoring SCD. A study recently found associations between 14 novel variants and haematological indices when using whole genome sequencing (WGS) data. The association between these variants and haematological indices have not yet been investigated in African SCD patients. This study aims to determine the associations between these variants and haematological indices in SCD patients from Cameroon and Senegal compared to non-SCD individuals. Aim: Investigating the association of 13 novel variants previously found in the Trans-Omics for Precision Medicine (TOPMed) project to haematological indices in African SCD patients. Methods: The 14 variants from the TOPMed study will be genotyped using a TaqMan Assay platform. Sanger sequencing will then be performed on 10% of the cohort to validate the TaqMan Assay result. The application R will be used for the descriptive statistics using sociodemographic, clinical, and haematological indices to evaluate the cases and controls. The Hardy-Weinberg Equilibrium will be used to ensure quality control. A linear regression analysis will be performed in R to determine the association between the variants and RBC phenotypes. Expected results: We are expecting to find an association between the SNPs and haematologocial indices. Conclusion: The variants of focus in this study were prioritized using WGS data, therefore capturing variants that are usually not included in GWAS arrays or imputation reference panels. The associations of the 13 novel variants with haematological indices have not yet been tested in African SCD populations. The results of this investigation could positively impact future studies into predictive modelling using haematological indices.

Non-Parametric MCMC Gibbs Sampler Approach and Misclassification Assessment of Estimating Haplotype Frequencies among Related Statistical Approaches

Introduction: Haplotype analysis allows higher resolution analysis in genetic association studies and is used as a reference panel for genotype imputation in genome-wide association studies. Haplotypes estimates from genotypes among unrelated individuals but misclassification of the haplotype reconstruction will directly affect the accuracy of the results.
 Methods: This study proposes a novel statistical method Gibbs sampler algorithm to estimate haplotype frequency and quantify the influence of misclassification bias of the estimate haplotype. The performance of the algorithm is evaluated on simulated datasets assuming that linkage phase unknown. The simulation used different minor allele frequencies at each single nucleotide polymorphism (SNP) and different linkagedisequilibrium between the SNPs.
 Results: The Gibbs sampler algorithm presents higher accuracy among over seven SNPs or less, validated, and deals with missing genotype compared to previous related statistical approaches. Misclassification of estimated haplotypes leads to non-differential bias in exposure and affects haplotype estimates in haplotype analysis. The observed odds ratio underestimates the association between haplotype and phenotype by 36% to 99%.
 Conclusion: The Gibbs sampler algorithm provides higher accuracy and robust effectiveness performance, handles missing genotypes and provides uncertain probabilities of haplotype frequencies. The misclassification bias of the estimate haplotype underestimates the genetic association by more than forty percent.

Genome-Wide Association Study of Age-Related Macular Degeneration Reveals 2 New Loci Implying Shared Genetic Components with Central Serous Chorioretinopathy

To investigate the genetic architecture of age-related macular degeneration (AMD) in a Japanese population. Genome-wide association study (GWAS). Three thousand seven hundred seventy-two patients with AMD and 16 770 control participants from the Japanese population were enrolled in the association analyses. We conducted a meta-analysis of 2 independent GWASs that included a total of 2663 patients with AMD and 9471 control participants using the imputation reference panel for genotype imputation specified for the Japanese population (n= 3541). A replication study was performed using an independent set of 1109 patients with AMD and 7299 control participants. Associations of genetic variants with AMD. A meta-analysis of the 2 GWASs identified 6 loci significantly associated with AMD (P < 5.0× 10-8). Of these loci, 4 were known to be associated with AMD (CFH, C2/FB, TNFRSF10A, and ARMS2), and 2 were novel (rs4147157 near WBP1L and rs76228488 near GATA5). The newly identified associations were confirmed in a replication study (P < 0.01). After the meta-analysis of all datasets, we observed strong associations in these loci (P= 1.88× 10-12 and P= 1.35× 10-9 for meta-analysis for rs4147157 and rs76228488, respectively). When we looked up the associations in the reported central serous chorioretinopathy (CSC) GWAS conducted in the Japanese population, both loci were associated significantly with CSC (P= 4.86× 10-3 and P= 4.28× 10-3 for rs4147157 and rs76228488, respectively). We performed a genetic colocalization analysis for these loci and estimated that the posterior probabilities of shared causal variants between AMD and CSC were 0.39 and 0.60 for WBP1L and GATA5, respectively. Genetic correlation analysis focusing on the epidemiologically suggested clinical risk factors implicated shared polygenic architecture between AMD and smoking cessation (rg [the measure of genetic correlation] = -0.33; P= 0.01; false discovery rate, 0.099). Our findings imply shared genetic components conferring the risk of both AMD and CSC. Proprietary or commercial disclosure may be found after the references.

PGG.MHC: toward understanding the diversity of major histocompatibility complexes in human populations.

The human leukocyte antigen (HLA) system, or the human version of the major histocompatibility complex (MHC), is known for its extreme polymorphic nature and high heterogeneity. Taking advantage of whole-genome and whole-exome sequencing data, we developed PGG.MHC to provide a platform to explore the diversity of the MHC in Asia as well as in global populations. PGG.MHC currently archives high-resolution HLA alleles of 53 254 samples representing 190 populations spanning 66 countries. PGG.MHC provides: (i) high-quality allele frequencies for eight classical HLA loci (HLA-A, -B, -C, -DQA1, -DQB1, -DRB1, -DPA1 and -DPB1); (ii) visualization of population prevalence of HLA alleles on global, regional, and country-wide levels; (iii) haplotype structure of 134 populations; (iv) two online analysis tools including 'HLA imputation' for inferring HLA alleles from SNP genotyping data and 'HLA association' to perform case/control studies for HLA-related phenotypes and (v) East Asian-specific reference panels for HLA imputation. Equipped with high-quality frequency data and user-friendly computer tools, we expect that the PGG.MHC database can advance the understanding and facilitate applications of MHC genomic diversity in both evolutionary and medical studies. The PGG.MHC database is freely accessible via https://pog.fudan.edu.cn/pggmhc or https://www.pggmhc.org/pggmhc.

Analyses of rare predisposing variants of lung cancer in 6,004 whole genomes in Chinese.

We present the largest whole-genome sequencing (WGS) study of non-small cell lung cancer (NSCLC) to date among 6,004 individuals of Chinese ancestry, coupled with 23,049 individuals genotyped by SNP array. We construct a high-quality haplotype reference panel for imputation and identify 20 common and low-frequency loci (minor allele frequency [MAF] ≥ 0.5%), including five loci that have never been reported before. For rare loss-of-function (LoF) variants (MAF<0.5%), we identify BRCA2 and 18 other cancer predisposition genes that affect 5.29% of individuals with NSCLC, and 98.91% (181 of 183) of LoF variants have not been linked previously to NSCLC risk. Promoter variants of BRCA2 also have a substantial effect on NSCLC risk, and their prevalence is comparable with BRCA2 LoF variants. The associations are validated in an independent case-control study including 4,410 individuals and a prospective cohort study including 23,826 individuals. Our findings not only provide a high-quality reference panel for future array-based association studies but depict the whole picture of rare pathogenic variants for NSCLC.

Evaluation of Whole-Genome Sequence Imputation Strategies in Korean Hanwoo Cattle.

Simple SummaryIn this study, we evaluated various imputation strategies for the Korean Hanwoo cattle. We observed that a large reference panel consisting of many cattle breeds did not improve the imputation accuracy when compared to a proportionally small purebred Hanwoo reference. This was because the multi-breed reference did not contain animals sufficiently related to the Hanwoo to improve the accuracies and, although not detrimental, in effect, only added to the computational burden of the imputation. Despite the large multi-breed reference, when the Hanwoo were removed from the reference, the imputation accuracies were low. These results suggest additional sequencing efforts are needed for underrepresented breeds, particularly those less genetically related to the main European breeds.This study evaluated the accuracy of sequence imputation in Hanwoo beef cattle using different reference panels: a large multi-breed reference with no Hanwoo (n = 6269), a much smaller Hanwoo purebred reference (n = 88), and both datasets combined (n = 6357). The target animals were 136 cattle both sequenced and genotyped with the Illumina BovineSNP50 v2 (50K). The average imputation accuracy measured by the Pearson correlation (R) was 0.695 with the multi-breed reference, 0.876 with the purebred Hanwoo, and 0.887 with the combined data; the average concordance rates (CR) were 88.16%, 94.49%, and 94.84%, respectively. The accuracy gains from adding a large multi-breed reference of 6269 samples to only 88 Hanwoo was marginal; however, the concordance rate for the heterozygotes decreased from 85% to 82%, and the concordance rate for fixed SNPs in Hanwoo also decreased from 99.98% to 98.73%. Although the multi-breed panel was large, it was not sufficiently representative of the breed for accurate imputation without the Hanwoo animals. Additionally, we evaluated the value of high-density 700K genotypes (n = 991) as an intermediary step in the imputation process. The imputation accuracy differences were negligible between a single-step imputation strategy from 50K directly to sequence and a two-step imputation approach (50K-700K-sequence). We also observed that imputed sequence data can be used as a reference panel for imputation (mean R = 0.9650, mean CR = 98.35%). Finally, we identified 31 poorly imputed genomic regions in the Hanwoo genome and demonstrated that imputation accuracies were particularly lower at the chromosomal ends.

Extent to which array genotyping and imputation with large reference panels approximate deep whole-genome sequencing

Understanding the genetic basis of human diseases and traits is dependent on the identification and accurate genotyping of genetic variants. Deep whole-genome sequencing (WGS), the gold standard technology for SNP and indel identification and genotyping, remains very expensive for most large studies. Here, we quantify the extent to which array genotyping followed by genotype imputation can approximate WGS in studies of individuals of African, Hispanic/Latino, and European ancestry in the US and of Finnish ancestry in Finland (a population isolate). For each study, we performed genotype imputation by using the genetic variants present on the Illumina Core, OmniExpress, MEGA, and Omni 2.5M arrays with the 1000G, HRC, and TOPMed imputation reference panels. Using the Omni 2.5M array and the TOPMed panel, ≥90% of bi-allelic single-nucleotide variants (SNVs) are well imputed (r2>0.8) down to minor-allele frequencies (MAFs) of 0.14% in African, 0.11% in Hispanic/Latino, 0.35% in European, and 0.85% in Finnish ancestries. There was little difference in TOPMed-based imputation quality among the arrays with >700k variants. Individual-level imputation quality varied widely between and within the three US studies. Imputation quality also varied across genomic regions, producing regions where even common (MAF>5%) variants were consistently not well imputed across ancestries. The extent to which array genotyping and imputation can approximate WGS therefore depends on reference panel, genotype array, sample ancestry, and genomic location. Imputation quality by variant or genomic region can be queried with our new tool, RsqBrowser, now deployed on the Michigan Imputation Server.

Development and validation of a horse reference panel for genotype imputation

BackgroundGenotype imputation is a cost-effective method to generate sequence-level genotypes for a large number of animals. Its application can improve the power of genomic studies, provided that the accuracy of imputation is sufficiently high. The purpose of this study was to develop an optimal strategy for genotype imputation from genotyping array data to sequence level in German warmblood horses, and to investigate the effect of different factors on the accuracy of imputation. Publicly available whole-genome sequence data from 317 horses of 46 breeds was used to conduct the analyses.ResultsDepending on the size and composition of the reference panel, the accuracy of imputation from medium marker density (60K) to sequence level using the software Beagle 5.1 ranged from 0.64 to 0.70 for horse chromosome 3. Generally, imputation accuracy increased as the size of the reference panel increased, but if genetically distant individuals were included in the panel, the accuracy dropped. Imputation was most precise when using a reference panel of multiple but related breeds and the software Beagle 5.1, which outperformed the other two tested computer programs, Impute 5 and Minimac 4. Genome-wide imputation for this scenario resulted in a mean accuracy of 0.66. Stepwise imputation from 60K to 670K markers and subsequently to sequence level did not improve the accuracy of imputation. However, imputation from higher density (670K) was considerably more accurate (about 0.90) than from medium density. Likewise, imputation in genomic regions with a low marker coverage resulted in a reduced accuracy of imputation.ConclusionsThe accuracy of imputation in horses was influenced by the size and composition of the reference panel, the marker density of the genotyping array, and the imputation software. Genotype imputation can be used to extend the limited amount of available sequence-level data from horses in order to boost the power of downstream analyses, such as genome-wide association studies, or the detection of embryonic lethal variants.

Genotype imputation and polygenic score estimation in northwestern Russian population.

Numerous studies demonstrated the lack of transferability of polygenic score (PGS) models across populations and the problem arising from unequal presentation of ancestries across genetic studies. However, even within European ancestry there are ethnic groups that are rarely presented in genetic studies. For instance, Russians, being one of the largest, diverse, and yet understudied group in Europe. In this study, we evaluated the reliability of genotype imputation for the Russian cohort by testing several commonly used imputation reference panels (e.g. HRC, 1000G, HGDP). HRC, in comparison with two other panels, showed the most accurate results based on both imputation accuracy and allele frequency concordance between masked and imputed genotypes. We built polygenic score models based on GWAS results from the UK biobank, measured the explained phenotypic variance in the Russian cohort attributed to polygenic scores for 11 phenotypes, collected in the clinic for each participant, and finally explored the role of allele frequency discordance between the UK biobank and the study cohort in the resulting PGS performance.

Decoding the diversity of killer immunoglobulin-like receptors by deep sequencing and a high-resolution imputation method

The killer cell immunoglobulin-like receptor (KIR) recognizes human leukocyte antigen (HLA) class I molecules and modulates the function of natural killer cells. Despite its role in immunity, the complex genomic structure has limited a deep understanding of the KIR genomic landscape. Here we conduct deep sequencing of 16 KIR genes in 1,173 individuals. We devise a bioinformatics pipeline incorporating copy number estimation and insertion or deletion (indel) calling for high-resolution KIR genotyping. We define 118 alleles in 13 genes and demonstrate a linkage disequilibrium structure within and across KIR centromeric and telomeric regions. We construct a KIR imputation reference panel (nreference= 689, imputation accuracy= 99.7%), apply it to biobank genotype (ntotal= 169,907), and perform phenome-wide association studies of 85 traits. We observe a dearth of genome-wide significant associations, even in immune traits implicated previously to be associated with KIR (the smallest p= 1.5× 10-4). Our pipeline presents a broadly applicable framework to evaluate innate immunity in large-scale datasets.

Using population-specific add-on polymorphisms to improve genotype imputation in underrepresented populations.

Genome-wide association studies rely on the statistical inference of untyped variants, called imputation, to increase the coverage of genotyping arrays. However, the results are often suboptimal in populations underrepresented in existing reference panels and array designs, since the selected single nucleotide polymorphisms (SNPs) may fail to capture population-specific haplotype structures, hence the full extent of common genetic variation. Here, we propose to sequence the full genomes of a small subset of an underrepresented study cohort to inform the selection of population-specific add-on tag SNPs and to generate an internal population-specific imputation reference panel, such that the remaining array-genotyped cohort could be more accurately imputed. Using a Tanzania-based cohort as a proof-of-concept, we demonstrate the validity of our approach by showing improvements in imputation accuracy after the addition of our designed add-on tags to the base H3Africa array.

Leveraging TOPMed imputation server and constructing a cohort-specific imputation reference panel to enhance genotype imputation among cystic fibrosis patients.

Cystic fibrosis (CF) is a severe genetic disorder that can cause multiple comorbidities affecting the lungs, the pancreas, the luminal digestive system and beyond. In our previous genome-wide association studies (GWAS), we genotyped approximately 8,000 CF samples using a mixture of different genotyping platforms. More recently, the Cystic Fibrosis Genome Project (CFGP) performed deep (approximately 30×) whole genome sequencing (WGS) of 5,095 samples to better understand the genetic mechanisms underlying clinical heterogeneity among patients with CF. For mixtures of GWAS array and WGS data, genotype imputation has proven effective in increasing effective sample size. Therefore, we first performed imputation for the approximately 8,000 CF samples with GWAS array genotype using the Trans-Omics for Precision Medicine (TOPMed) freeze 8 reference panel. Our results demonstrate that TOPMed can provide high-quality imputation for patients with CF, boosting genomic coverage from approximately 0.3–4.2 million genotyped markers to approximately 11–43 million well-imputed markers, and significantly improving polygenic risk score (PRS) prediction accuracy. Furthermore, we built a CF-specific CFGP reference panel based on WGS data of patients with CF. We demonstrate that despite having approximately 3% the sample size of TOPMed, our CFGP reference panel can still outperform TOPMed when imputing some CF disease-causing variants, likely owing to allele and haplotype differences between patients with CF and general populations. We anticipate our imputed data for 4,656 samples without WGS data will benefit our subsequent genetic association studies, and the CFGP reference panel built from CF WGS samples will benefit other investigators studying CF.

COMPARISON OF GENOTYPE PHASING METODS FOR THE HIGH THROUGHPUT SEQUENCING DATA OF CLINICAL EXOMES

The aim of this investigation was to search for an algorithm for phasing of exome NGS data that would be optimal for analyzing the cis-/transposition of closely located polymorphic loci in patients with hereditary diseases. The NGS data on clinical exome sequences from 149 patients was collected and analyzed, and its primary processing was carried out. A combined reference panel for phasing and genotype imputation “Belref1000G” was created by adding 131 samples of Belarusian patients to the panel from the “1000 Genomes” project. The most effective methods of phasing have been determined: the Michigan Imputation Server (online service) allows achieving the highest accuracy of phasing data from clinical exomes in Belarus. In cases where an online format of phasing is not appropriate for whatever reason, Beagle software with the combined reference panel “Belref1000G” is recommended for use. Beagle software with the “1000G” reference panel should be used to obtain the longest phasing blocks.

Suggestive linkage and association of preserved cognition to chromosome 18 in genetically at-risk Amish.

Previous studies have identified genetic risk factors for Alzheimer Disease (AD). In the Amish community, many individuals maintain normal cognitive function at older ages, despite having relatively high genetic risk of AD. These individuals may carry protective alleles that buffer the genetic risk. To find such loci, we utilized the Amish family structures to apply a modified genetic linkage analytical tool that considered genetic risk of AD. Cognitive status in adult Amish subjects (n=1,855) was assigned using the modified mini-mental status exam (3MS). Individuals aged ≥ 75 with 3MS ≥ 87 were classified as CN. Those with 3MS < 87 were classified impaired (CI), regardless of age. Genome-wide genotypes were used to impute ∼7 million variants using the HRC imputation reference panel. For genome-wide linkage analysis, we selected 2,167 SNPs and analyzed 110 sibships with ≥ 2 CN members. Adjusting for known genetic risk, we applied ordered subsets analysis (OSA) to rank families by mean Genetic Risk Score (GRS) of CN members. GRS was derived from 27 genome-wide AD loci (Kunkle et al, 2019). Regions with significant increases in linkage (LOD*) were assessed by mixed model genetic association testing. Ranking sibships by high-to-low mean GRS, significant increases in the LOD* after OSA were observed at ∼40 Mb on chromosome (Chr) 19 (peak LOD*=2.176, p=0.009) and ∼49 Mb on chr 18 (peak LOD*=2.23, p=0.03). While linkage to chr 19 may result from known AD loci, the observed linkage to chr 18 does not overlap known risk loci. Mixed model association testing within a 5 Mb region of the peak linkage to chr 18 was performed in 1,036 SNPs with minor allele frequency ≥ 0.01. Individuals with a GRS ≥ 3.65, corresponding to the GRS in the 17 families with the strongest LOD* increase, were considered (n=236). The strongest association occurred at rs12969463 (p=0.002, OR = 2). This study detected increased evidence of linkage and association for protective AD loci on chr 18, considering CN individuals with the highest risk from known AD loci. One or more protective alleles that buffer this increased risk may be located in this region.

A globally diverse reference alignment and panel for imputation of mitochondrial DNA variants

BackgroundVariation in mitochondrial DNA (mtDNA) identified by genotyping microarrays or by sequencing only the hypervariable regions of the genome may be insufficient to reliably assign mitochondrial genomes to phylogenetic lineages or haplogroups. This lack of resolution can limit functional and clinical interpretation of a substantial body of existing mtDNA data. To address this limitation, we developed and evaluated a large, curated reference alignment of complete mtDNA sequences as part of a pipeline for imputing missing mtDNA single nucleotide variants (mtSNVs). We call our reference alignment and pipeline MitoImpute.ResultsWe aligned the sequences of 36,960 complete human mitochondrial genomes downloaded from GenBank, filtered and controlled for quality. These sequences were reformatted for use in imputation software, IMPUTE2. We assessed the imputation accuracy of MitoImpute by measuring haplogroup and genotype concordance in data from the 1000 Genomes Project and the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The mean improvement of haplogroup assignment in the 1000 Genomes samples was 42.7% (Matthew’s correlation coefficient = 0.64). In the ADNI cohort, we imputed missing single nucleotide variants.ConclusionThese results show that our reference alignment and panel can be used to impute missing mtSNVs in existing data obtained from using microarrays, thereby broadening the scope of functional and clinical investigation of mtDNA. This improvement may be particularly useful in studies where participants have been recruited over time and mtDNA data obtained using different methods, enabling better integration of early data collected using less accurate methods with more recent sequence data.

RefRGim: an intelligent reference panel reconstruction method for genotype imputation with convolutional neural networks.

Genotype imputation is a statistical method for estimating missing genotypes from a denser haplotype reference panel. Existing methods usually performed well on common variants, but they may not be ideal for low-frequency and rare variants. Previous studies showed that the population similarity between study and reference panels is one of the key factors influencing the imputation accuracy. Here, we developed an imputation reference panel reconstruction method (RefRGim) using convolutional neural networks (CNNs), which can generate a study-specified reference panel for each input data based on the genetic similarity of individuals from current study and references. The CNNs were pretrained with single nucleotide polymorphism data from the 1000 Genomes Project. Our evaluations showed that genotype imputation with RefRGim can achieve higher accuracies than original reference panel, especially for low-frequency and rare variants. RefRGim will serve as an efficient reference panel reconstruction method for genotype imputation. RefRGim is freely available via GitHub: https://github.com/shishuo16/RefRGim

The impact of non-additive genetic associations on age-related complex diseases

Genome-wide association studies (GWAS) are not fully comprehensive, as current strategies typically test only the additive model, exclude the X chromosome, and use only one reference panel for genotype imputation. We implement an extensive GWAS strategy, GUIDANCE, which improves genotype imputation by using multiple reference panels and includes the analysis of the X chromosome and non-additive models to test for association. We apply this methodology to 62,281 subjects across 22 age-related diseases and identify 94 genome-wide associated loci, including 26 previously unreported. Moreover, we observe that 27.7% of the 94 loci are missed if we use standard imputation strategies with a single reference panel, such as HRC, and only test the additive model. Among the new findings, we identify three novel low-frequency recessive variants with odds ratios larger than 4, which need at least a three-fold larger sample size to be detected under the additive model. This study highlights the benefits of applying innovative strategies to better uncover the genetic architecture of complex diseases.