Local Haplotype Research Articles

Local read haplotagging enables accurate long-read small variant calling

Long-read sequencing technology has enabled variant detection in difficult-to-map regions of the genome and enabled rapid genetic diagnosis in clinical settings. Rapidly evolving third-generation sequencing platforms like Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) are introducing newer platforms and data types. It has been demonstrated that variant calling methods based on deep neural networks can use local haplotyping information with long-reads to improve the genotyping accuracy. However, using local haplotype information creates an overhead as variant calling needs to be performed multiple times which ultimately makes it difficult to extend to new data types and platforms as they get introduced. In this work, we have developed a local haplotype approximate method that enables state-of-the-art variant calling performance with multiple sequencing platforms including PacBio Revio system, ONT R10.4 simplex and duplex data. This addition of local haplotype approximation simplifies long-read variant calling with DeepVariant.

The interplay of local adaptation and gene flow may lead to the formation of supergenes.

Supergenes are genetic architectures resulting in the segregation of alternative combinations of alleles underlying complex phenotypes. The co-segregation of alleles at linked loci is often facilitated by polymorphic chromosomal rearrangements suppressing recombination locally. Supergenes are involved in many complex polymorphisms, including sexual, colour or behavioural polymorphisms in numerous plants, fungi, mammals, fish, and insects. Despite a long history of empirical and theoretical research, the formation of supergenes remains poorly understood. Here, using a two-island population genetic model, we explore how gene flow and the evolution of overdominant chromosomal inversions may jointly lead to the formation of supergenes. We show that the evolution of inversions in differentiated populations, both under disruptive selection, leads to an increase in frequency of poorly adapted, immigrant haplotypes. Indeed, rare allelic combinations, such as immigrant haplotypes, are more frequently reshuffled by recombination than common allelic combinations, and therefore benefit from the recombination suppression generated by inversions. When an inversion capturing a locally adapted haplotype spreads but is associated with a fitness cost hampering its fixation (e.g. a recessive mutation load), the maintenance of a non-inverted haplotype in the population is enhanced; under certain conditions, the immigrant haplotype persists alongside the inverted local haplotype, while the standard local haplotype disappears. This establishes a stable, local polymorphism with two non-recombining haplotypes encoding alternative adaptive strategies, that is, a supergene. These results bring new light to the importance of local adaptation, overdominance, and gene flow in the formation of supergenes and inversion polymorphisms in general.

Evaluation of the genetic basis of familial-associated early-onset hematologic cancers in an ancestral/ethnically diverse population.

Genetic predisposition to hematologic malignancies has historically been addressed utilizing patients recruited from clinical trials and pedigrees constructed at major treatment centers. Such efforts leave unexplored the genetic basis of variations in risk by race/ethnic group shown in population-based surveillance data where cancer registration, compulsory by law, delivers universal enrollment. To address this, we performed exome sequencing on DNA isolated from newborn bloodspots derived from sibling pairs with early-onset cancers across California in which at least one of the siblings developed a hematologic cancer, using unbiased recruitment from the full state population. We identified pathogenic/likely pathogenic (P/ LP) variants among 1,172 selected cancer genes that were private or present at low allele frequencies in reference populations. Within 64 subjects from 32 families, we found 9 LP variants shared between siblings, and an additional 7 such variants in singleton children (not shared with their sibling). In 8 of the shared cases, the ancestral origin of the local haplotype that carries P/LP variants matched the dominant global ancestry of study participant families. This was the case for Latino sibling pairs on FLG and CBLB, non-Latino White sibling pairs in TP53 and NOD2, and a shared GATA2 variant for a non-Latino Black sibling pair. A new inherited mutation in HABP2 was identified in a sibling pair, one with diffuse large B-cell lymphoma and the other with neuroblastoma. Overall, the profile of P/LP germline variants across ancestral/ethnic groups suggests that rare alleles contributing to hematologic diseases originate within their race/ethnic origin parental populations, demonstrating the value of this discovery process in diverse, population-based registries.

Comparative analysis of chloroplast genomes of two Chinese local citrus varieties and haplotype analysis with other citrus species

Citrus plants, comprising some of the most widely cultivated fruit crops worldwide, have great research value. We assembled and annotated the chloroplast genomes of two local important citrus varieties of China, ‘Tuju’ and ‘Zangju’, and comparatively analyzed their sequences. The full lengths chloroplast genomes of ‘Tuju’ and ‘Zangju’ were 160,121 and 160,694 bp, respectively. Both featured a quadripartite structure with two copies of a large inverted repeat (IR) separated by small (SSC) and large (LSC) single-copy regions. ‘Tuju’ chloroplast genome contained 134 genes. ‘Zangju’ had one less protein-coding gene (rpl22) than ‘Tuju’. We also found 27 and 21 optimal codons, all of which ended with A or U, for chloroplast protein-coding genes of ‘Tuju’ and ‘Zangju’, respectively. There were 32 haplotypes obtained from 43 citrus chloroplast genomes, some different species or cultivars could share one haplotype. Of them, ‘Tuju’, one individual of Citrus limon, one individual of Citrus maxima and Citrus platymamma shared one haplotype, which implied they had the same maternal origin. The phylogenetic trees supported that all haplotypes could be divided five groups, while haplotype network analysis tended to get three groups. Our study reveals the complex lineage relationship of citrus plants at a certain level, and provides a valuable reference tool for future research on citrus plants.

Using association mapping and local interval haplotype association analysis to improve the cotton drought stress response

Drought stress has a serious impact on the growth and development of cotton. To explore the relevant molecular mechanism of the drought stress response in cotton, gene mapping based on the QTL interval mapped by simplified genome BSA-seq of the drought-resistance-related RIL population was performed. A QTL region spanning 2.02 Mb on chromosome D07 was selected, and 201 resource materials were genotyped using 9 KASP markers in the interval. After local interval haplotype association analysis, the overlap of the 110 kb peak region confirmed the reliability of this region, and at the same time, the role of GhGF14–30, the only gene in the overlapping region, was modeled in the response of cotton to drought stress. qRTPCR analysis of the materials and population parents proved that this gene plays a role in the drought stress response in cotton. Virus-induced gene silencing proved the importance of this gene in drought-sensitive materials, and drought-resistance-related marker genes also proved that the GhGF14–30 gene may play an important role in the ABA and SOS signaling pathways. This study provides a basis for mining drought stress response functional genes in cotton and lays the foundation for the molecular mechanism of the GhGF14–30 gene in response to drought stress in cotton.

Minimal positional substring cover is a haplotype threading alternative to Li and Stephens model.

The Li and Stephens (LS) hidden Markov model (HMM) models the process of reconstructing a haplotype as a mosaic copy of haplotypes in a reference panel. For small panels, the probabilistic parameterization of LS enables modeling the uncertainties of such mosaics. However, LS becomes inefficient when sample size is large, because of its linear time complexity. Recently the PBWT, an efficient data structure capturing the local haplotype matching among haplotypes, was proposed to offer a fast method for giving some optimal solution (Viterbi) to the LS HMM. Previously, we introduced the minimal positional substring cover (MPSC) problem as an alternative formulation of LS whose objective is to cover a query haplotype by a minimum number of segments from haplotypes in a reference panel. The MPSC formulation allows the generation of a haplotype threading in time constant to sample size (O(N)). This allows haplotype threading on very large biobank-scale panels on which the LS model is infeasible. Here, we present new results on the solution space of the MPSC. In addition, we derived a number of optimal algorithms for MPSC, including solution enumerations, the length maximal MPSC, and h-MPSC solutions. In doing so, our algorithms reveal the solution space of LS for large panels. We show that our method is informative in terms of revealing the characteristics of biobank-scale data sets and can improve genotype imputation.

Haptools: a toolkit for admixture and haplotype analysis.

Leveraging local ancestry and haplotype information in genome-wide association studies and downstream analyses can improve the utility of genomics for individuals from diverse and recently admixed ancestries. However, most existing simulation, visualization and variant analysis frameworks are based on variant-level analysis and do not automatically handle these features. We present haptools, an open-source toolkit for performing local ancestry aware and haplotype-based analysis of complex traits. Haptools supports fast simulation of admixed genomes, visualization of admixture tracks, simulation of haplotype- and local ancestry-specific phenotype effects and a variety of file operations and statistics computed in a haplotype-aware manner. Haptools is freely available at https://github.com/cast-genomics/haptools. Detailed documentation is available at https://haptools.readthedocs.io. Supplementary data are available at Bioinformatics online.

Inverted genomic regions between reference genome builds in humans impact imputation accuracy and decrease the power of association testing.

Over the last two decades, the human reference genome has undergone multiple updates as we complete a linear representation of our genome. Two versions of human references are currently used in the biomedical literature, GRCh37/hg19 and GRCh38. Conversions between these versions are critical for quality control, imputation, and association analysis. In the present study, we show that single-nucleotide variants (SNVs) in regions inverted between different builds of the reference genome are often mishandled bioinformatically. Depending on the array type, SNVs are found in approximately 2-5 Mb of the genome that are inverted between reference builds. Coordinate conversions of these variants are mishandled by both the TOPMed imputation server as well as routine in-house quality control pipelines, leading to underrecognized downstream analytical consequences. Specifically, we observe that undetected allelic conversion errors for palindromic (i.e., A/T or C/G) variants in these inverted regions would destabilize the local haplotype structure, leading to loss of imputation accuracy and power in association analyses. Though only a small proportion of the genome is affected, these regions include important disease susceptibility variants that would be affected. For example, the p value of a known locus associated with prostate cancer on chromosome 10 (chr10) would drop from 2.86×10-7 to 0.0011 in a case-control analysis of 20,286 Africans and African Americans (10,643 cases and 9,643 controls). We devise a straight-forward heuristic based on the popular tool, liftOver, that can easily detect and correct these variants in the inverted regions between genome builds to locally improve imputation accuracy.

A Role of DNA Methylation within the CYP17A1 Gene in the Association of Genetic and Environmental Risk Factors with Stress-Related Manifestations of Schizophrenia.

As genetic and environmental influences on schizophrenia might converge on DNA methylation (DNAm) within loci which are both associated with the disease and implicated in response to environmental stress, we examined whether DNAm within CYP17A1, a hypothalamus–pituitary–adrenal axis gene which is situated within the schizophrenia risk locus 10q24.32, would mediate genetic and environmental effects on stress-related schizophrenia symptoms. DNAm within an exonic–intronic fragment of CYP17A1 was assessed in the blood of 66 schizophrenia patients and 63 controls using single-molecule real-time bisulfite sequencing. Additionally, the VNTR polymorphism of the AS3MT gene, a plausible causal variant within the 10q24.32 locus, was genotyped in extended patient and control samples (n = 700). The effects of local haplotype, VNTR and a polyenviromic risk score (PERS) on DNAm, episodic verbal memory, executive functions, depression, and suicidality of patients were assessed. Haplotype and PERS differentially influenced DNAm at four variably methylated sites identified within the fragment, with stochastic, additive, and allele-specific effects being found. An allele-specific DNAm at CpG-SNP rs3781286 mediated the relationship between the local haplotype and verbal fluency. Our findings do not confirm that the interrogated DNA fragment is a place where genetic and environmental risk factors converge to influence schizophrenia symptoms through DNAm.

Human papillomavirus integration perspective in small cell cervical carcinoma

Small cell cervical carcinoma (SCCC) is a rare but aggressive malignancy. Here, we report human papillomavirus features and genomic landscape in SCCC via high-throughput HPV captured sequencing, whole-genome sequencing, whole-transcriptome sequencing, and OncoScan microarrays. HPV18 infections and integrations are commonly detected. Besides MYC family genes (37.9%), we identify SOX (8.4%), NR4A (6.3%), ANKRD (7.4%), and CEA (3.2%) family genes as HPV-integrated hotspots. We construct the genomic local haplotype around HPV-integrated sites, and find tandem duplications and amplified HPV long control regions (LCR). We propose three prominent HPV integration patterns: duplicating oncogenes (MYCN, MYC, and NR4A2), forming fusions (FGFR3–TACC3 and ANKRD12–NDUFV2), and activating genes (MYC) via the cis-regulations of viral LCRs. Moreover, focal CNA amplification peaks harbor canonical cancer genes including the HPV-integrated hotspots within MYC family, SOX2, and others. Our findings may provide potential molecular criteria for the accurate diagnosis and efficacious therapies for this lethal disease.

REViewer: haplotype-resolved visualization of read alignments in and around tandem repeats

BackgroundExpansions of short tandem repeats are the cause of many neurogenetic disorders including familial amyotrophic lateral sclerosis, Huntington disease, and many others. Multiple methods have been recently developed that can identify repeat expansions in whole genome or exome sequencing data. Despite the widely recognized need for visual assessment of variant calls in clinical settings, current computational tools lack the ability to produce such visualizations for repeat expansions. Expanded repeats are difficult to visualize because they correspond to large insertions relative to the reference genome and involve many misaligning and ambiguously aligning reads.ResultsWe implemented REViewer, a computational method for visualization of sequencing data in genomic regions containing long repeat expansions and FlipBook, a companion image viewer designed for manual curation of large collections of REViewer images. To generate a read pileup, REViewer reconstructs local haplotype sequences and distributes reads to these haplotypes in a way that is most consistent with the fragment lengths and evenness of read coverage. To create appropriate training materials for onboarding new users, we performed a concordance study involving 12 scientists involved in short tandem repeat research. We used the results of this study to create a user guide that describes the basic principles of using REViewer as well as a guide to the typical features of read pileups that correspond to low confidence repeat genotype calls. Additionally, we demonstrated that REViewer can be used to annotate clinically relevant repeat interruptions by comparing visual assessment results of 44 FMR1 repeat alleles with the results of triplet repeat primed PCR. For 38 of these alleles, the results of visual assessment were consistent with triplet repeat primed PCR.ConclusionsRead pileup plots generated by REViewer offer an intuitive way to visualize sequencing data in regions containing long repeat expansions. Laboratories can use REViewer and FlipBook to assess the quality of repeat genotype calls as well as to visually detect interruptions or other imperfections in the repeat sequence and the surrounding flanking regions. REViewer and FlipBook are available under open-source licenses at https://github.com/illumina/REViewer and https://github.com/broadinstitute/flipbook respectively.

Exploring DNA methylation within the CYP17A gene as a potential mediator between childhood adversity and stress-related phenotypes in schizophrenia

IntroductionStress caused by childhood adversity (CA) is known to contribute to schizophrenia risk and symptoms. Its effects might be mediated by epigenetic mechanisms, specifically DNA methylation (meDNA) within relevant genes, and predominantly influence the hippocampus and prefrontal cortex (PFC). CYP17A1 is a candidate, as it situates within a schizophrenia risk locus and is involved in glucocorticoid synthesis.ObjectivesTo explore meDNA within CYP17A and its relations to hippocampus- and PFC-dependent schizophrenia symptoms: depression and deficits of declarative memory and executive functions.MethodsWe assessed meDNA at each CpG within a CYP17A fragment (chr10:104594471-104595887, hg19) in blood of 66 schizophrenia patients using the third-generation sequencing. Immediate memory, depression, cognitive shifting and cognitive inhibition (CI) were assessed with the RAVLT, PANSS, TMT-B and Stroop word-color test, respectively. ANCOVA and regression models adjusted for sex and age were applied to explore the relations between the phenotypes, local haplotype, meDNA and CA, defined as the presence of parental alcoholism or psychiatric illness.ResultsMeDNA at CpG-SNP rs3781286 correlated with CI (corrected p=0.01). However, there were no main or interaction effects of CA either on meDNA at this site or on CI. Both CI and meDNA associated with haplotype, but subsequent analysis showed that meDNA did not mediate the relation between haplotype and CI.ConclusionsOur findings suggest that CYP17A associates with PFC-dependent cognitive deficits in schizophrenia but did not support the hypothesis that CA plays a role in this association via meDNA or any other mechanism. Grant support: 21-15-00124/Russian Science Foundation https://rscf.ru/project/21-15-00124/.DisclosureNo significant relationships.

Cholesterol and matrisome pathways dysregulated in astrocytes and microglia

The impact of apolipoprotein E ε4 (APOE4), the strongest genetic risk factor for Alzheimer's disease (AD), on human brain cellular function remains unclear. Here, we investigated the effects of APOE4 on brain cell types derived from population and isogenic human induced pluripotent stem cells, post-mortem brain, and APOE targeted replacement mice. Population and isogenic models demonstrate that APOE4 local haplotype, rather than a single risk allele, contributes to risk. Global transcriptomic analyses reveal human-specific, APOE4-driven lipid metabolic dysregulation in astrocytes and microglia. APOE4 enhances de novo cholesterol synthesis despite elevated intracellular cholesterol due to lysosomal cholesterol sequestration in astrocytes. Further, matrisome dysregulation is associated with upregulated chemotaxis, glial activation, and lipid biosynthesis in astrocytes co-cultured with neurons, which recapitulates altered astrocyte matrisome signaling in human brain. Thus, APOE4 initiates glia-specific cell and non-cell autonomous dysregulation that may contribute to increased AD risk.

SNP and Haplotype Regional Heritability Mapping (SNHap-RHM): Joint Mapping of Common and Rare Variation Affecting Complex Traits.

We describe a genome-wide analytical approach, SNP and Haplotype Regional Heritability Mapping (SNHap-RHM), that provides regional estimates of the heritability across locally defined regions in the genome. This approach utilises relationship matrices that are based on sharing of SNP and haplotype alleles at local haplotype blocks delimited by recombination boundaries in the genome. We implemented the approach on simulated data and show that the haplotype-based regional GRMs capture variation that is complementary to that captured by SNP-based regional GRMs, and thus justifying the fitting of the two GRMs jointly in a single analysis (SNHap-RHM). SNHap-RHM captures regions in the genome contributing to the phenotypic variation that existing genome-wide analysis methods may fail to capture. We further demonstrate that there are real benefits to be gained from this approach by applying it to real data from about 20,000 individuals from the Generation Scotland: Scottish Family Health Study. We analysed height and major depressive disorder (MDD). We identified seven genomic regions that are genome-wide significant for height, and three regions significant at a suggestive threshold (p-value < 1 × 10−5) for MDD. These significant regions have genes mapped to within 400 kb of them. The genes mapped for height have been reported to be associated with height in humans. Similarly, those mapped for MDD have been reported to be associated with major depressive disorder and other psychiatry phenotypes. The results show that SNHap-RHM presents an exciting new opportunity to analyse complex traits by allowing the joint mapping of novel genomic regions tagged by either SNPs or haplotypes, potentially leading to the recovery of some of the “missing” heritability.

Unravelling cucumber resistance to several viruses via genome-wide association studies highlighted resistance hotspots and new QTLs.

The mapping and introduction of sustainable resistance to viruses in crops is a major challenge in modern breeding, especially regarding vegetables. We hence assembled a panel of cucumber elite lines and landraces from different horticultural groups for testing with six virus species. We mapped 18 quantitative trait loci (QTL) with a multiloci genome wide association studies (GWAS), some of which have already been described in the literature. We detected two resistance hotspots, one on chromosome 5 for resistance to the cucumber mosaic virus (CMV), cucumber vein yellowing virus (CVYV), cucumber green mottle mosaic virus (CGMMV) and watermelon mosaic virus (WMV), colocalizing with the RDR1 gene, and another on chromosome 6 for resistance to the zucchini yellowing mosaic virus (ZYMV) and papaya ringspot virus (PRSV) close to the putative VPS4 gene location. We observed clear structuring of resistance among horticultural groups due to plant virus coevolution and modern breeding which have impacted linkage disequilibrium (LD) in resistance QTLs. The inclusion of genetic structure in GWAS models enhanced the GWAS accuracy in this study. The dissection of resistance hotspots by local LD and haplotype construction helped gain insight into the panel's resistance introduction history. ZYMV and CMV resistance were both introduced from different donors in the panel, resulting in multiple resistant haplotypes at same locus for ZYMV, and in multiple resistant QTLs for CMV.

Discovery of the REN11 Locus From Vitis aestivalis for Stable Resistance to Grapevine Powdery Mildew in a Family Segregating for Several Unstable and Tissue-Specific Quantitative Resistance Loci.

Race-specific resistance loci, whether having qualitative or quantitative effects, present plant-breeding challenges for phenotypic selection and deciding which loci to select or stack with other resistance loci for improved durability. Previously, resistance to grapevine powdery mildew (GPM, caused by Erysiphe necator) was predicted to be conferred by at least three race-specific loci in the mapping family B37-28 × C56-11 segregating for GPM resistance from Vitis aestivalis. In this study, 9 years of vineyard GPM disease severity ratings plus a greenhouse and laboratory assays were genetically mapped, using a rhAmpSeq core genome marker platform with 2,000 local haplotype markers. A new qualitative resistance locus, named REN11, on the chromosome (Chr) 15 was found to be effective in nearly all (11 of 12) vineyard environments on leaves, rachis, berries, and most of the time (7 of 12) stems. REN11 was independently validated in a pseudo-testcross with the grandparent source of resistance, “Tamiami.” Five other loci significantly predicted GPM severity on leaves in only one or two environments, which could indicate race-specific resistance or their roles in different timepoints in epidemic progress. Loci on Chr 8 and 9 reproducibly predicted disease severity on stems but not on other tissues and had additive effects with REN11 on the stems. The rhAmpSeq local haplotype sequences published in this study for REN11 and Chr 8 and 9 stem quantitative trait locus (QTL) can be used directly for marker-assisted selection or converted to SNP assays. In screening for REN11 in a diversity panel of 20,651 vines representing the diversity of Vitis, this rhAmpSeq haplotype had a false positive rate of 0.034% or less. The effects of the other foliar resistance loci detected in this study seem too unstable for genetic improvement regardless of quantitative effect size, whether due to race specificity or other environmental variables.

Molecular epidemiology of Plasmodium falciparum by multiplexed amplicon deep sequencing in Senegal

BackgroundMolecular epidemiology can provide important information regarding the genetic diversity and transmission of Plasmodium falciparum, which can assist in designing and monitoring elimination efforts. However, malaria molecular epidemiology including understanding the genetic diversity of the parasite and performing molecular surveillance of transmission has been poorly documented in Senegal. Next Generation Sequencing (NGS) offers a practical, fast and high-throughput approach to understand malaria population genetics. This study aims to unravel the population structure of P. falciparum and to estimate the allelic diversity, multiplicity of infection (MOI), and evolutionary patterns of the malaria parasite using the NGS platform.MethodsMultiplex amplicon deep sequencing of merozoite surface protein 1 (PfMSP1) and merozoite surface protein 2 (PfMSP2) in fifty-three P. falciparum isolates from two epidemiologically different areas in the South and North of Senegal, was carried out.ResultsA total of 76 Pfmsp1 and 116 Pfmsp2 clones were identified and 135 different alleles were found, 56 and 79 belonged to the pfmsp1 and pfmsp2 genes, respectively. K1 and IC3D7 allelic families were most predominant in both sites. The local haplotype diversity (Hd) and nucleotide diversity (π) were higher in the South than in the North for both genes. For pfmsp1, a high positive Tajima’s D (TD) value was observed in the South (D = 2.0453) while negative TD value was recorded in the North (D = − 1.46045) and F-Statistic (Fst) was 0.19505. For pfmsp2, non-directional selection was found with a highly positive TD test in both areas and Fst was 0.02111. The mean MOI for both genes was 3.07 and 1.76 for the South and the North, respectively, with a statistically significant difference between areas (p = 0.001).ConclusionThis study revealed a high genetic diversity of pfmsp1 and pfmsp2 genes and low genetic differentiation in P. falciparum population in Senegal. The MOI means were significantly different between the Southern and Northern areas. Findings also showed that multiplexed amplicon deep sequencing is a useful technique to investigate genetic diversity and molecular epidemiology of P. falciparum infections.

Somatic variant analysis of linked-reads sequencing data with Lancet.

We present a new version of the popular somatic variant caller, Lancet, that supports the analysis of linked-reads sequencing data. By seamlessly integrating barcodes and haplotype read assignments within the colored De Bruijn graph local-assembly framework, Lancet computes a barcode-aware coverage and identifies variants that disagree with the local haplotype structure. Lancet is implemented in C++ and available for academic and non-commercial research purposes as an open-source package at https://github.com/nygenome/lancet. Supplementary data are available at Bioinformatics online.

Two Synthetic 18-Way Outcrossed Populations of Diploid Budding Yeast with Utility for Complex Trait Dissection.

Advanced-generation multiparent populations (MPPs) are a valuable tool for dissecting complex traits, having more power than genome-wide association studies to detect rare variants and higher resolution than F2 linkage mapping. To extend the advantages of MPPs in budding yeast, we describe the creation and characterization of two outbred MPPs derived from 18 genetically diverse founding strains. We carried out de novo assemblies of the genomes of the 18 founder strains, such that virtually all variation segregating between these strains is known, and represented those assemblies as Santa Cruz Genome Browser tracks. We discovered complex patterns of structural variation segregating among the founders, including a large deletion within the vacuolar ATPase VMA1, several different deletions within the osmosensor MSB2, a series of deletions and insertions at PRM7 and the adjacent BSC1, as well as copy number variation at the dehydrogenase ALD2 Resequenced haploid recombinant clones from the two MPPs have a median unrecombined block size of 66 kb, demonstrating that the population is highly recombined. We pool-sequenced the two MPPs to 3270× and 2226× coverage and demonstrated that we can accurately estimate local haplotype frequencies using pooled data. We further downsampled the pool-sequenced data to ∼20-40× and showed that local haplotype frequency estimates remained accurate, with median error rates 0.8 and 0.6% at 20× and 40×, respectively. Haplotypes frequencies are estimated much more accurately than SNP frequencies obtained directly from the same data. Deep sequencing of the two populations revealed that 10 or more founders are present at a detectable frequency for > 98% of the genome, validating the utility of this resource for the exploration of the role of standing variation in the architecture of complex traits.

A Random Forests Framework for Modeling Haplotypes as Mosaics of Reference Haplotypes

Many genomic data analyses such as phasing, genotype imputation, or local ancestry inference share a common core task: matching pairs of haplotypes at any position along the chromosome, thereby inferring a target haplotype as a succession of pieces from reference haplotypes, commonly called a mosaic of reference haplotypes. For that purpose, these analyses combine information provided by linkage disequilibrium, linkage and/or genealogy through a set of heuristic rules or, most often, by a hidden Markov model. Here, we develop an extremely randomized trees framework to address the issue of local haplotype matching. In our approach, a supervised classifier using extra-trees (a particular type of random forests) learns how to identify the best local matches between haplotypes using a collection of observed examples. For each example, various features related to the different sources of information are observed, such as the length of a segment shared between haplotypes, or estimates of relationships between individuals, gametes, and haplotypes. The random forests framework was fed with 30 relevant features for local haplotype matching. Repeated cross-validations allowed ranking these features in regard to their importance for local haplotype matching. The distance to the edge of a segment shared by both haplotypes being matched was found to be the most important feature. Similarity comparisons between predicted and true whole-genome sequence haplotypes showed that the random forests framework was more efficient than a hidden Markov model in reconstructing a target haplotype as a mosaic of reference haplotypes. To further evaluate its efficiency, the random forests framework was applied to imputation of whole-genome sequence from 50k genotypes and it yielded average reliabilities similar or slightly better than IMPUTE2. Through this exploratory study, we lay the foundations of a new framework to automatically learn local haplotype matching and we show that extra-trees are a promising approach for such purposes. The use of this new technique also reveals some useful lessons on the relevant features for the purpose of haplotype matching. We also discuss potential improvements for routine implementation.