Imputing Genotypes in Biallelic Populations from Low-Coverage Sequence Data

Abstract

Low-coverage next-generation sequencing methodologies are routinely employed to genotype large populations. Missing data in these populations manifest both as missing markers and markers with incomplete allele recovery. False homozygous calls at heterozygous sites resulting from incomplete allele recovery confound many existing imputation algorithms. These types of systematic errors can be minimized by incorporating depth-of-sequencing read coverage into the imputation algorithm. Accordingly, we developed Low-Coverage Biallelic Impute (LB-Impute) to resolve missing data issues. LB-Impute uses a hidden Markov model that incorporates marker read coverage to determine variable emission probabilities. Robust, highly accurate imputation results were reliably obtained with LB-Impute, even at extremely low (,13) average per-marker coverage. This finding will have implications for the design of genotype imputation algorithms in the future. LB-Impute is publicly available on GitHub at https://github.com/dellaportalaboratory/LB-Impute.KEYWORDS hidden Markov models; imputation; next-generation sequencing; population genetics; plant genomics T HE imputation of missing genotype data has been a key research topic in statistical genetics since well before the advent of next-generation sequencing (NGS) technologies. The goal of many of these algorithms was to reconstruct haplotypes from Sanger or microarray-based genotyping, usually on human populations. Strategies employing the expectation-maximization algorithm (Hawley and Kidd 1995;Long et al. 1995;Qin et al. 2002;Scheet and Stephens 2006), Bayesian inference Stephens and Donnelly 2003), or Markovian methodology (Stephens et al. 2001;Broman et al. 2003;Broman and Sen 2009), local ancestry and gametic phase, could be used to resolve missing markers within a population (Browning and Browning 2011). In these cases, missing genotypes were assigned based on the most likely proximal haplotypes. These computational methods greatly increased the informative content of genotyping information, especially for population studies (Spencer et al. 2009;Cleveland et al. 2011). While these programs were powerful and accurate, they also could be computationally expensive. Further, they assumed that available genotypes were largely correct, which could cause issues with sequencing data sets.The development of programs that focused primarily on the imputation of missing data and haplotype phasing was likely motivated by several factors. Genome-wide association studies could be enhanced by the inference of additional markers using large multipopulation data sets such as the International HapMap Project (International HapMap Consortium et al. 2010). The emergence of the meta-analysis led to a need for algorithms that could merge disparate data sets Howie et al. 2009;Li et al. 2010;Liu et al. 2013;Fuchsberger et al. 2015). These algorithms often employed large haplotype reference panels to improve imputation (Marchini et al. 2007;Browning and Browning 2009;Howie et al. 2009). In bialleli...

Accurate sequence variant genotyping in cattle using variation-aware genome graphs

“…Irrespective of the method evaluated, we observed heterozygous under-calling in animals that have been sequenced at low coverage, i.e., heterozygous variants were erroneously genotyped as homozygous due to an insufficient number of sequencing reads supporting the heterozygous genotype [10,[45][46][47]. In agreement with previous studies [2,5], Beagle imputation improved genotype concordance and reduced heterozygous under-calling particularly in cattle that had been sequenced at low coverage.…”

NOISYmputer: genotype imputation in bi-parental populations for noisy low-coverage next-generation sequencing data

“…Given the results obtained with simulated and real data, we recommend the use of NOISYmputer to impute noisy or high-quality GBS or WGS data. LB-Impute or Tassel-FSFHap can also be used for imputation of high-quality data, with a preference for LB-Impute in case of very low-coverage (< 0.4 X), as discussed in (Fragoso et al, 2016).…”

“…To test NOISYmputer and other methods' accuracy on real data, we used the high-quality Rice_GBS dataset. We compared map sizes to the map obtained with LB-Impute combined with BP-Impute and a custom R script that selects the most probable genotype, as extensive testing on this dataset showed that this method provided accurate imputation (Fragoso et al, 2016.…”

“…Very efficient methods have been recently developed to impute genotypic data derived from LC-NG assays in biparental populations. For instance, Tassel-FSFHap (Swarts et al, 2015), LB-Impute (Fragoso, Heffelfinger, Zhao, & Dellaporta, 2016) or Genotype-Corrector (Miao et al, 2018) can all address issues 1 and 2 accurately. Yet, we found in preliminary analyses that these methods can produce inaccurate results when the errors mentioned in issues 3 and 4 -thereafter called "noisy data" -are too frequent.…”

Accounting for Errors in Low Coverage High-Throughput Sequencing Data when Constructing Genetic Maps using Biparental Outcrossed Populations

“…Constructing linkage maps using sequencing data is complicated by the presence of two types of missing data that can result when the sequencing depth is low. The first is a missing genotype resulting from no alleles being called, while the second consists of a heterozygous genotype being called as homozygous due to only one of the parental alleles being sequenced at a particular locus (Fragoso et al . 2016; Dodds et al .…”