Scholarlink

Background <p>There is a lack of genotype imputation software tailored specifically for polyploids and pooled samples without phased haplotype information and reference panel of genotypes. Numerous important crops are polyploids, while pool sequencing is an emerging cost-effective approach for the genomic characterisation of breeding populations, families, and other segregating lines. The scarcity of imputation tools for these datasets results in the reliance on diploid-specific software, potentially leading to suboptimal outcomes. This necessitates the development of allele frequency imputation tools which accommodate the unique computational challenges presented by polyploid genomes and pooled sequencing data.</p> Results <p>We developed imputef, an allele frequency imputation tool for polyploid individuals and pooled samples lacking the rich genomic information typically available to model species. Missing allele frequencies are imputed using the genetic distance-weighted mean of non-missing allele frequencies from k-nearest neighbours. Genetic distance is estimated as the mean absolute difference in allele frequencies across linked loci, with linkage estimated as Pearson’s correlation between loci. The minimum loci correlation and maximum genetic distance thresholds can be optimised per locus to minimise imputation error.</p> <p>Imputef using the default parameters (minimum loci correlation, maximum genetic distance, minimum number of linked loci, and minimum number of nearest neighbours set to 0.9, 0.1, 20, and 5, respectively), generally outperformed mean value imputation and performed well across the range of sparsity levels (1% to 90% simulated missing data, and 6% to 50% empirical sparsity levels), and minor allele frequencies (1% and 5%). Imputation accuracy was further enhanced using the built-in per-locus optimisation. The maximum computation time ranged from 41 h (380 tetraploids; 233,627 loci) without optimisation to 6 days (275 polyploids; 1,526,917 loci) with per locus optimisation using 32 compute cores at 3.7 GHz and 60 to 320 GB of memory. Additionally, practical imputation approaches for large datasets with unordered loci were demonstrated using imputef.</p> Conclusion <p>Imputef is an accurate allele frequency imputation tool for polyploid individuals and pooled samples lacking rich genomic information. This tool is expected to improve the statistical power of genomic analyses involving polyploids and pools, and enhance the adoption of pool-based methods in crop breeding and ecological genomics.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Imputef: imputation of polyploid genotype classes and allele frequencies

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月22日

查看全文
  • Jefferson Paril,
  • Noel O. I. Cogan,
  • M. Michelle Malmberg

摘要

Background

There is a lack of genotype imputation software tailored specifically for polyploids and pooled samples without phased haplotype information and reference panel of genotypes. Numerous important crops are polyploids, while pool sequencing is an emerging cost-effective approach for the genomic characterisation of breeding populations, families, and other segregating lines. The scarcity of imputation tools for these datasets results in the reliance on diploid-specific software, potentially leading to suboptimal outcomes. This necessitates the development of allele frequency imputation tools which accommodate the unique computational challenges presented by polyploid genomes and pooled sequencing data.

Results

We developed imputef, an allele frequency imputation tool for polyploid individuals and pooled samples lacking the rich genomic information typically available to model species. Missing allele frequencies are imputed using the genetic distance-weighted mean of non-missing allele frequencies from k-nearest neighbours. Genetic distance is estimated as the mean absolute difference in allele frequencies across linked loci, with linkage estimated as Pearson’s correlation between loci. The minimum loci correlation and maximum genetic distance thresholds can be optimised per locus to minimise imputation error.

Imputef using the default parameters (minimum loci correlation, maximum genetic distance, minimum number of linked loci, and minimum number of nearest neighbours set to 0.9, 0.1, 20, and 5, respectively), generally outperformed mean value imputation and performed well across the range of sparsity levels (1% to 90% simulated missing data, and 6% to 50% empirical sparsity levels), and minor allele frequencies (1% and 5%). Imputation accuracy was further enhanced using the built-in per-locus optimisation. The maximum computation time ranged from 41 h (380 tetraploids; 233,627 loci) without optimisation to 6 days (275 polyploids; 1,526,917 loci) with per locus optimisation using 32 compute cores at 3.7 GHz and 60 to 320 GB of memory. Additionally, practical imputation approaches for large datasets with unordered loci were demonstrated using imputef.

Conclusion

Imputef is an accurate allele frequency imputation tool for polyploid individuals and pooled samples lacking rich genomic information. This tool is expected to improve the statistical power of genomic analyses involving polyploids and pools, and enhance the adoption of pool-based methods in crop breeding and ecological genomics.