High performance imputation of structural and single nucleotide variants using low-coverage whole genome sequencing

Area of Science:

• Population genomics
• Aquaculture genetics
• Bioinformatics

Background:

• Whole genome sequencing (WGS) offers advantages but hasn't replaced targeted genotyping for single nucleotide variants (SNVs).
• Structural variants (SVs) have significant trait effects but are difficult to genotype accurately.
• Low-coverage WGS with genotype imputation is a cost-effective strategy for genome-wide variant coverage, yet its application to SVs is underexplored.

Purpose of the Study:

• To investigate the efficacy of combined SNV and SV imputation using low-coverage WGS data in Atlantic salmon.
• To evaluate imputation performance across various WGS depths (1x–4x) for samples within and external to a reference panel.

Main Methods:

• Utilized a reference panel of 365 wild Atlantic salmon with high-confidence SNV and SV genotypes.
• Generated 15x WGS data for 20 commercial population samples external to the reference panel.
• Employed the GLIMPSE imputation method, assessing performance at WGS depths of 1x, 2x, 3x, and 4x.

Main Results:

• SNVs were imputed with high accuracy and recall across all tested WGS depths, even for samples outside the reference panel.
• SV imputation accuracy improved when incorporating SV genotype likelihoods (GLs) alongside SNV linkage disequilibrium (LD), with optimal performance at 3-4x depth.
• The combined strategy captured 84% of reference panel deletions with 87% accuracy at 1x depth; SV length impacted imputation performance, with longer SVs benefiting most from SV GLs.

Conclusions:

• Reference panel imputation using low-coverage WGS shows significant promise for genotyping both SNVs and SVs.
• This approach offers new avenues to enhance the resolution of genome-wide association studies by incorporating SV data.
• The findings support the cost-effectiveness and accuracy of low-coverage WGS for comprehensive genomic variant analysis in aquaculture.