Imputation strategies for genomic prediction using nanopore sequencing.

Area of Science:

• Genomics
• Animal Breeding
• Bioinformatics

Background:

• Genomic prediction utilizes SNP genotypes for complex trait prediction in various species.
• Genotyping-by-sequencing (GBS) with genotype imputation is a growing method for genomic prediction.
• Oxford Nanopore Technologies (ONT) MinION offers portable and rapid GBS.

Purpose of the Study:

• Evaluate speed and accuracy of genomic predictions using low-coverage ONT sequence data in cattle.
• Assess four imputation approaches and the effect of SNP reference panel size on imputation performance.

Main Methods:

• Used SNP array and ONT sequence data from 62 beef heifers for genomic estimated breeding value (GEBV) calculations.
• Employed four imputation methods, including the QUILT package.
• Investigated varying sequencing coverages (as low as 0.1×) and SNP reference panel sizes (up to 48 million SNPs).

Main Results:

• GEBV accuracy significantly increased when using genome-wide flanking SNPs from sequence data for imputation.
• Correlations between ONT and low-density SNP array GEBVs exceeded 0.91, reaching 0.97 at 0.1× coverage.
• Imputation time was reduced by decreasing flanking sequence SNPs; 0.1× coverage GBS was more accurate than low-density SNP array imputation.

Conclusions:

• Accurate genomic prediction is achievable with ONT sequence data at coverages as low as 0.1×.
• Imputation can be performed rapidly, with times as short as 10 minutes per sample.
• Low-coverage GBS (0.1×) can outperform imputation from low-density SNP arrays in cattle.