Genomic Variance Estimation Based on Genotyping-by-Sequencing with Different Coverage in Perennial Ryegrass

TY - ABST

T1 - Genomic Variance Estimation Based on Genotyping-by-Sequencing with Different Coverage in Perennial Ryegrass

AU - Ashraf, Bilal

AU - Fé, Dario

AU - Jensen, Just

AU - Byrne, Stephen

AU - Asp, Torben

AU - Greve-Pedersen, Morten

AU - Lenk, Ingo

AU - Jensen, Christian Sig

AU - Janss, Luc

PY - 2014/1/11

Y1 - 2014/1/11

N2 - Advancement in next generation sequencing (NGS) technologies has significantly decreased the cost of DNA sequencing enabling increased use of genotyping by sequencing (GBS) in several plant species. In contrast to array-based genotyping GBS also allows for easy estimation of allele frequencies at each SNP in family pools or polyploids. There are, however, several statistical challenges associated with this method, including low sequencing depth and missing values. Low sequencing depth results in inaccuracies in estimates of allele frequencies for each SNP. In this work we have focused on optimizing methods and models utilizing F2 family phenotype records and NGS information from F2 family pools in perennial ryegrass. Genomic variance was estimated using genomic relationship matrices based on different coverage depths to verify effects of coverage depth. Example traits were seed yield, rust score and heading date. A total of 995 F2 families were genotyped via GBS, resulting in allele frequency estimates at 1 million SNPs in each family, the coverage within family ranging from 0 to 60. Results from both real and simulated data show that genomic variance is overestimated at lower coverage depth. This overestimation is caused by inflated diagonals in the genomic relationship matrix based on sequence data. The estimate of environmental variance is independent of sequencing coverage.We conclude that it is feasible to estimate genetic variance among F2 families using an NGS approach, which is promising for the application of genomic prediction and genomic selection

AB - Advancement in next generation sequencing (NGS) technologies has significantly decreased the cost of DNA sequencing enabling increased use of genotyping by sequencing (GBS) in several plant species. In contrast to array-based genotyping GBS also allows for easy estimation of allele frequencies at each SNP in family pools or polyploids. There are, however, several statistical challenges associated with this method, including low sequencing depth and missing values. Low sequencing depth results in inaccuracies in estimates of allele frequencies for each SNP. In this work we have focused on optimizing methods and models utilizing F2 family phenotype records and NGS information from F2 family pools in perennial ryegrass. Genomic variance was estimated using genomic relationship matrices based on different coverage depths to verify effects of coverage depth. Example traits were seed yield, rust score and heading date. A total of 995 F2 families were genotyped via GBS, resulting in allele frequency estimates at 1 million SNPs in each family, the coverage within family ranging from 0 to 60. Results from both real and simulated data show that genomic variance is overestimated at lower coverage depth. This overestimation is caused by inflated diagonals in the genomic relationship matrix based on sequence data. The estimate of environmental variance is independent of sequencing coverage.We conclude that it is feasible to estimate genetic variance among F2 families using an NGS approach, which is promising for the application of genomic prediction and genomic selection

M3 - Conference abstract in proceedings

SP - 157

EP - 157

BT - Proceeding of Plant & Animal Genome XXII

CY - San Diego, USA

ER -