Comparing BeadChip and WGS Genotyping: Non-Technical Failed Calling Is Attributable to Additional Variation within the Probe Target Sequence

Abstract

Microarray-based genomic selection is a central tool to increase the genetic gain of economically significant traits in dairy cattle. Yet, the effectivity of this tool is slightly limited, as estimates based on genotype data only partially explain the observed heritability. In the analysis of the genomes of 17 Israeli Holstein bulls, we compared genotyping accuracy between whole-genome sequencing (WGS) and microarray-based techniques. Using the standard GATK pipeline, the short-variant discovery within sequence reads mapped to the reference genome (ARS-UCD1.2) was compared to the genotypes from Illumina BovineSNP50 BeadChip and to an alternative method, which computationally mimics the hybridization procedure by mapping reads to 50 bp spanning the BeadChip source sequences. The number of mismatches between the BeadChip and WGS genotypes was low (0.2%). However, 17,197 (40% of the informative SNPs) had extra variation within 50 bp of the targeted SNP site, which might interfere with hybridization-based genotyping. Consequently, with respect to genotyping errors, BeadChip varied significantly and systematically from WGS genotyping, introducing null allele-like effects and Mendelian errors (<0.5%), whereas the GATK algorithm of local de novo assembly of haplotypes successfully resolved the genotypes in the extra-variable regions. These findings suggest that the microarray design should avoid polymorphic genomic regions that are prone to extra variation and that WGS data may be used to resolve erroneous genotyping, which may partially explain missing heritability.

Re-Evaluation of Genotyping Methodologies in Cattle: The Proficiency of Imputation

“…To further validate the findings of a QTL for RFI on BTA14 that have been based on a sample of 114 cows [ 12 ], we enlarged this sample to 192 cows all genotyped with the Illumina BovineHD BeadChip (777K SNPs). We also performed whole genome sequencing (WGS) for 19 Israeli Holstein sires, representing major lineages of the Israeli herd [ 20 ] with no RFI evaluations.…”

Complex population structure and haplotype patterns in the Western European honey bee from sequencing a large panel of haploid drones

“…To investigate the effects of the high SNP density found on the honey bee genome on the potential quality of the genotyping results using the chip, we looked for additional SNPs close to each target SNP that could interfere with the genotyping process and cause allele dropout. Indeed, additional variation in the probe sequence within 50 bp flanking the target SNP will interfere with the hybridization‐based genotyping (Gershoni et al, 2022 ) and it is therefore common practice to have at most one extra variant within 50 bp either side of the target SNP. For SNPs with an MAF >0.2, we find that 51,178 out of 61,320 meet this criterion (16.5% of markers lost).…”

“…Indeed, additional variation in the probe sequence within 50 bp flanking the target SNP will interfere with the hybridization-based genotyping (Gershoni et al, 2022) Conversely, for lower density chips, the spacing of markers can be optimized by taking the haplotype structure into account, thus avoiding redundancy while maintaining the highest possible level of genetic information. Another advantage of sequencing haploid samples is that the whole data set represents phased chromosomes.…”

Genetic and Genomic Analysis of Cow Mortality in the Israeli Holstein Population

“…As genotyping of these bulls was performed using several SNP chip platforms with different qualities and coverage, we performed imputation using 2114 Holstein bulls with genomic evaluations. A total of 50,392 genetic markers were selected for further analysis, according to [ 22 ]. Following additional formatting to fit the requirements of the imputation software, the imputation process was carried out using the findhap.f90 v3.0 program [ 23 ] (USDA-ARS, Beltsville, MD, USA, , accessed on 2 January 2023).…”