Abstract
Haplotypes from sequencing data may improve the prediction accuracy in genomic evaluations as haplotypes are in stronger linkage disequilibrium with quantitative trait loci than markers from SNP chips. This study focuses first, on the creation of haplotypes in a population sample of 450 Holstein animals, with full-sequence data from the 1000 bull genomes project; and second, on incorporating them into the whole genome prediction model. In total, 38,319,258 SNPs (and indels) from Next Generation Sequencing were included in the analysis. After filtering variants with minor allele frequency (MAF< 0.025) 13,912,326 SNPs were available for the haplotypes extraction with findhap.f90. The number of SNPs in the haploblocks was on average 924 SNP (166,552 bp). Unique haplotypes were around 97% in all chromosomes and were ignored leaving 153,428 haplotypes. Estimated haplotypes had a large contribution to the total variance of genomic estimated breeding values for kilogram of protein, Global Type Index, Somatic Cell Score and Days Open (between 32 and 99.9%). Haploblocks containing haplotypes with large effects were selected by filtering for each trait, haplotypes whose effect was larger/lower than the mean plus/minus 3 times the standard deviation (SD) and 1 SD above the mean of the haplotypes effect distribution. Results showed that filtering by 3 SD would not be enough to capture a large proportion of genetic variance, whereas filtering by 1 SD could be useful but model convergence should be considered. Additionally, sequence haplotypes were able to capture additional genetic variance to the polygenic effect for traits undergoing lower selection intensity like fertility and health traits.
Highlights
New technological advances such as Single Nucleotide Polymorphism (SNP) discovery using high-throughput SNP genotyping has led to a new strategy of selection called genomic selection (GS) that has revolutionized breeding in some species such as dairy cattle in the last decade (Hayes et al, 2009; Ibañez-Escriche & Gonzalez-Recio, 2011)
The BTA1 showed the highest number of haplotype blocks (961) as well as haplotypes (9363), while BTA25 showed the smallest number of blocks (261) and haplotypes (2788)
These haplotypes were not used in this analysis due to the difficulty of finding statistical effects when the haplotype is present in only a couple of individuals in our sample
Summary
New technological advances such as Single Nucleotide Polymorphism (SNP) discovery using high-throughput SNP genotyping has led to a new strategy of selection called genomic selection (GS) that has revolutionized breeding in some species such as dairy cattle in the last decade (Hayes et al, 2009; Ibañez-Escriche & Gonzalez-Recio, 2011). The achievable genetic gain is proportional to the accuracy of predictions, which depends on the proportion of the genetic variance that can be captured by the marker information. This is function of the linkage disequilibrium (LD) between the SNP and the causative mutations affecting the trait, size of the genotyped population, heritability of the traits, among other factors (Druet et al, 2014). Druet et al (2014) showed that the accuracy of genomic breeding value may improve in the range of 2-30% (depending on the trait), if the variation from rare alleles could be captured from the whole genome sequence data and exploited in genomic predictions. Whole genome imputation accuracy is larger than 0.95, except for variants with minor allele frequency (MAF)>0.10, which can be as low as 0.50 (Van Binsbergen et al, 2014)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
