Bovinesnp50 beadchip

The 36 cows used for the RNA sequencing were also genotyped on the Illumina BovineSNP50 BeadChip (54 K, Illumina, San Diego). Blood samples were collected by certified personnel, and DNA extraction and genotyping on the Illumina BovineSNP50 BeadChip (54 K, Illumina, San Diego) were performed according to the manufacturer’s protocol. Genotypes were quality checked and imputed to sequence density as described above.

Samples were genotyped with the Illumina BovineSNP50 BeadChip [56] (link). Autosomal SNPs and a single pseudo-autosomal SNP were analyzed, because the data set from Gautier et al. [4] (link) excluded SNPs located exclusively on the X chromosome. We also filtered all SNPs which mapped to “chromosome unknown” of the UMD3.1 assembly [57] (link). In PLINK [58] (link), [59] , we removed SNPs with greater than 10% missing genotypes and with minor allele frequencies less than 0.0005 (1/[2*Number of Samples] = 0.000324, thus the minor allele had to be observed at least once in our data set). The average total genotype call rate in the remaining individuals was 0.993. Genotype data were deposited at DRYAD (doi:10.5061/dryad.th092) [60] .

A German Holstein genomic prediction population comprising 5024 bulls, provided by Vereinigte Informationssysteme Tierhaltung w.V., was used to validate our new strategy based on the BLUP|GA approach. All bulls were genotyped with the Illumina Bovine SNP50 Beadchip (Matukumalli et al. 2009 (link)). After quality control, 42,551 SNPs remained for our further analyses. Highly reliable conventional estimated breeding values (EBVs) of three traits, milk fat percentage (FP), milk yield (MY), and somatic cell score (SCS), were available for all bulls. Statistics of the phenotypes are shown in Table 1. These three traits were used in the present study because of the well-established knowledge from previous studies (Hu et al. 2013 (link); Zhang et al. 2014 (link)) and their representative distinct genetic architectures. They may represent three genetic architectures of complex traits that are composed of (1) one major gene and a large number of small effect loci (FP), (2) few moderate effect loci and many small effect loci (MY), and (3) many loci with small effects (SCS), respectively. EBVs of the three traits were used as “phenotypes” in our WGP model. To consider the performance of WGP methods in smaller population size, we randomly selected subsets of 4000, 2000, 1000, 500, 250, and 125 bulls from the full dataset, respectively. This datasets are available in File S1 and File S2.

We next sought to simulate phenotypes associated with genomic data and associated annotations (as described below) to evaluate our models. For this purpose, we used real Illumina Bovine SNP50 BeadChip genotyping data from $n=2605$ Montbéliarde bulls. Using these data as a base for our simulations has the advantage of including realistic population and linkage disequilibrium (LD) structures in our simulations. We excluded SNPs with a minor allele frequency (MAF) less than 0.01, leaving a total of $p=46,178$ SNPs. We divided individuals into learning and validation sets, respectively consisting of 80% of the oldest (2083 bulls) and 20% of the youngest bulls (522 bulls).

Among the animals with phenotypic records, 3,040 heifers and 4,483 cows were genotyped with Illumina BovineSNP50 BeadChip version 1 or 2 (Illumina Inc., San Diego, CA)^{23 (link)}. A total of 55,298 single nucleotide polymorphisms (SNPs) on 29 Bos taurus autosomes and the X chromosome were available. The numbers of SNPs per chromosome ranged from 894 on the X chromosome to 2,502 on the chromosome 1. A routine quality control was conducted using the software Plink^{24 (link)} to remove SNPs with minor allele frequencies <5%, call rate <90% or a significant deviation from the Hardy–Weinberg proportion (P-value < 10⁻⁵ for χ² test with one degree of freedom). After quality control, the number of SNPs remaining was 35,391. The UMD v3.1 assembly^{25 (link)} was used as the reference for genomic position of the SNPs.

All animals were genotyped with the Illumina BovineSNP50 BeadChip (Illumina Inc., San Diego, CA, USA). This 50 K SNP panel contained 41,847 SNPs after the quality control following [26 (link)], with an average of 51.5 Kb spacing between SNPs. Only autosomal chromosomes and SNPs with known genome positions according to the UMD_3.1 bovine assembly map [27 (link)] were used in the present study. The imputed high-density (777 K) SNP panel contained 305,341 SNPs evenly spaced across the genome after imputation from the 50 K SNP panel genotype.
The pedigree data contained 79,613 Holstein cattle from Canada, the USA, Germany, Denmark, Spain, Finland, France, Great Britain, Italy, and the Netherlands. There were 11,227 sires, 51,350 dams, and 9363 founder animals. Within the pedigree, there were 41,264 inbred animals with an average inbreeding coefficient of 0.0441, obtained using the software CFC [28 ].

A total of 122 French bulls were used as test set in this study. The GEBVs of milk yield, fat percentage, protein percentage, confirmation and feet_legs evaluated using American RP and French RP separately was provided by Gènes DIFFSUION. The GEBVs of these 122 French bulls using Chinese RPs were estimated in this study. The Chinese RP consisted of 1,568 Chinese cows with both genotype and phenotype. De-regressed proof (DRP) was used as the response variable for genomic prediction in this study. Genotypes of 270 French bulls, 270 American bulls and 270 Chinese bulls were used to compare the relationship among three populations. These 270 French bulls were the progenies of the imported French bulls and cows. So did the American bulls. The Chinese bulls were randomly selected from the native population. All the animals were genotyped using Illumina Bovine SNP50 BeadChip (Illumina, San Diego, CA, USA). After deleting SNPs with a minor allele frequency smaller than 0.01, 45,404 SNPs on 29 autosomes were retained.