The serious drawback underlying the biological annotation of whole-genome sequence data is the p >> n problem, which means that the number of polymorphic variants (p) is much larger than the number of available phenotypic records (n). We propose a way to circumvent the problem by combining a LASSO logistic regression with deep learning to classify cows as susceptible or resistant to mastitis, based on single nucleotide polymorphism (SNP) genotypes. Among several architectures, the one with 204,642 SNPs was selected as the best. This architecture was composed of two layers with, respectively, 7 and 46 units per layer implementing respective drop-out rates of 0.210 and 0.358. The classification of the test data resulted in AUC = 0.750, accuracy = 0.650, sensitivity = 0.600, and specificity = 0.700. Significant SNPs were selected based on the SHapley Additive exPlanation (SHAP). As a final result, one GO term related to the biological process and thirteen GO terms related to molecular function were significantly enriched in the gene set that corresponded to the significant SNPs. Our findings revealed that the optimal approach can correctly predict susceptibility or resistance status for approximately 65% of cows. Genes marked by the most significant SNPs are related to the immune response and protein synthesis.

Heritabilities of workability (WT) traits—milking speed (MS) and temperament (MT)—as well as genetic and phenotypic correlations between these traits in the population of Polish Holstein-Friesian (PHF) cows were estimated. The estimation of genetic parameters was performed twice: first with the use of pedigree data; and second with the use of pedigree and genomic data. Phenotypic data from routinely conducted MS and MT evaluations for 1,045,511 cows born from 2004 to 2013 were available; the cows were evaluated from 2011 to 2015. The main dataset was reduced based on imposed restrictions (e.g., on age of calving, stage of lactation and day of first trial milking). The dataset prepared in this manner comprised 391,615 cows. It was then reduced to daughters of 10% randomly selected sires for computational reasons. Finally, for genetic parameter estimation, 13,280 records of cows were used. The linear observation model included additive random effects of animal, fixed effects of herd-year-season of calving subclass (HYS) and lactation phase, fixed regressions on cow age at calving and the percent of HF breed genes in the cow genotype. Heritabilities estimated based on pedigree data were 0.12 (±0.0067) for MS and 0.08 (±0.0063) for MT, the genetic correlation between MS and MT was estimated at 0.05 (±0.0002) and the phenotypic correlation coefficient was estimated at 0.14 (±0.0004). The inclusion of genomic information of sire bulls had no clear effect on the size of the estimated WT genetic parameters. The heritabilities of MS and MT were 0.11 (±0.0065) and 0.09 (±0.0012), respectively. The genetic and phenotypic correlation coefficients were 0.07 (±0.0003) and 0.12 (±0.0005), respectively. The sizes of the obtained heritabilities of WT and of the genetic and phenotypic correlation between these traits indicate the possibility of effective population improvement for both WT traits.

Inbreeding was analysed in a population of 14,144 Polish Black-and-White Holstein-Friesian (PBWHF) young bulls born between 1994 and 2017 and bred under both conventional and genomic breeding programmes. The inbreeding coefficients were computed using a model with genetic groups, according to the algorithm given by VanRaden. It was found that in the analysed population all bulls are inbred (100% of the population), with the mean coefficient of inbreeding ranging from 0.09% to 26.95%. Pedigree analysis also showed a relationship between the changing number of bulls over the years and the dynamics of population inbreeding. These trends are connected with changes in the breeding scheme, related to the implementation of genomic selection in the breeding programme for PBWHF cattle in 2014. The increasing number of weaned young bulls in Poland was paralleled by a fairly consistent increase in the mean inbreeding, but the inbreeding dynamics were relatively small. A reverse trend was observed in the group of young bulls born after 2013. As the number of bulls very rapidly decreased in successive birth years, the mean inbreeding for successive birth-year groups very rapidly increased. As a result, the estimated linear trend was equal to 0.02% inbreeding per year of birth in the group of bulls raised before genomic selection (~20 birth-year) whereas in the group of bulls raised after genomic selection (~4 birth-year) the trend was much higher and amounted to 0.56% inbreeding per year of birth. The high mean inbreeding found in the group of the genomically selected young bulls may translate into higher inbreeding in the whole population of PBWHF cattle, because these bulls are now intensively used as sires. The results of our study also show that the implementation of genomic selection in the breeding programme caused a very rapid increase in the inbreeding rate per birth-year in young bulls.

Genome-wide patterns of copy number variation in case-control mastitis study of Polish HF cows using high-density Illumina BovineSNP50 v3 BeadChip array