The use of a genomic relationship matrix for breed assignment of cattle breeds: comparison and combination with a machine learning method.

To develop a breed assignment model, three main steps are generally followed: 1) The selection of breed informative single nucleotide polymorphism (SNP); 2) The training of a model, based on a reference population, that allows to classify animals to their breed of origin; and 3) The validation of the developed model on external animals i.e., that were not used in previous steps. However, there is no consensus in the literature about which methodology to follow for the first step, nor about the number of SNP to be selected. This can raise many questions when developing the model and lead to the use of sophisticated methodologies for selecting SNP (e.g., with iterative algorithms, partitions of SNP, or combination of several methods). Therefore, it may be of interest to avoid the first step by the use of all the available SNP. For this purpose, we propose the use of a genomic relationship matrix (GRM), combined or not with a machine learning method, for breed assignment. We compared it with a previously developed model based on selected informative SNP. Four methodologies were investigated: 1) The PLS_NSC methodology: selection of SNP based on a partial least square-discriminant analysis (PLS-DA) and breed assignment by classification based on the nearest shrunken centroids (NSC) method; 2) Breed assignment based on the highest mean relatedness of an animal to the reference populations of each breed (referred to mean_GRM); 3) Breed assignment based on the highest SD of the relatedness of an animal to the reference populations of each breed (referred to SD_GRM) and 4) The GRM_SVM methodology: the use of means and SD of the relatedness defined in mean_GRM and SD_GRM methodologies combined with the linear support vector machine (SVM), a machine learning method used for classification. Regarding mean global accuracies, results showed that the use of mean_GRM or GRM_SVM was not significantly different (Bonferroni corrected P > 0.0083) than the model based on a reduced SNP panel (PLS_NSC). Moreover, the mean_GRM and GRM_SVM methodology were more efficient than PLS_NSC as it was faster to compute. Therefore, it is possible to bypass the selection of SNP and, by the use of a GRM, to develop an efficient breed assignment model. In routine, we recommend the use of GRM_SVM over mean_GRM as it gave a slightly increased global accuracy, which can help endangered breeds to be maintained. The script to execute the different methodologies can be accessed on: https://github.com/hwilmot675/Breed_assignment.

Breed assignment models generally rely on three main steps: 1) Selection of markers that allow to distinguish the breeds under study; 2) Development of a classification model that assigns each animal to its breed of origin; and 3) Validation of the developed model with new animals, to verify that the developed model is not overfitted. The first step often raises several questions about the methodology to select the best markers or about the number of markers to select. That is why it can be interesting to avoid this first step and to use an appropriate methodology that performs similarly without the need for single nucleotide polymorphism (SNP) selection. In this study, we developed different methodologies based on the genomic relationship matrix (GRM), combined or not with a machine learning method, to assign animals to their breed of origin. The results showed that the model based on a GRM combined with a machine learning method showed equivalent percentage of correct assignment to a previously developed model relying on SNP selection while being substantially faster to compute. It is therefore possible to assign animals to their breed by the use of a GRM and to bypass the first step of selection of SNP.

Genome-wide association for milk urea concentration in Dual-Purpose Belgian Blue cows.

The objectives of this study were to estimate genetic parameters and identify genomic regions associated with milk urea concentration (MU) in Dual-Purpose Belgian Blue (DPBB) cows. The data were 29,693 test-day records of milk yield (MY), fat yield (FY), protein yield (PY), fat percentage (FP), protein percentage (PP) and MU collected between 2014 and 2020 on 2498 first parity cows (16,935 test-day records) and 1939 second-parity cows (12,758 test-day records) from 49 herds in the Walloon Region of Belgium. Data of 28,266 single nucleotide polymorphisms (SNP), located on 29 Bos taurus autosomes (BTA), on 1699 animals (639 males and 1060 females) were used. Random regression test-day models were used to estimate genetic parameters through the Bayesian Gibbs sampling method using a single chain of 100,000 iterations after a burn-in period of 20,000. SNP solutions were estimated using a single-step genomic best linear unbiased prediction approach. The proportion of genetic variance explained by windows of 25 consecutive SNPs (with an average size of ~2 Mb) was calculated, and regions accounting for at least 1.0% of the total additive genetic variance were used to search for candidate genes. The mean (SD) of MU was 22.89 (10.07) and 22.35 (10.07) mg/dl for first and second parity, respectively. The mean (SD) heritability estimates for daily MU were 0.18 (0.01) and 0.22 (0.02), for first and second parity, respectively. The mean (SD) genetic correlations between daily MU and MY, FY, PY, FP and PP were -0.05 (0.09), -0.07 (0.11), -0.03 (0.13), -0.05 (0.08) and -0.03 (0.11) for first parity, respectively. The corresponding values estimated for second parity were 0.02 (0.10), -0.02 (0.09), 0.02 (0.08), -0.08 (0.06) and -0.05 (0.05). The genome-wide association analyses identified three genomic regions (BTA2, BTA3 and BTA13) associated with MU. The identified regions showed contrasting results between parities and among different stages within each parity. This suggests that different groups of candidate genes underlie the phenotypic expression of MU between parities and among different lactation stages within a parity. The results of this study can be used for future implementation and use of genomic evaluation to reduce MU in DPBB cows.

The pattern of linkage disequilibrium in Dual-Purpose Belgian Blue cattle.

Quantifying the level of linkage disequilibrium (LD), non-random association of alleles at two or more loci, is important to determine the number of markers needed for genomic selection. The aims of this study were to evaluate the extent of LD in Dual-Purpose Belgian Blue (DPBB) and to compare the level of LD in DPBB with that of Walloon Holstein. Data of 28,427 single nucleotide polymorphisms (SNP), located on 29 Bos taurus autosomes (BTA), of 639 DPBB and 398 Holstein bulls were used. The level of LD between pairwise SNPs separated by up to 10 Mb was evaluated, separately for each breed, using the squared correlation of the alleles at two loci. The analysis of molecular variance showed that the percentage of variation within populations (85.48%) was higher than between populations (14.52%). However, permutation tests showed a significant genetic differentiation between the two studied populations (p < .01). The average LD found between adjacent SNP pairs in DPBB (0.16 (SD = 0.22)) was generally lower than in Holstein (0.23 (SD = 0.27)). The proportion of SNPs in useful LD (r2  > 0.30) within a genomic distance of ≤0.10 Mb between SNPs was 18.58% and 28.23% in DPBB and Holstein bulls, respectively. In both breeds, the effective population size decreased over generations; however, the decline was greater in DPBB than that in Holstein. Based on results, it can be concluded that at least 68,000 SNPs are needed for implementing genomic selection in DPBB cattle with enough accuracy.

Development of a genomic tool for breed assignment by comparison of different classification models: Application to three local cattle breeds.

Assignment of individual cattle to a specific breed can often not rely on pedigree information. This is especially the case for local breeds for which the development of genomic assignment tools is required to allow individuals of unknown origin to be included to their herd books. A breed assignment model can be based on two specific stages: (a) the selection of breed-informative markers and (b) the assignment of individuals to a breed with a classification method. However, the performance of combination of methods used in these two stages has been rarely studied until now. In this study, the combination of 16 different SNP panels with four classification methods was developed on 562 reference genotypes from 12 cattle breeds. Based on their performances, best models were validated on three local breeds of interest. In cross-validation, 14 models had a global cross-validation accuracy higher than 90%, with a maximum of 98.22%. In validation, best models used 7,153 or 2,005 SNPs, based on a partial least squares-discriminant analysis (PLS-DA) and assigned individuals to breeds based on nearest shrunken centroids. The average validation sensitivity of the first two best models for the three local breeds of interest were 98.33% and 97.5%. Moreover, results reported in this study suggest that further studies should consider the PLS-DA method when selecting breed-informative SNPs.