Accurate determination of breed origin of alleles in a simulated smallholder crossbred dairy cattle population

Abstract

Background: Accurate assignment of breed origin of alleles (BOA) at a heterozygote locus may help to introduce a resilient or adaptive haplotype in crossbreeding. In this study, we developed and tested a method to assign breed of origin for individual alleles in crossbred dairy cattle. After generations of mating within and between local breeds as well as the importation of exotic bulls, five rounds of selected crossbred cows were simulated to mimic a dairy breeding program in the low- and middle-income countries (LMICs). In each round of selection, the alleles of those crossbred animals were phased and assigned to their breed of origin (being either local or exotic).

Results: Across all core lengths and modes of phasing (with offset-move 50% of the core length forward or no-offset), the average percentage of alleles correctly assigned a breed origin was 95.76%, with only 1.39% incorrectly assigned and 2.85% missing or unassigned. On consensus, the average percentage of alleles correctly assigned a breed origin was 93.21%, with only 0.46% incorrectly assigned and 6.33% missing or unassigned. This high proportion of alleles correctly assigned a breed origin resulted in a high core-based mean accuracy of 0.99 and a very high consensus-based (most frequently observed assignment across all the scenarios) mean accuracy of 1.00. The algorithm's assignment yield and accuracy were affected by the choice of threshold levels for the best match of assignments. The threshold level had the opposite effect on assignment yield and assignment accuracy. A less stringent threshold generated higher assignment yields and lower assignment accuracy.

Conclusions: We developed an algorithm that accurately assigns a breed origin to alleles of crossbred animals designed to represent breeding programs in the LMICs. The developed algorithm is straightforward in its application and does not require prior knowledge of pedigree, which makes it more relevant and applicable in LMICs breeding programs.

Fig. 1

Schematic representation of the simulated breeding program. A founder population on the top of the figure is split into exotic and local breeds

Fig. 2

Illustration of a core and offset. Phasing was performed in two modes: either using the whole length of a core or by moving it forward 50% of the core length (offset) to define the beginning of a given core

Fig. 3

Phasing and assignment algorithm overview. a Phasing was performed using AlphaPhase, b Hypothetical haplotype libraries based on a core length of ten SNPs. To assign origin to the haplotype of an individual (bottom genotype sequence), the algorithm searches for the best match in each position in the exotic (top left genotype sequence) and local (top right genotype sequence) haplotypes. In this case, the individual’s haplotype should be assigned as a local haplotype because the local haplotype library contains the haplotype with the fewest number of errors, i.e., mismatches (red)

Fig. 4

Plot of principal component analysis of SNP genotypes (PC1 vs. PC2 and PC1 vs. PC3). a Genetic data structure of the founders and the first crossbred cows, b of all animals across generations

Fig. 5

Effect of core length on assignment yield. a Phasing yield was very high for all core lengths but more concise for longer core lengths (core length of 220–280 SNPs), b Assignment yield was optimal for a core length of 200 SNPs

Fig. 6

Percentage of allele assignment yield and accuracy of assignment. a Yield using the consensus-based allele assignment algorithm as a function of threshold level, b accuracy using the consensus-based allele assignment algorithm as a function of threshold level

Fig. 7

Percentage of allele assignment yield and accuracy of assignment. a Yield using the consensus-based allele assignment algorithm as a function of crossbreeding (admixture) level, b accuracy using the consensus-based allele assignment algorithm as a function of crossbreeding (admixture) level