Should we use the single nucleotide polymorphism linked to in genomic evaluation of French trotter?

An A/C mutation responsible for the ability to pace in horses was recently discovered in the DMRT3 gene. It has also been proven that allele C has a negative effect on trotters' performances. However, in French trotters (FT), the frequency of allele A is only 77% due to an unexpected positive effect of allele C in late-career FT performances. Here we set out to ascertain whether the genotype at SNP BIEC2-620109 (linked to DMRT3) should be used to compute EBV for FT. We used the genotypes of 630 horses, with 41,711 SNP retained. The pedigree comprised 5,699 horses. Qualification status (trotters need to complete a 2,000-m race within a limited time to begin their career) and earnings at different ages were precorrected for fixed effects and evaluated with a multitrait model. Estimated breeding values were computed with and without the genotype at SNP BIEC2-620109 as a fixed effect in the model. The analyses were performed using pedigree only via BLUP and using the genotypes via genomic BLUP (GBLUP). The geno-type at SNP BIEC2-620109 was removed from the file of genotypes when already taken into account as a fixed effect. Alternatively, 3 groups of 100 candidates were used for validation. Validations were also performed on 50 random-clustered groups of 126 candidates and compared against the results of the 3 disjoint sets. For performances on which DMRT3 has a minor effect, the coefficients of correlation were not improved when the genotype at SNP BIEC2-620109 was a fixed effect in the model (earnings at 3 and 4 yr). However, for traits proven strongly related to DMRT3, the accuracy of evaluation was improved, increasing +0.17 for earnings at 2 yr, +0.04 for earnings at 5 yr and older, and +0.09 for qualification status (with the GBLUP method). For all traits, the bias was reduced when the SNP linked to DMRT3 was a fixed effect in the model. This work finds a clear rationale for using the genotype at DMRT3 for this multitrait evaluation. Genomic selection seemed to achieve better results than classic selection.