QTL detection

Training sets obtained by U, SU, CD, S or R sampling of the genotype panel were used to identify QTL that became part of the prediction model. QTL were identified by a genome-wide association mapping scan (GWAS), following Equation 5.

In Equation 5, y_i stands for the phenotype of genotype i, μ is the intercept, x_ik is a vector that represents information of genotype i at marker k (0 and 2 for homozygous and 1 for heterozygous), and α_k is the additive QTL effect (fixed) for marker k. G_i represents a polygenic effect for genotype i, and e_i is the nongenetic residual (ei∼N(0,σe2)). The distribution of G_i is ∼N(0,Aσg2). A is the additive genetic relationship matrix calculated from the molecular marker information as in Rincent et al. (2014a). In this method, a specific A is calculated for each linkage group by excluding the markers on that particular linkage group. A significance threshold equivalent to a genome-wide significance level of 0.01 was used for the four data sets, following the Li and Ji (2005) multiple-testing correction. We performed the GWAS as implemented in GenStat 17th edition (VSN-International 2015).