Summary of the third strategy. This approach was divided into two major steps. First, non-recombinant animals in the GGA5 AF QTL confidence interval were selected. Discriminant analysis (DA) or Logistic Regression (RL) were then carried out with the 5 genes to distinguish a specific transcriptome profile between the Q or q allele at the causative mutation. Second, previously established DA or RL were used to predict a most probable Q or q allele at the causative mutation for each recombinant animal in the GGA5 AF QTL confidence interval. New markers were then developed and genotyped (not shown on the figure) to define each recombination breakpoint. Comparison between recombinant offspring then enabled us to define the most valuable location of the causative mutation (line in red).