Cross-by-QTL interaction analysis in DABC and PVGBC.

The statistical validation of the cross-by-QTL interaction was performed using the fit-multiple QTL analysis (for details see Materials and Methods). A full model comprised all identified QTLs (from forward selection with reverse elimination, as in Table S1) and CROSS x QTL interaction terms for the QTLs that displayed parent-of-origin effect (i.e. QTLs that could be mapped only in one of the crosses). In the next stage the effect of each QTL or CROSS x QTL interaction was subtracted from the full model and the contribution of the subtracted term to the full model was evaluated and expressed in p-values. The model used was Phenotype ∼ QTL1a + QTL1a*CROSS + QTL1b + QTL3 + QTL3*CROSS + QTL4a + QTL4a*CROSS + QTL4b + QTL5b + QTL5b*CROSS + QTL6 + QTL6*CROSS + QTL7a + QTL7b + QTL7b*CROSS + QTL10a + QTL10a*CROSS + QTL10b + QTL10c + QTL10c*CROSS + QTL10d + QTL11 + QTL12 + QTL14 + QTL14*CROSS + QTL18 + QTL18*CROSS + QTL19 + ε. Presented in the table are only p-values for the parent-of-origin dependent QTLs x CROSS terms. The p-values from the fit-multiple QTL analysis in the PVGBC malesa (N = 232), PVGBC femalesb (N = 239), DABC malesc (N = 208) and DABC femalesd (N = 213) are given. The population with the most significant p-values is shown for each QTL and specified in superscript, with the additional populations that show significant CROSS x QTL interaction indicated in superscript parenthesis. n/a, analysis could not be applied due to lack of evidence of a QTL.