Comparison of sever al approaches for predicting maize hybrid performance using marker data
1997
Denis, J.B. | Charcosset, Alain, A. | Bonnisseau, B. | Touchebeuf, O. | Burstin, Judith | Dubreuil, Pierre | Barrière, Yves, Y. | Gallais, A. | Unité de biométrie et intelligence artificielle de Jouy | Institut National de la Recherche Agronomique | Génétique Végétale | Institut National de la Recherche Agronomique -Université Paris-Sud - Paris -Institut National Agronomique Paris-Grignon -Centre National de la Recherche Scientifique | Unité de recherche Génétique et Amélioration des Plantes Fourragères
International audience
Show more [+] Less [-]English. Several statistical ways of predicting the performance of maize hybrids are compared in the framework of silage performances available from a 21 x 21 complete half diallel mating design. Each inbred was characterized by different kind of loci (142 in total). Four types of models were considered: (i) using Roger's multilocus distance a s covariate in the expectation of SCA (specific combining ability), (ii) using different subsets of PCA scores obtained from marker data in factorial regression equations and (iii) calculating BLUP for SCA introducing a covariance matrix for a random SCA term, based on Roger's multilocus distance. The additive model (no SCA) was included as control of the efficiency of the marker information. Comparisons were made by means of a cross-validation technique: the 210 hybrids were split in one set for prediction and the complementary set for validation according to different schemes. For each of them, 100 different random samples were simulated. Each time, the predictions were calculated and stored together with the observed supposed "true" values. Improvements were always obtained when introducing the marker information, especially when hybrids between related parents are present.
Show more [+] Less [-]Bibliographic information
This bibliographic record has been provided by National Institute for Agricultural Research