Publications des agents du Cirad


Pedigree reconstruction for cosexual species using simulated annealing: case study of oil palm (Elaeis guineensis) : P-124

Cros D., Fernandez J., Bouvet J.M., Sanchez L.. 2012. In : ICQG 4 Conference. Programme and book of abstracts of the 4th International Conference of Quantitative Genetics:Understanding Variation in Complex Traits, Edinburgh, UK, 17-22 june 2012. s.l. : s.n., p. 123-123. International Conference of Quantitative Genetics. 4, 2012-06-17/2012-06-22, Edinburgh (Royaume-Uni).

The approach of Fernandez and Toro (2006) to reconstruct pedigrees from molecular data of contemporaneous individuals using a simulated annealing algorithm was extended to hermaphroditic and monoecious species. New features also include the possibility of selfings, accounting for a predefined coancestry matrix between founders and specifying different number of individuals per generation. The new method was validated using 16 individuals from the last generation of the Yangambi breeding population of oil palm. Their pedigree was known for 5 generations and they were genotyped with 166 SSR. In the study, the number of used SSR varied from 6 to 166 and the percentage of unknown parentages from 20% to 100%. The Pearson correlation between the pedigree-based coancestries calculated on the true and on the reconstructed genealogies ranged from 0.74 to 0.99. The RMSE ranged from 0.02 to 0.12. When pedigree was assumed completely unknown, reliable reconstruction required at least 38 SSR. Using 100 SSR or more, the Pearson correlation was very high (0.98) and the RMSE very low (0.06). The new method was also applied to 104 individuals from the last generation of a key breeding population (Deli) originated from 4 oil palms. The individuals were genotyped with 160 SSR. Records of their pedigree only existed for the recent past. Results of pedigree reconstruction detected a family coming from old selfings looking as outliers, with pedigree-based coancestries much higher than molecular coancestries, indicating old selfings were erroneous. After correcting the recorded pedigree, pedigree-based coancestries calculated on the reconstructed genealogy and molecular coancestries were highly correlated (> 0.9) when using 80 markers or more. In conclusion, this method gave likely pedigrees with satisfactory reliability for cosexual species, using a realistic number of polymorphic markers. Also, it seems very helpful to correct historical pedigrees. The methodology has been implemented in the software MOL_COANC_v2. (Texte integral)

Documents associés

Communication de congrès

Agents Cirad, auteurs de cette publication :