Skip to Main content Skip to Navigation
Master Thesis

Évaluation de la précision des modèles de prédiction de la qualité boulangère du blé tendre (Triticum aestivum L.)

Abstract : Bread-making qualities are crucial to register a new variety in France. It is a complex trait that can be measured either by bread-making tests or by indirect physical measurements of the dough (alveograph). Genomic selection could increase genetic gain per time or cost unit for this trait. The objective is to define the size of the training population, the number of molecular markers and the optimal statistical model to build the prediction equation. In total, 716 lines and 10 124 SNPs from Limagrain’s breeding program, 370 lines and 172 074 SNPs from INRA-AgriObtentions are analysed. The architecture of 14 traits is determined using association analyses. Different prediction models adapted to traits controlled by many genes with small effect (GBLUP) or a few genes with large effect ( Bayes Cπ ) are tested. Although random samples of 5000 molecular markers allow optimal predictions for GBLUP, 1000 independent markers associated with the trait give significantly higher accuracy for all traits, up to three times higher for bread-making ability at Limagrain. The Bayes Cπ model gives similar results than GBLUP. The size of the training population needs to be increased to stabilize predictions. These models necessitate validation on independent data sets.
Document type :
Master Thesis
Complete list of metadatas

Cited literature [38 references]  Display  Hide  Download

https://dumas.ccsd.cnrs.fr/dumas-01404001
Contributor : Agrocampus Ouest <>
Submitted on : Monday, November 28, 2016 - 11:27:35 AM
Last modification on : Monday, July 6, 2020 - 3:38:21 PM
Long-term archiving on: : Monday, March 20, 2017 - 6:16:05 PM

Identifiers

  • HAL Id : dumas-01404001, version 1

Collections

Citation

Pierre Colin. Évaluation de la précision des modèles de prédiction de la qualité boulangère du blé tendre (Triticum aestivum L.). Sciences du Vivant [q-bio]. 2016. ⟨dumas-01404001⟩

Share

Metrics

Record views

135

Files downloads

885