Skip to Main content Skip to Navigation
Master Thesis

Évaluation d'une stratégie de sélection génomique dans 3 dispositifs expérimentaux chez la tomate

Abstract : Selection to improve traits of economical interest in crop was achieved through MAS or GWA which capture with difficulty loci of minor effect, and through steps phenotyping often long and costly. Genomic selection (GS) is a new tool for selecting elite plants in a breeding program by predicting the performance of traits of interest in application of statistical model; GS has the potential to catch the effect of all markers underlying the genetic architecture of the studied traits. We aimed at evaluating the use of GS into tomato to evaluate the prediction accuracy for 25 traits related to the fruit quality for three different type of population (GWA, MAGIC and RIL panel). We conducted a cross validation approach while estimating the relative weight of parameters (training and testing populations sizes, GxE interactions, predictive statistical models, markers density, heritability of the trait, type of population) onto the prediction accuracy. Our results demonstrated that the GS approach seems powerful at predicting phenotypes values with accuracies ranging from 0.32 to 0.93 in the panel when training set is optimizing, for example. We observe that (1) if the population is the bigger and if it contains individuals the most close genetically, and that were phenotyped in similar environments compared to PV, prediction accuracy is better (2) optimizing the training set with the larger genetic diversity, the better the predictions are (+9% on average) (3) the larger the heritability of the trait value is, the better the predictions are (3) the statistical prediction models perform very similarly to estimate accuracies (4) statistical models lead to very similar predictions (5) the optimal markers density is depending of the population LD. To conclude, although SG still a practice which is still in its infancy in plants, applying GS in tomato seems very promising and could be used in tomato breeding programs.
Document type :
Master Thesis
Complete list of metadata

https://dumas.ccsd.cnrs.fr/dumas-01219994
Contributor : Agrocampus Ouest <>
Submitted on : Friday, October 23, 2015 - 4:29:28 PM
Last modification on : Monday, July 6, 2020 - 3:38:21 PM
Long-term archiving on: : Friday, April 28, 2017 - 6:41:15 AM

Identifiers

  • HAL Id : dumas-01219994, version 1

Collections

Citation

Coralie Picard. Évaluation d'une stratégie de sélection génomique dans 3 dispositifs expérimentaux chez la tomate. Sciences agricoles. 2015. ⟨dumas-01219994⟩

Share

Metrics

Record views

298

Files downloads

2361