Skip to Main content Skip to Navigation

Réduction de modèle, estimation des paramètres pour une population de génotypes et analyse du contrôle génétique : Cas du métabolisme des sucres dans la pêche

Abstract : Genotype-phenotype models are seen as future tools for designing new varieties, as they can help to test the performance of genotypes under different climates and cultural practices. However, progress is still needed to include complex genetic control in process-basedmodels and to achieve the integration of information (metabolism, enzyme control, quantitative genetics) from gene to plant. A kinetic model of sugar metabolism was developed by Desnoues et al. (2018) to simulate the concentrations of different sugars during the development of the peach fruit. The objectives of my thesis work are (a) to estimate the inter-genotype variability of themodel parameter values and (b) to study the architecture of genetic control of genotype-dependent parameters. To achieve these objectives, it is necessary to estimate the values of the influential parameters of the model for all 106 genotypes for which few data are available. The number of parameters and the non-linearity of the modelmake the calibration of the model inaccurate (in particular because of the large number of parameters compared to the number of available data and the correlations between the parameters) and costly in terms of calculation time. Therefore, we have developed a reduction strategy for the initial model to reduce the number of parameters and to simplify the model structure while maintaining the network structure and the identity of the variables to facilitate their biological interpretation. The estimation of the 9 parameters of the reduced model obtained was carried out at population scale using a nonlinear mixedmodel(Baey et al. 2018) approach, which makes it possible to simultaneously estimate the values of the parameters of all genotypes. This strategy was compared to more conventional methods which carry out the estimation of the values of the parameters individually, genotype by genotype. We showed that the reliability of the estimates is greatly improved and those correlations between the estimated parameters are reduced. The last step consisted in estimating the links between the genome markers and the genotype-dependent parameters (QTL analyses). We compared the results of a so-called independent analyse (estimation of parameters then detection of the genomic regions involved) with a joint analysis method (Onogi 2020). This PhD work constitutes a crucial step towards the development of a tool that takes into account a complex genetic control of traits to design ideotypes.
Document type :
Complete list of metadata
Contributor : Abes Star :  Contact
Submitted on : Thursday, July 8, 2021 - 9:31:10 AM
Last modification on : Friday, July 9, 2021 - 4:33:19 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03281201, version 1



Hussein Kanso. Réduction de modèle, estimation des paramètres pour une population de génotypes et analyse du contrôle génétique : Cas du métabolisme des sucres dans la pêche. Statistiques [math.ST]. Université d'Avignon, 2021. Français. ⟨NNT : 2021AVIG0727⟩. ⟨tel-03281201⟩



Record views


Files downloads