lire cette page en :
Français

Modelling the effects of the genetic variability: which models for which uses?

François Tardieu
LEPSE, INRA
le 18/03/2011 à 15:30

Résumé

The plant phenotype is determined by a programme which drives organ appearance and development, and by biophysical laws which drive transports and expansion. Both are controlled by gene networks and environmental conditions. It has been proposed that a model can cross levels of integrations from gene to phenotype by taking advantage of the fact that networks at different levels of biological integration share similar structural and dynamic properties. However, this would result in a deluge of parameters whose genetic variability cannot be analysed. I propose that different models can be used at different scales, which communicate via emergent properties established at each level. (i) Several examples will be presented of models at whole plant level which cannot be interpreted simply in terms of mechanisms at cellular or gene level. Conversely, they can help to discover these mechanisms via a genetic analysis of "hidden variables" derived from models, such as absolute rates or conductances and their responses to environmental conditions. These variables are usually more heritable than raw phenotypic traits and therefore more amenable to a genetic analysis. Mechanistic models are probably not appropriate for identifying these variables, first because they are based on physiological hypotheses which may or may not be compatible with the genes whose polymorphisms will eventually be identified, second because mechanistic models use too many parameters for a genetic analysis. (ii) A second set of models evaluate the consequences, at the whole plant level, of the effect of a causal polymorphism which affects specific functions of the plant. In this case, mechanistic models are essential, over-parametrisation does not necessarily cause major problems provided that most parameters are kept common among the tested genotypes.