Exemplar ProjectsVPH NoE Exemplar Projects and the VPH ToolKit - 7. CIGENE: Integrating genetic theory and genomic data with multiscale models in a population context
7. CIGENE: Integrating genetic theory and genomic data with multiscale models in a population context
Coordinator: Stig Omholt, Norwegian University of Life Sciences
Partners: University of Auckland, King’s College London
One of the central goals of the VPH is to link genotype to phenotype through multiscale models of physiological structure and function, at the levels of cells, tissues, organs and organ systems. The vertical integration is a major challenge, from both ends of the spectrum. Given the serious challenges for the effectiveness of genome-wide association studies (GWAS) for drug development, and since mathematical models represent a rigorous compendium of the knowledge about biological processes at many scales, the time is ripe to begin using them as a platform for multi-parametric in silico studies aimed at understanding the functional implications of genome-level perturbations, i.e., to work towards modelling-based genome-to-phenotype maps.
The group of Stig Omholt in Norway has been moving in this direction for the last several years. They call it "causally cohesive genotype-phenotype (cGP) models" (Rajasingh et al. 2008), and they will now, in the context of an NoE Exemplar Project, team up with Universities of Auckland and Oxford to connect genetic information with multiscale and multiphysics models in a population context, with specific application to dynamic multiscale modelling of the heart.
In a well-validated multiscale model describing one or more phenotypic features and capable of accounting for observed variation in a population, the effects of genetic and environmental variability on the phenotypic features are manifested in the parametric variations of the model. To the extent that model parameters represent phenotypes, the phenotypic variation that emerges from a multiscale model when its parameters are varied is an in silico manifestation of how lower-level phenotypic variation causally contributes to higher-level phenotypic variation; a deep enough sensitivity analysis of the model will provide valuable insight into how this lower-level variation percolates up to the higher level. By producing large populations of "virtual individuals" through randomization of selected parameter values, the models thus become a testbed for the possible roles of low-level parameters (e.g., cell- or molecular-level phenotypes affected directly by genetic variation) in determining high-level physiological behavior. Since the in silico models are not subject to the same limitations facing experimental models such as mice, the two could be combined in a powerful new approach. The modeling could thus serve as a companion and guide to directed GWA studies, in which the contributions of individual genetic variants all too often seem to contribute little to macro phenotypic variation. This EP will explore this new territory.