Séminaire INRIabcd

Multi-resolution and multi-level models in biology: from methods to applications

par M. Gunnar Cedersun (Linköping University)

Europe/Paris
432 (4ème étage) (INRIA- antenne Lyon- La Doua, Bâtiment CEI-2 (tram: T1-IUT FEYSSINE))

432 (4ème étage)

INRIA- antenne Lyon- La Doua, Bâtiment CEI-2 (tram: T1-IUT FEYSSINE)

INRIA- antenne Lyon- La Doua, Bâtiment CEI-2 (tram: T1-IUT FEYSSINE)
Description
Biological and medical sciences are today facing an irreversible shift: one from reasoning and statistics used to detect differences between a few data points, to computer-aided reasoning and mechanistic modelling, based on large-scale datasets. This shift requires a rapid development of both new concepts and new technologies, both within the biological and the mathematical/computer sciences. In this presentation, I will outline some of my contributions to these developments, which involves: i) How to borrow data and knowledge from neighboring cells using nonlinear mixed-effects modelling, ii) How to uniquely identify parameters and predictions despite parameter unidentifiability, iii) How to scale mechanistic models to the omics level, also encompassing multiple cell-types and levels (cell, organ, whole-body). The result of this is a multi-resolution model, which combines all prior knowledge and data into a single descriptive, and which has different known degrees of "resolution" - i.e. detail and predication accuracy - in its different parts. I will illustrate how we have used this to draw useful conclusions that could not have been drawn without the modelling, both regarding basic research, and regarding clinical and pharmaceutical end-usage.