Computational Modeling of Cell Signaling Dynamics: Hypothesis Management and Parameter Estimation Methods Applied to the Akt Pathway

Abstract

The dynamics of a cell signaling pathway can be used to elucidate the mechanisms of the pathway. We simulated Akt activation dynamics using ODE (ordinary differential equation) models based on previous time-series datasets. Simulations of the canonical Akt activation pathway were unable to reconcile the levels of phosphatidylinositol(3,4,5)-trisphosphate (PIP3), which peaked at 2min, and the dynamics of Akt activation, which peaked at 30min. By systematically simulating non-canonical mechanisms for resolving this time difference, we identified five potential hypotheses and constructed ensembles of ODE models to evaluate each hypothesis. Simulations motivated additional experiments, and computational analyses of the results were consistent with two novel mechanisms of augmented Akt phosphorylation. Also part of this thesis is a computational method for estimating the rate parameters in ODE models. Our novel method is called SPEDRE (Systematic Parameter Estimation of Data-Rich Experiments).