Introduction

UQSA is an R-package for modelling and calibration of biochemical (and other) reaction networks.

Example of a Chemical Reaction Network

An important part of this is the parameter estimation (calibration), and to make sure that the uncertainty in the parameter estimates and predictions are accounted for. UQSA allow you to perform Baysian uncertainty quantification while calibrating your model. With UQSA you can also do a global sensitivity analysis to guide further experiments. (To get to know UQSA quickly try out our smallest and easiest example: AKAR4.)

UQSA workflow

UQSA is specially constructed for systems biology models as we use the SBtab format to define the model and calibration data (see intro to SBtab). But SBtab can be translated to and from other formats like SBML.

Models (reaction networks) written in SBtab can automatically be translated to ODE or stochastic models within UQSA. These mathematical models can next be simulated and calibrated. As part of the calibration uncertainty quantification is performed. We use likelihood based Bayesian approaches for the ODE models and Approximate Bayesian Comutation (ABC) for the stochastic models to do this. Finally a global sensitivity analysis can be don on a independent or non-independent parameter space.

You can read more about UQSA in our paper and there are several examples to try out within this documentation.