Modelling of biochemical reaction networks is an important element of modern systems biology. It aims to describe and explain how basicbiochemical mechanisms give rise  to more complex biological phenomena. Theoretical models, however, involve numerous parameters, which are usually unknown or have only vaguely constrained values. Therefore reliable utilisation of a model demands tools to extract relevant information even in the case when parameters are not entirely determined.
Some sensitivity analysis methods, for instance, enable to quantify the impact of parameters on the model responses, even if parametersare not precisely determined but constrained to a broad range of values. The existing methods, however, are not well adapted to a multi-parameter scenario, typical for models of biochemical dynamics.We have developed a sensitivity analysis method, based on information theoretic measure, to understand the joint impact of parameters on model dynamics.  The method  hierarchically determines  the  subgroups  of  most significant parameters, assigning the sensitivity index to the subgroup. To quantify the sensitivity of the model, we randomly search the parameter space, seeking the parameters with largest weights based on sensitivity indexing.  Our approach improves available methods by finding groups of crucial parameters that significantly influence overall model dynamics. Moreover the method enables a modeller to specify feature of interest and determine which group of parameters is relevant for the specified property. The advantage of the proposed methodology is its relative simplicity and easy applicability for multi-parameters models.

Sensitivity analysis; multi-parameter analysis; mutual information; entropy