Using heterogeneous data sources in a systems biology approach to modeling the Sonic Hedgehog signaling pathway

Abstract

Experimental data from biological pathways come in many forms: qualitative or quantitative, static or dynamic. By combining a variety of these heterogeneous sources of data, we construct a mathematical model of a critical regulatory network in vertebrate development, the Sonic Hedgehog signaling pathway. The structure of our model is first constrained by several well-established pathway interactions. On top of this, we develop a hierarchical genetic algorithm that is capable of integrating different types of experimental data collected on the pathway's function, including qualitative as well as static and dynamic quantitative data, in order to estimate model parameters. The result is a dynamical model that fits the observed data and is robust to perturbations in its parameters. Since it is based on a canonical power-law representation of biochemical pathways whose parameters can be directly translated into physical interactions between network components, our model provides insight into the nature and strength of pathway interactions and suggests directions for future research.