Journal Club
Ramsey, S.; Orrell, D. and Bolouri, H. (2005) Dizzy: stochastic simulation of large-scale genetic regulatory networks. Journal of Bioinformatics and Computational Biology, 3, 415-436. Abstract
Most mathematical models of biological systems are deterministic systems of differential equations, in which the variables of the system (e.g. the metabolites involved) are concentrations (a continuum variable). These models are adequate insofar the microscopic deviations from the average macroscopic values can be discarded. But there are situations in which microscopic fluctuations are important in biological systems:
when there is a small number of molecules involved. For example, for a mitochondrion with typical dimensions of 4×10-13 cm3, a concentration of 42 nM is equivalent to consider 1 molecule on average per mitochondrion. Another example, the number of copies of a single gene in a nucleus is 2 for a diploid organism.
When the system under study has small dimensions. Biological compartments can often be subdivided into small microcompartments with distinctive physico-chemical properties. The number of molecules of an abundant cellular species in such microcompartment can be very small.
The system operates near an instability point of a deterministic model. In this case the small microscopic fluctuations can be amplified producing macroscopic effects (the well-known "butterfly effect").
In all these cases, stochastic microscopic fluctuations are the essence of the phenomenon and therefore deterministic models are not adequate to model such systems. Stochastic models must be used. In these models, variables are the number of molecules instead of the macroscopic average concentrations, and the rate of processes is based on the molecular collisions between the individual molecules and not on the mass action principle. The disadvantage of stochastic models is the high computational cost, which even for today fast computers can be prohibitive.
Dizzy, is a novel stochastic software programme focused on the simulation of large-scale regulatory networks. Besides carrying a user-friendly interface which allow for the non-expert to carry its own stochastic simulations, Dizzy contains two features that are essential to successfully model complex biological systems:
Dizzy allows approximating complex reaction kinetics as a simplified probability function that can be implemented in a stochastic model. A priori any complex reaction before could be simulated at a stochastic level has to be decomposed into its elemental reactions steps. For the vast majority of biological processes this decomposition is not possible due to the lack of knowledge of the constituent elemental steps, and even if these were known the computational cost of stochastic simulating them would be prohibitive.
Dizzy allows lumping multi-step reactions, such as elongation, or the multiple reactions of a metabolic pathway, into a probability function with a phenomenological time delay that approximates well the whole process at a stochastic level.
Dizzy comes with an implementation of four deterministic algorithms and four stochastic algorithms. These features together with an integrated graphic interface and portability, which allows exchanging models with other simulation software, makes Dizzy an excellent tool for those interested in stochastic modeling of complex reaction networks.
Dizzy is free and open-source software, distributed under the GNU Lesser General Public License (LGPL). Dizzy software is available for download. 