There is a growing set of tools available to researchers for simulating intracellular biochemical networks, but a new package from the Molecular Sciences Institute claims to have raised the bar a few notches.
The software, Moleculizer 1.0, generates reaction networks during the simulation process — a capability that other packages do not have, according to Larry Lok and Roger Brent of MSI, who published a paper on the stochastic simulation method in the January 2005 issue of Nature Biotechnology.
Other packages can either generate reaction networks or run the simulation, but Moleculizer can do both, and therefore “generates molecular complexes on the fly,” as the simulation is running, Brent said.
Moreover, the method addresses a major drawback of computationally predicting cellular networks — the so-called combinatorial explosion that occurs when all potential reactions and complexes are generated at once — by generating only those complexes that are needed as the simulation progresses. This generates a much smaller reaction network than other methods, which makes it easier for computers — and researchers — to process the output.
As an example, Brent noted that in the primary pathway the MSI research team is studying — the yeast mating pheromone signal transduction, or alpha, pathway — one cascade alone is comprised of 480 known molecules that could interact in more than 20,000 possible combinations — an “absurd” number for a biologist to wade through, Brent said. If the entire pathway were to be considered, he said, that would generate “an inconceivable number of possible [molecular] species.”
Moleculizer’s reaction network generator begins with a single “triggering” molecule, along with its specifications for binding, unbinding, and other reactions. The program builds out the network in an iterative manner, creating all possible reactions for a single molecule at a time, which may or may not be used by the program’s simulator as it progresses. Only those molecular species required by the simulator continue in the iterative process as the network is created.
Brent and Lok said that the networks generated by Moleculizer could be used as input for other simulation software packages, and that the team plans to enable conversion of Moleculizer-generated networks to SBML (Systems Biology Markup Language), Level 3, in order to facilitate this exchange.
Lok said that the MSI team is currently using the software as an “heuristic tool” to aid MSI’s wet lab activities, which currently account for around 80 percent of the institute’s effort to model the alpha pathway.
The software is available under the GPL at http://www.molsci.org/~lok/moleculizer/downloads/index.html. Lok warned, however, that Moleculizer “is not for the faint of heart.” Rather, he said, it’s designed with the requirements of “hardcore simulation geeks” in mind — an extremely small community, by Lok and Brent’s estimation.
Nevertheless, future development plans call for a web interface through which researchers will be able to run simulations online, as well as an improved network-generation method that accounts for the geometry of molecules.
The MSI developers are integrating Moleculizer with a number of collaborative biosimulation packages, including E-Cell, MONOD (Modeler’s Notebook and Datastore), ChemCell, and the Systems Biology Workbench.