James R. Faeder
Department of Computational Biology
University of Pittsburgh School of Medicine
Title: An exact hybrid particle/population approach for modeling
biochemical systems
Justin S. Hogg, Leonard A. Harris, Lori J. Stover, Niketh S. Nair and
James R. Faeder
Abstract: Practical simulation of realistic biochemical networks is
often challenging due to a combination of high state-space complexity
and large molecular populations that is characteristic of many
biological systems. Recent particle-based simulation approaches have
been developed that can efficiently handle systems with large state
spaces. However, memory and computation costs, which increase linearly
with the number of particles, remain an impediment to the practical
application of these methods. Here, we present an exact hybrid
particle/population modeling approach that can significantly reduce
memory costs by treating a subset of molecular species as population
variables rather than as particles. The approach is cast within the
framework of rule-based modeling. We demonstrate how a model cast
within a rule-based modeling language, such as the BioNetGen language
(BNGL), can be transformed into a form suitable for hybrid simulation
via a population-adapted particle-based method. The transformation
method is implemented within the open-source rule-based modeling
platform BioNetGen and the resulting hybrid model can be simulated using
the particle-based simulation engine NFsim. Benchmark tests show that
significant memory savings can be achieved using the new approach and a
cost analysis provides a practical measure of its utility. We also
discuss the intriguing possibility of applying accelerated-stochastic
methods, such as $\tau$\/~leaping, to the population sub-network of the
hybrid model to improve computational efficiency.