Parallel Computing Platform for the Agent-Based Modeling of Multicellular Biological Systems

Abstract

Abstract. Agent-based simulation of large multicellular biological systems has become a viable option owing to affordable parallel computers, such as Beowulf-style clusters. We describe a scalable modular software platform that (i) provides for an easy integration of different solvers computing internal dynamics of the cells, and (ii) dynamically adapts to the changing loads on the cluster nodes using a load balancing algorithm. Simulations of systems of about 100,000 bacterial cells have been shown to be feasible using the platform.

Colorectal tumour simulation using agent based modelling and high performance computing

“…Toh Da-Jun et al [28] present an ABMS platform for multicellular biological systems, which, similarly to our proposal, incorporates a load balancing mechanism including migration of cells. However, apart for being a specialised platform (FLAME is far more generalist), this load balancing mechanism is centralised and, in consequence, not scalable to nowadays systems.…”

Design of a web-based application of the coupled multi-agent system model and environmental model for watershed management analysis using Hadoop

“…For example, Da-Jun et al (2005) design the parallel computing platform for the specific Beowulf-style clusters, and Tang et al (2011) use supercomputer grids, TeraGrid, for the parallel agentbased modeling. However, in reality the time and monetary cost to access these computing resources can limit their applications.…”

Computer experiments for epidemiological studies : theory, methodology and models

“…We have studied the conventional Differential Equation (DE) based mathematical models [76][77][78][79], and also the more recent Agent-Based Models (ABMs) [80][81][82][83][84]. Given the modeling requirements imposed by the domain, we further studied the existing modeling strategies that address the multi-scale issues [85,86], the explicit modeling of the environment and space [87,88], scalability of models [89], and the adoption of distributed and parallel processing [90][91][92][93]. Due to the length constraint, in this Chapter we only present our reviews of literature that are directly relevant to the discussions on computer experiments.…”