Abstract

Abstract: The adoption of a particular fate by cellular mechanisms is usually thought of as being deterministic. However, several findings have highlighted the relevance of noise and hence of stochasticity. Many simulator tools have been developed to capture, reproduce and capitalize on this property. They use formal models to produce individual markovian trajectories, usually calculated by the SSA or one of its variants. Statistical inference on the simulation output may require a fairly large number of trajectories in order to generate accurate results. On the other side, with the aim to tackle complexity of nature and to faithfully reproduce its (intrinsically parallel) behaviours, some algorithms have been designed to deal with chemical motion through cellular compartments and with asynchronous firing of independent chemical reactions. Any of such attempt led scientists to simulate in parallel, but under heavy assumptions.

Having in mind that biological systems consist of networks of interacting biological components, we define an optimised stochastic simulation algorithm that takes advantage of the partitioning of models. The resulting algorithm enjoys the following properties: it is free from any restrictive assumptions (apart from those directly dealing with SSA ); it is proactive to self-adjusting in run-time when the system changes; it is scalable and correct. Then, we show how to deal with a basic matter of parallel stochastic simulation, namely the maintenance of correct simulation time amongst the simulated parts. Finally, we give some insights on message-passing policy of communications among computing-cores and parallel random seed generations: both of which are fundamentals aspects in such a framework.

Biography: Tommaso Mazza was born in Catanzaro, Italy, on 8 June 1979. He studied Computer Science Engineering at the University of Calabria and received a PhD in Computer Science and Biomedical Engineering in November 2007 from the ‘Magna Graecia’ University of Catanzaro. In 2004, he worked on the Cofin 2003 Project titled “Tumori ereditari della mammella: studi genetici ed analisi del proteoma” (“Hereditary tumours: genetic studies and proteomic analysis”). In 2005, he worked on the design of a common interface to translate mass spectrometry raw data (MALDI-TOF and LC) into mzData. In March 2006, he visited CoSBi, where he dynamically collaborated in the design and implementation of Cyto-Sim, a stochastic simulator of biochemical processes. Moreover, he worked on the definition of a common interface for unmarshalling SBML models into the Cyto-Sim syntax and vice-versa. In summer 2006, Tommaso visited Microsoft Research Cambridge where he investigated the possibility of parallelizing and distributing simulation algorithms. In 2007, he joined the Bioinformatics Italian Society (B.IT.S.). Tommaso joined CoSBi in January 2008.