Section: Software

BioRica: Multi-scale Stochastic Modeling

Participants : David James Sherman, Macha Nikolski [ correspondant ] , Hayssam Soueidan, Nicolás Loira, Grégoire Sutre.

Multi-scale modeling provides one avenue to better integrate continuous and event-based modules into a single scheme. The word multi-scale itself can be interpreted both at the level of building the model, and at the level of model simulation. At the modeling level, it involves building modular and hierarchical models. An attractive feature of such modeling is that it provides a systematic means to balance the need for greater biological detail against the need for simplicity. At the execution level, it implies the co-existence of phenomena operating at different time scales in an integrated fashion. This is a very lively research topic by itself, and has promising applications to biology, such as for example in [46] .

We are developing BioRica , a high-level modeling framework integrating discrete and continuous multi-scale dynamics within the same semantics field. BioRica has been adopted as an INRIA Technology Development Action (ADT).

The co-existence of continuous and discrete dynamics is assured by a pre-computation of the continuous parts of the model. Once computed, these parts of the model act as components that can be queried for the function value, but also modified, therefore accounting for any trajectory modification induced by discrete parts of the model. To achieve this we extensively rely on methods for solving and simulation of continuous systems by numerical algorithms. As for the discrete part of the model, its role is that of a controller.

As a means to counteract the over-genericity of re-usable modular models and their underlying simulation complexity, BioRica will provide an abstraction module, whose aim is to preserve only the pertinent information for a given task. The soundness of this approach is ensured by a formal study of the operational semantics of BioRica models[22] that adopts the theoretical framework of abstract interpretation [34] .

The current stage of development extends the AltaRica modeling language to Stochastic AltaRica Dataflow [55] semantics, but also provides parsers for widely used SBML [41] data exchange format. The corresponding simulator is easy to use and computationally efficient.

See also the web page http://www.labri.fr/ .