Working group: "Computational Biology"

2010, the 16th of November

Ferdinanda Camporesi

Title: On model reduction


Molecular biological models usually suffer from a large combinatorial explosion. Indeed, proteins form complexes and modify each others, which leads to the formation of a huge number of distinct chemical species. Thus we cannot generate explicitly the quantitative semantics of these models, and even less compute their properties. Model reduction aims at reducing this complexity by providing another grain of observation.

Rules-based languages offers a convenient and compact way for describing this kind of system and combinatorial complexity is partially avoided thanks to context-free rules, in which the set of all potential contexts of application for an interaction does not need to be written explicitly. Anyway that is not enough when considering quantitative aspects. Starting from the fact that rules cannot observe observe the correlation between specific parts of some chemical species, we are able to cut them in autonomous fragments whose behavior is an exact projection of the original system's behavior. This reduction can be further improved by taking into account some symmetries between sites.

We propose a generic framework to formalize and combine model reductions and we apply it both to the differential semantics and to the stochastic one.