A functional representation of the simulation data of biochemical models based on molecular activity SIMON HARDY, PIERRE N. ROBILLARD Département de génie informatique et génie logiciel École Polytechnique de Montréal C.P. 6079, succ. Centre-Ville, Montréal (Québec) H3C 3A7 CANADA simon.hardy@mssm.edu, pierre-n.robillard@polymtl.ca Abstract: - Interpreting the simulation data of a complex biochemical model to understand its dynamic behavior is a difficult task. Traditional data representations display simulation outputs as concentration plots. To study the dynamic behavior of a model from these plots, it is necessary to have in mind the topology of the modeled system, know the function of the individual elements of the system and be able to describe their activity. Only with this mental image of the model can the dynamic behavior be deciphered. In this paper, we suggest exploiting this knowledge to create a preprocessing filter for the simulation data. This data filter is based on the concept of molecular activity and transforms the simulation data from a concentration perspective to a molecular activity perspective. This is done in two steps: identify the functional groups of the system, and mathematically describe the molecular activity of these groups. In this paper, we demonstrate this new data representation approach with a complex model of the signal transduction system of long-term potentiation in the hippocampal post-synapse, a model exhibiting a bistable behavior. To facilitate viewing of the resulting data matrix, the preprocessed data are displayed with known visualization techniques, followed by the production of an animated and a spectral functional representation. One advantage of the functional data filter is that, once created, it can be applied to a large number of simulation runs while at the same time performing parametric and structural modifications on the model in order to quickly explore the impacts on the model’s behavior. Key-Words: - biochemical modeling, simulation data, system dynamics, function, visualization 1 Introduction The simulation of biochemical models based on kinetic reactions generates data time series of concentration. To interpret these time series, the data are usually displayed on Cartesian graphs where time is the x-axis and concentration the y- axis. These plotted data are used to study the dynamic behavior of simulated models by combining sets of graphs. This simple type of representation is convenient for the study of simple models, but using it to study the dynamic behavior of complex models is a difficult task. Biochemical systems are highly organized and composed of many heterogeneous processes, formed by the interaction of the activities of different molecules. The signaling pathways of the cell are typical cases of such processes, where signals are transmitted by the perturbation of enzymatic activities and the actions of molecular messengers. An example of enzymatic activity is phosphorylation: some proteins are turned "on" and "off" by the addition and removal of a phosphate group, catalyzed by a kinase. Thus, it is possible to link a structural modification of a molecule to its role in a biochemical process. Researchers develop models of signaling pathway networks on the basis of experimentally established relationships between molecules, their function and the property of a biological system under study. Analyses of the simulation data of these models require a deep understanding of the underlying molecular or biochemical processes. To make sense of the simulation data, they are processed and filtered according to the researcher’s understanding of the activities of the system components. Our analysis will focus mostly on some relevant dynamic behaviors of model components. The original approach presented in this paper enables researchers to explore a new functional interpretation of a signaling pathway based on their existing knowledge of molecular activities by using simulation data. This functional interpretation transforms the data from a molecular concentration WSEAS TRANSACTIONS on BIOLOGY and BIOMEDICINE Simon Hardy, Pierre N. Robillard ISSN: 1109-9518 161 Issue 11, Volume 4, November 2007