user login 

BITS2007 Meeting
BITS2007 Meeting

26-28 April 2007 Napoli, Italy

email support
Home > Contribution details
get PDF of this contribution get XML of this contribution get ICal of this contribution
Development of a systems biology infrastructure for mathematical modelling and parameter estimation
Nowadays it is important to describe integrated biological processes such as the cell
cycle or signal transduction in quantitative functional terms [1]. Since these
pathways are not just an assembly of genes and proteins, some properties cannot be
fully understood merely by drawing diagrams of their interconnections [2]. In order
to model the cell cycle mechanism it is appropriate to use ordinary differential
equations (ODEs), if the two basic assumptions of this approach, the `well-stirred`
chemical reactor and sufficient great concentrations, are satisfied [3,4]. According
to this approach, each component of the modelled system is associated with a
differential equation which describes its temporal rate of change. One of the main
problems in developing new mathematical models, using the ODEs approach, lies in the
experimental outline of the specific kinetic parameters, such as rate and association
constants, needed in the kinetic laws. 
Moreover, although standards for codification of biological models in
computer-readable formats (SBML and CellML) exist, many published models are not
included in models repositories, like BioModels, CellML models repositories and JWS
on-line. To overcome these limitations we have developed an infrastructure with the
aim to improve the creation, the sharing and the simulation of models based on ODEs
in the context of the systems biology.

The technology for the simulation consists in five modules: a model repository, an
XML based parser, the MathML to HTML converter, the ODEs simulation engine, and the
parameter estimation tool. PHP code links all the modules and generates web
interfaces  which allow users to interact with the system. Most of the models
collected in the repository comes from the BioModels database at EBI, some other
models derive from the model authors web sites, or have been manually implemented
using the JigCell software [5]. The models are encoded in Systems Biology Markup
Language (SBML), an XML-based language specific for biological models. The parser has
been written in object oriented PHP and its structure follows the SBML data structure
specifications. The implemented system allows users to translate formulas, which in
SBML files are encoded with MathML specifications, to HTML using an XSLT library [6]
and TTH, a tool to translate TEX to HTML [7]. The simulation engine relies on XPPAUT 
[8] and the output results are show using GNUPLOT [9]. The parameter estimation is
performed using a evolution strategy algorithm, named Stochastic Ranking Evolution
Strategy [10] coupled to sundials library [11] for numerical calculations of the
differential equations. Parameter estimation execution is parallelized thanks to a C
based Message Passing Interface implementation of SRES algorithm [12] on high
performance computing device.

The web interface allows the user to acquire information about models in two
sections. The first section regards the model publication paper information,
including model wiring-diagram. The second allows the users to explore the SBML
components of the selected model, including its complete mathematical definition,
which consists in species initial concentration and parameters values, differential
equations, algebraic equations,  assignment  rules and events. 
Our technology allows the real-time simulation of the models: this feature is
essential to fully understand the complex behaviours of the pathway, such as
oscillations. Moreover, it is possible to observe model response against change of
initial conditions, i.e. proteins species concentrations and kinetic parameter
values. Results are instantaneously plotted on 2D graphs which show time courses and
phase diagrams. In the parameter estimation section it is possible to submit a novel
model and optimize its kinetics parameter values according to experimental data
provided. The algorithm we used, SRES, has demonstrated good performance in the
estimation of continuous variables [13]. Since the computational load of this
optimization problem scales rapidly both with the cost of a single model simulation
and with the number of its kinetic parameters, the parameter estimation job is
executed using high performance computing.
Id: 197
Place: Napoli, Italy
Centro Congressi "Federico II"
Via Partenope 36
Starting date:
-- not yet scheduled --   
Duration: 01h00'
Contribution type: Poster
Primary Authors: MOSCA, Ettore (Department of Biotechnology and Biosciences, University of Milano-Bicocca, Milano; Institute for Biomedical Technologies, National Research Council, Segrate (Milano))
Co-Authors: MERELLI, Ivan (Institute for Biomedical Technologies, National Research Council, Segrate (Milano))
MILANESI, Luciano (Institute for Biomedical Technologies, National Research Council, Segrate (Milano); CILEA, Consorzio Intrauniversitario per l`Elaborazione Automatica, Segrate (Milano))
ALFIERI, Roberta (Institute for Biomedical Technologies, National Research Council, Segrate (Milano); CILEA, Consorzio Intrauniversitario per l`Elaborazione Automatica, Segrate (Milano))
Presenters: MOSCA, Ettore
Included in session: Poster Session
Included in track: Gene expression and system biology | Last modified 08 July 2009 10:35 |

Powered by Indico 0.90.3