Qualitative Modeling of Cellular Networks with CellNetAnalyzer and ...

Tutorial ICSB 2010 (Edinburgh)
Qualitative Modeling of Cellular Networks with CellNetAnalyzer and Odefy and
linking it to high-throughput data with DataRail and CellNetOptimizer
Steffen Klamt 1* , Julio Saez-Rodriguez 3* , Oliver Hädicke 1 , Regina Samaga 1 , Jan Krumsiek 2 , Leonidas Alexopoulos 4
1 Max Planck Institute for Dynamics of Complex Technical Systems, D-39106 Magdeburg, Germany; 2 Helmholtz
Zentrum München, D-85764 Neuherberg, Germany; 3 EMBL-EBI European Bioinformatics Institute, Hinxton, UK;
4 National Technical University of Athens, Greece
* Contact: klamt@mpi-magdeburg.mpg.de, saezrodriguez@gmail.com
Summary: This tutorial will present methods and software tools for structural analysis and qualitative
modeling of metabolic and signaling networks (first part) and for linking qualitative modeling of signaling
networks with high-throughput proteomic data (second part). The tutorial will consist of presentations and
hands-on exercises. For the latter, participants are encouraged to bring their own computers (with
MATLAB and software packages installed).
Part I (~130 min): Structural analysis and qualitative simulation of cellular networks using
CellNetAnalyzer and Odefy
Break (~20 min)
Part II (~90 min): From high-throughput protein activity data to cell-specific logic models using
DataRail and CellNetOptimizer
Total Duration: 4 hours; Expected number of participants: 30-40
Part I: Structural Analysis and Qualitative Simulation of Cellular Networks
with CellNetAnalyzer and Odefy
Steffen Klamt, Oliver Hädicke, Regina Samaga, Jan Krumsiek, Fabian Theis
Introduction: An important class of modeling techniques in Systems Biology focuses on the investigation of
topological and qualitative properties of cellular networks without requiring detailed kinetic descriptions of
the underlying molecular mechanisms.
CellNetAnalyzer is a MATLAB toolbox providing a graphical user interface and a comprehensive set of
methods with various, partially unique, functions and algorithms for exploring structural properties of cellular
networks (Klamt et al., BMC Syst. Biol. 1:2, 2007). Both mass-flow (metabolic) and signal-flow
(signaling, regulatory) networks are supported. A particular strength of CellNetAnalyzer are methods for
functional network analysis, i.e. for characterizing functional states, for detecting functional dependencies,
for identifying intervention strategies, and for giving qualitative predictions on the effects of perturbations.
The capabilities of CellNetAnalyzer for analyzing signaling networks were recently extended by integrating
a plugin version of Odefy. Odefy is a MATLAB and Octave compatible toolbox which can be used to convert
Boolean models (including those in CellNetAnalyzer format) into systems of ordinary differential equations
(ODEs) approximating standard kinetic laws (Wittmann et al., BMC Syst. Biol. 3:98, 2009). Using
Odefy, coarse-grained Boolean models can be translated into such qualitative ODEs allowing one to study
essential dynamic features of regulatory and signaling networks in combination with measurement data.
Contents: We start with an overview of existing approaches for structural and qualitative analysis of
metabolic and signaling networks. Then, after a general introduction into CellNetAnalyzer, several
modeling techniques will be exemplified by examining toy models and realistic networks with
CellNetAnalyzer. Functionalities of Odefy will be demonstrated as well. Participants are encouraged to
bring their own laptops and to follow the live demonstrations and hands-on exercises. Topics that will be
discussed are:

� structural network analysis vs. ODE based modeling
� mass flows (metabolic networks) vs. signal flows (signaling networks)
� general set-up and basic usage of CellNetAnalyzer
� methods for mass-flow networks (with live demonstration of CellNetAnalyzer): graph-theoretical
properties, conservation relations, null-space analysis, constraint-based modeling, metabolic flux
analysis, flux balance analysis, pathway analysis based on elementary modes, minimal cut sets
� methods for signal-flow networks (with live demonstration of CellNetAnalyzer and Odefy):
- interaction graphs: feedback loops and signaling paths, network-wide interdependencies and
dependency matrix, predicting qualitative effects of perturbations
- logical/Boolean networks: analyzing the qualitative input/output behavior of signaling networks
using logical steady states, structural couplings of signal flows, minimal intervention sets (identification
of combinatorial interventions preventing or provoking certain qualitative responses)
- Odefy: methodology for transforming Boolean models into ODEs; simulation of Boolean and ODE
models with Odefy; large-scale example
� other functionalities of CellNetAnalyzer (displaying experimental data, API, import/export)
For academic use, CellNetAnalyzer (together with plugin Odefy) can be downloaded for free from:
www.mpi-magdeburg.mpg.de/projects/cna/cna.html. MATLAB (version 7.1 or higher) is required.
Part II: From high-throughput protein activity data to cell-specific logic models
using DataRail and CellNetOptimizer
Julio Saez-Rodriguez, Leonidas Alexopoulos
Purpose and background: The process of constructing and testing models against high-throughput data,
particularly those models that incorporate significant prior knowledge, involves multiple steps that are
currently very poorly integrated. Together with colleagues in the groups of Peter Sorger and Douglas
Lauffenburger at the CellDecisionProcess Center at M.I.T. and Harvard Medical School, we have
developed SB-Pipeline to create an effective workflow based on public standards and modern software
practice. SB-Pipeline is a multi-faceted software platform that pulls together all of the steps involved in
collecting and transforming primary data; constructing, annotating and calibrating models; and distributing
and sharing simulations and analyses. SB-Pipeline is primarily concerned with data and model management
for the purpose of calibration, and implements a robust system for tracking the provenance of data, links
between data and models, and the origins of model assumptions in data or the literature. SB-Pipeline is a
collection of discrete but interoperable software tools, rather than a single integrated system, and
incorporates standard protocols for import and export of data.
In this second part of the tutorial we will present two modules of SB-Pipeline: DataRail, and
CellNetOptimizer and use a data-set of high-throughput functional data of signal transduction in liver cells
(Alexopoulos et al., submitted) to illustrate its use.
- DataRail is an open source MATLAB toolbox for managing, transforming, visualizing, and modeling
data, in particular the varied high-throughput data encountered in Systems Biology (Saez-Rodriguez et al.,
Bioinformatics, 24(6):840-7, 2008). It supports data-driven models, in particular Multiple Linear
Regression (MLR), Partial Least Squares Regression (PLSR), and Bayesian Inference.
- CellNetOptimizer (CNO) is a MATLAB toolbox to turning pathway maps into logical models (Boolean or
Fuzzy) that can be calibrated against experimental data (Saez-Rodriguez et al., Mol. Syst. Biol. 5:331,
2009), generating functional, predictive, cell-type specific models of mammalian signal transduction. Representation
and simulation of logical networks follows the same rules as in CellNetAnalyzer.
DataRail and CellNetOptimizer are complementary and compatible with CellNetAnalyzer and Odefy, tools
focused on the analysis of cellular networks presented in the first part of the tutorial.
SB-Pipeline resources can be downloaded from http://code.google.com/p/sbpipeline/ and
http://www.cdpcenter.org/resources/. MATLAB is required to run DataRail and CellNetOptimizer.