A platform for in silico modeling of physiological systems II. CellML compatibility and other extended capabilities.

Abstract

The number of biological models published in peer reviewed journals and complexity of each of those models are rapidly increasing, making it difficult to reproduce simulation results of the published models and to reuse the models by third persons. This paper is a continuation of our previous report on a software platform development as a solution to such difficulties. We describe progresses of our development. Those include improvement in functional capabilities to import and simulate published models in the CellML model repository, to browse and edit CellML models and then to export them as new models either with the CellML format or with a XML format defined for our platform (ISML), and to newly construct large scale models by connecting CellML/ISML models. Several advantages to use ISML in parallel with CellML are; 1) ISML can deal with geometry (morphology) of a model, enabling the user to perform geometry dependent modeling and simulations. 2) ISML can deal with time series data, both simulated and experimentally acquired data, for visualization of dynamics.

Full text

Page 1

30th Annual International IEEE EMBS Conference
Vancouver, British Columbia, Canada, August 20-24, 2008


?
Abstract?The number of biological models published in peer
reviewed journals and complexity of each of those models are
rapidly increasing, making it difficult to reproduce simulation
results of the published models and to reuse the models by third
persons. This paper is a continuation of our previous report on a
software platform development as a solution to such difficulties.
We describe progresses of our development. Those include
improvement in functional capabilities to import and simulate
published models in the CellML model repository, to browse and
edit CellML models and then to export them as new models
either with the CellML format or with a XML format defined
for our platform (ISML), and to newly construct large scale
models by connecting CellML/ISML models. Several advantages
to use ISML in parallel with CellML are; 1) ISML can deal with
geometry (morphology) of a model, enabling the user to perform
geometry dependent modeling and simulations. 2) ISML can
deal with time series data, both simulated and experimentally
acquired data, for visualization of dynamics.

I. INTRODUCTION
HYSIOME and systems biology have been recognized as
emerging and important research areas that can integrate
growing experimental data at multiple levels and scales of the
human body for better understanding of human physiological
functions [1,2]. Mathematical models, in particular dynamical
system models, of physiological functions play a key role to
integrate the vast amount of pieces of knowledge. This can be
argued with the following reasons. 1) Mathematical models
are capable of describing time evolution of states of biological
systems based upon physical and chemical principles or
phenomenological logics governing behaviors of the system.
Solutions of a given system can be associated quantitatively

Manuscript received April 7, 2008. This work was supported in part by the
???????????????????????????????????????????????????????????? Y. Suzuki, T.
Kawazu, M. Nakanishi, Y. Taniguchi, and T. Nomura are with Graduate
School of Engineering Science and The Center for Advanced Medical
Engineering and Informatics at Osaka University, Toyonaka, Osaka,
560-8531 Japan.(phone: +81-6-6850-6534; fax: +81-6-6850-6534;e-mail:
suzuki@bpe.es.osaka-u.ac.jp, kawazu@bpe.es.osaka-u.ac.jp, nakanishi@
bpe.es.osaka-u.ac.jp, ytaniguchi@bpe.es.osaka-u.ac.jp, taishin@bpe.es.
osaka-u.ac.jp). Y. Asai is with The Center for Advanced Medical
Engineering and Informatics at Osaka University, Toyonaka, Osaka,
560-8531 Japan.(e-mail: asai@bpe.es.osaka-u.ac.jp). E. Heien and K.
Hagihara are with Graduate School of Information Science and Technology
at Osaka University, Toyonaka, Osaka, 560-8531 Japan.(e-mail:
e-heien@ist.osaka-u.ac.jp, hagihara@ist.osaka-u.ac.jp). Y. Kurachi is with
Graduate School of Medicine and The Center for Advanced Medical
Engineering and Informatics at Osaka University, Suita, Osaka, 565-0871
Japan.(e-mail: ykurachi@pharma2.med.osaka-u.ac.jp)

with changes in time of experimentally observed biological
signals and images. 2) Biophysical models require
morphology and geometry of biological organs and organelles
with high-precision measurement in order to perform accurate
numerical simulations of
structure-function linkage can be quantitatively revealed. 3) A
mathematical system model can be divided into subsystems
(modules), each of which may correspond to biological
entities at different levels and scales, leading to understanding
hierarchical mechanisms of physiological functions.
However, the number of mathematical models describing
biological functions published in peer reviewed journals is
rapidly increasing. Moreover, complexity of each of those
published models increases as the computational performance
increases. These make it difficult to reproduce simulation
results of the published models and to reuse the models by
third persons, apparently obstructing promotion of sciences
and the knowledge integration. The pioneering effort to
overcome this problem has been promoted by CellML project
[3,4] as a part of IUPS Physiome Project. CellML is a XML
based markup language which aims at describing
mathematical models of biological functions such as electrical
activities of cell membranes. Each model written in the
CellML format can be downloaded from the CellML model
repository, and includes all information necessary to
reproduce simulated results described in the corresponding
publication about the model. CellML project provides a
software called PCEnv to simulate dynamics of the CellML
formatted models.
This paper is a continuation of our previous report [5] on a
comparable and complementary effort to CellML project. We
have also proposed a XML based model description format
referred to as insilicoML (abbreviated as ISML). For ISML,
insilicoIDE is used to support the user for model construction
(generation of ISML formatted models) and for automatic
generation of C++ source codes to numerically simulate
dynamics of the models. The insilicoIDE, together with model
databases, is on the development to play a role as a platform
for in silico modeling of physiological systems. ISML and
insilicoIDE are CellML compatible and have several
additional unique features. In this paper, we describe
progresses of our platform development. Those include
improvement in functional capabilities to import and simulate
published models in the CellML model repository, to browse
and edit CellML models and then to export them as new
the model, by which
A Platform for in silico Modeling of Physiological Systems II.
CellML Compatibility and Other Extended Capabilities
Yasuyuki Suzuki, Yoshiyuki Asai, Toshihiro Kawazu, Masao Nakanishi, Yoshiki Taniguchi, Eric
Heien, Kenichi Hagihara, Yoshihisa Kurachi, and Taishin Nomura
P
978-1-4244-1815-2/08/$25.00 ©2008 IEEE. 573
Authorized licensed use limited to: OSAKA UNIVERSITY. Downloaded on December 10, 2008 at 19:25 from IEEE Xplore. Restrictions apply.

Page 2

models either with the CellML format or with ISML format,
and to construct new large scale models by connecting
numbers of CellML/ISML models. Several advantages to use
ISML in parallel with CellML are also illustrated. Those are;
1) ISML can deal with geometry (morphology) of a model,
enabling the user to perform geometry dependent modeling
and simulations with the morphology of the models. 2) ISML
can deal with time series data, both simulated and
experimentally acquired data, for visualization of dynamics.
II. SYSTEM OUTLINE
Let us briefly summarize features of the platform that we
have described previously[5]. Those include the ISML model
description, the insilicoIDE (IDE), and the model database.
Progresses on the CellML compatibility (functionalities to
import and export CellML files) and the morphology database
used with ISML are then illustrated in the following sections.
We consider a target biological system as an aggregate of
elements referred to as modules. In ISML, each module is
characterized by the XML tags such as ID number, edges,
state, parameters, dynamic-rules and morphology among
others. The edges represent the structural and functional
relationships among modules. The structural relationships
defined as the edge types are ?include? and ?constituent? to
represent hierarchical relationships, and ?attachment? to
represent modules are glued each other (Fig. 1). A set of
modules can form a module by declaring encapsulation which
makes an encapsulated module independent and to be easily
reused (Fig. 2). Two or more modules can be functionally
related with each other by another type of the edge, the
functional link. Ports of an encapsulated module are packed in
a class object which has properties describing information
necessary to use the module as a sub-module of larger models.
The state and dynamic-rules of a module are responsible for
dynamics of the module, where the state is time-updated using
the dynamic-rules.
The insilicoIDE represents modules graphically as ball-like
objects, where four kinds of relationships between modules
are displayed as different types of lines (Fig. 1). Functional
relationships connect between modules through the ports

Fig. 2. Encapsulation. An encapsulated module is surrounded by a
square. The user can connect encapsulated modules by a function link
without taking care of detailed structure of the modules (the
transparent part of the modules on the right panel).

Fig. 3. Outline of the platform. The insilicoIDE can load both ISML and CellML formatted models from the model databases. The users can develop their
own models on insilicoIDE and export them as either ISML or CellML models. The models developed on insilicoIDE can be simulated by automatically
generating C++ (MPI C++) source code files. Simulated data or experimental data (time series) can also be loaded for visualization. Note that, when an
ISML model uses geometry of biological organs, the corresponding morphological models are taken from the morphology database.

Fig. 1. Relationships among modules defined in ISML. Each ball is a
module. Modules are connected by several types of lines. From left to
right, it is the ?include? like hierarchical structure, ?constituent? like
hierarchical structure, ?attachment? like relationship, and ?functional?
relationship. The first two relationships represent hierarchical structure
of modules. The attachment is represented by bidirectional arrow line.
The functional relationship is uni-directional, and information flows
from non-marked end to marked end.
574
Authorized licensed use limited to: OSAKA UNIVERSITY. Downloaded on December 10, 2008 at 19:25 from IEEE Xplore. Restrictions apply.

Page 3

which are depicted as the small circles on the modules. Each
encapsulated module is indicated by the square surrounding
the ball-like module. The user can edit modules and
relationships (hierarchical structure of modules) using
mouse-operated GUI. For example, as shown in Fig. 2, the
aggregate of modules that are appropriately encapsulated can
be simply treated as a single object, and the user can use it
without taking care of detailed structure and complexity of
underlying modules. Once the model construction is
completed, the insilicoIDE can export the model either as a
C++/MPI-C++ source code or as a CellML/ISML file. The
exported ISML file may be registered, after publication, to the
model database to be used by other users. With the source
code files, the user can perform model simulation either
during IDE runtime without apparent compiling or off-line
with manual compiling and execution (Fig. 3). Note that, as
described in [4], ISML and IDE are capable of describing and
simulating multi-agent type models as well as models
described by a set of ordinary differential equations (ODEs).
III. CELLML COMPATIBILITY
The CellML model repository that has been in the public
domain provides more than 140 models in CellML format in
the public domain. The insilicoIDE is fully compatible to
CellML in the following senses. The insilicoIDE can import
any model files with CellML 1.1 specification as examined by
the models in CellML model repository as of March 2008, and
most of the models can be simulated to exhibit reasonable
dynamics. The insilicoIDE can export newly constructed
models described by ODEs as CellML 1.1 formatted files. The
exported model files by the IDE are validated by performing
dynamic simulations of the models on PCEnv, which is the
model-simulator provided by the CellML project. Figure 4
shows simulated waveforms on PCEnv for a model of
neuro-musculo-skeletal system constructed on the IDE. A
CellML formatted file used for this simulation on PCEnv is
not from the CellML model model repository but it is
generated and exported by the IDE, validating the CellML file
export functionality of the IDE.
Figure 5 exemplifies a collaborative simulation between
CellML and ISML. In this example, a cardiac cell model (Luo
and Rudy[6]) and a model of gap-junction are downloaded
from, respectively, the CellML model repository and the
ISML model database, and they are imported on the IDE as
modules. The cell model is then encapsulated with setting its
membrane potential at an output port and the stimulation
current at an input port. By copying those modules and
connecting between ports of the cell and gap-junction models,
the user can easily simulate dynamics of the diffusively
coupled cardiac cell models (a model with five cells coupled
by the gap-junctions in Fig. 5).
IV. USE OF MORPHOLOGY AND TIME SERIES DATA
In this section, we illustrate other extended capabilities of
ISML and the insilicoIDE, motivating the complimentary use
of this platform with CellML and its APIs. The extended
capabilities include the use of a morphology database and
time series data obtained in physiological experiments and/or
numerical simulations of models.
Figure 6 shows snapshots of the insilicoIDE in which
dynamics of a model of human neuro-musculo-skeletal system
[7] are simulated. The model includes three-dimensional
geometry of the skeletal system of human body, models of six
muscles for an upper arm attached on the skeletal system, and
neural motor controllers for the muscles, each of which is
described by ISML as a reusable module and registered in the

Fig. 5. Construction of a coupled cell models on insilicoIDE. A cardiac
cell model and a gap-junction model are imported from, respectively,
CellML and ISML databases. Copying and connecting the models
allow the user to simulate the constructed model.

Fig. 4. Dynamic simulation of a CellML formatted model exported by
the insilicoIDE on PCEnv. This graph shows a simulated result for the
neuro-musculo-skeletal system. The blue line represents the rotating
angle of shoulder, red line the rotating angle of elbow.
575
Authorized licensed use limited to: OSAKA UNIVERSITY. Downloaded on December 10, 2008 at 19:25 from IEEE Xplore. Restrictions apply.

Page 4

ISML model database (see Fig. 3). The morphology of each
skeletal link, in this case, is represented by a surface data
(VRML), and separately stored in the morphology database to
be reusable for general purposes. The ISML file of this
neuro-musculo-skeletal system model is described as a set of
different modules including the arm and the muscles, and has
links specifically to the morphological models in the
morphology database. The morphological models that can be
registered in the morphological database include surface and
volume data of three dimensional
morphological model is registered with a metadata file
describing the metric unit of the model and keywords to
characterize the model. The morphological models in the
database are used not only for visualization purposes but also
for dynamic simulations if a model utilizes morphometric
information such as in multi-agent simulations.
Another new functionality of the insilicoIDE is to relate
time series data obtained either by numerical simulations or
physiological experiments to models on the IDE. When
dynamics of a model are simulated using IDE, the obtained
time series data usually represent changes in the states
properties of all modules comprising the model. For
experimental data that are not obtained directly by simulations
on the IDE, the user specifies modules (properties of modules)
that can be underlying models of the data. In order to facilitate
describing relationships between a given set of time series
data and the corresponding modules on the IDE, a XML based
markup language, referred to as TSML, is designed, and time
series data are maintained with the TSML format.
Figures 6 and 7 exemplify use cases of the functionalities of
both morphology database and time series data. In Fig. 6, the
simulation data obtained from the neuro-musculo-skeletal
system used for Fig. 4 is represented by TSML which is
imported to the IDE to show the three-dimensional human arm
movement. In Fig. 7, a motion captured data of human body
movement is represented by TSML. More specifically,
changes in the three-dimensional positions of several markers
are observed experimentally, and the motion data of the
markers are saved by TSML format. The human skeletal
objects. Each
system with markers is selected (or constructed) on the IDE as
a model to be related to the data. This TSML and the model
are merged on the IDE to visualize the motion.
V. CONCLUSION
Several functionalities such as importing and simulating
published models in the CellML model repository, browsing
and editing a CellML model and then exporting it as a new
model either with the CellML or ISML format, and newly
constructing large scale models by connecting numbers of
CellML/ISML models are demonstrated. Benefits to use both
ISML and CellML are described. ISML can deal with
morphology of models, enabling the user to perform geometry
dependent modeling and simulations. ISML with the IDE can
deal with time series data, both simulated and experimentally
acquired data, for visualization of dynamics.
REFERENCES
[1] J. B. Bassingthwaighte. Strategies for the Physiome Project, Annals of
Biomedical Engineering, Vol. 28, pp. 1043-1058, 2000
[2] P. J. Hunter and Thomas K. Borg. Integration from proteins to organs:
The Physiome Project. Nature Review Mol Cell Biol, 4(3), pp. 237-243,
2003
[3] http://www.cellml.org/
[4] C. M. Lloyd, M. D.B. Halstead, P. F. Nielsen. CellML: its future,
present and past. Prog Biophys Mol Biol, Vol. 85, No.2-3 pp. 433-450,
2004
[5] T. Kawazu, M. Nakanishi, Y. Suzuki, T. Nomura. A Platform for in
Silico Modeling of Physiological Systems. Proceedings of the 29th
Annual International Conference of the IEEE EMBS Cité
Internationale, Lyon, France, August 23-26, 2007, pp. 1394-1397,
2007
[6] C. H. Luo and Y. Rudy. A model of the ventricular cardiac action
potential. Depolarization, repolarization, and their interaction.
Circulation Research, Vol. 68, No. 6, pp. 1501-1526, 1991
[7] S. Stroeve, ?Impedance characteristics of a neuro musculo skeletal
model of the human arm?, Biological Cybernetics, Vol. 81, pp.
475-494, 1999

Fig. 7. An experimentally obtained motion captured data during human
movement is represented by TSML (time series markup language) and
imported to the insilicoIDE to visualize the motion with a
three-dimensional of skeletal system model.

Fig. 6. Three-dimensional visualization of simulated dynamic behavior
of a model using the insilicoIDE.
576
Authorized licensed use limited to: OSAKA UNIVERSITY. Downloaded on December 10, 2008 at 19:25 from IEEE Xplore. Restrictions apply.