Agent-based
modelling - a tool for replicating biological systems
Agent-based
modelling - a tool for replicating biological systems
IntroductionThis implies that we
cannot
impose behaviour at a tissue
level on the individual cells – the behaviour at tissue level has to be
an
emergent property of individual cell behaviour. The emergent behaviour
has to be compatible
with continuum models describing the physical environment and
interactions, but
should not be driven by a continuum model – if it is, it is not
emergent
behaviour.
Our way of
approaching this is
to introduce the
concept of representing
biological systems by a virtual replicate. A replicate is defined as a
software
system (a model) which incorporates a 1:1 mapping of biological
entities
(protein molecules, receptors, cells etc) into software agents
(specifically,
into X-machines), with individual agents
having properties analogous to
those
of the biological entities that they represent. The X-machine is
formally
defined, as a result of which the virtual replicates are also formally
defined,
and their performance can be formally verified. The X-machine is Turing
complete, which has two consequences: any function within an X-machine
can be
replaced by another X-machine which can evaluate the function,
automatically
giving rise to hierarchical models; and the set of functions within an
X-machine could constitute a formal mathematical model (e.g. a
differential
equation model), thus providing a universal framework linking any type
of
model.
Rod
Smallwood, Mike
Holcombe, Jenny
Southgate, Sheila Mac
Neil, Richard
Clayton, Rodney Hose, Peter Hunter,
Rob Gaizauskas
Dawn
Walker (Dawn Wood),
Sun Tao, Alan Waterworth, Jiujiang Zhu, Phil McMinn, Simon Coakley, Nik
Georgopoulos, Steven Wood, Andrew Leathard
The aim of the
Epitheliome Project is to develop
a computational model that is able
to predict the social behaviour of cells in epithelial tissues.
Epithelial tissues form the barriers between us
and the outside world - our skin, the lining of all our body cavities
(mouth,
lungs, cervix, bladder, prostate gland, our intestines). They are very
thin -
typically about 0.5 mm thick, perhaps 10 cells - but have specialised
functions. Key to epithelial behaviour is the protective barrier
function
coupled with enormous repair potential. Thus skin prevents us
dehydrating and
protects us from disease organisms, the bladder epithelium (the
urothelium) is
watertight and prevents urine damage or contamination of circulating
blood, the lining of our intestines
protects us from potentially damaging ingested material (eg bacteria)
while
selectively absorbing nutrients. It is not surprising giving the role
of
epithelia and their proliferative potential that all cancers, other
than those
originating from haemopoietic and mesenchymal cells, originate in
epithelial
tissues, which are relatively simple.
Epithelial tissues are obviously important - we can't live without
them! They are also relatively simple - they contain a limited number
of different cell types, no blood vessels, no nerve endings. They are
the source of important clinical problems - cancer, wound healing,
diabetic ulcers, skin graft contraction. The ultimate aim of our
modelling is to better understand these problems, and thus be able to
do something about them.
All the tissues in our bodies (to be more general, all multi-cellular
creatures) self-assemble. The 'rules' for doing this are in each cell -
in the genetic material. There is no information at a higher level of
organisation than the individual cell, so all the organisation in
tissues and organs and organisms is an 'emergent property' of the
interaction of large numbers of individual cells - 1013 in a human.
That is what we are interested in - how does
this social interaction of the cells produce properly functioning and
structured creatures?
There is a lot more about the project on the Epitheliome
Project web pages - models, publications, and links to other
interesting sites.
NF-κB signallingEva
Qwarnstrom, Mike
Holcombe,
Rod
Smallwood
Mark Pogson, Hong Bum Kim, Ian Palmer
The
intracellular NF-κB signalling pathway is vital to immune response
regulation.
To understand better its operation, a suitably detailed model is
required to
account for both spatial and temporal aspects of the pathway. We have
developed
a novel agent-based model to deal with the complexities of the system
and to
extend the capabilities of previous models.
The agent-based model is extensible and robust, and provides an
intuitive method to determine and explore the key features of the
system. This
is a systems biology model that addresses pathway behaviour from
initiation at
receptors on the cell membrane down to gene activation and regulation.
The
model agrees extremely well with experimental data obtained using
analysis of
single cells in vitro, and from the
model we predict some key properties that have not, as yet, been
investigated
experimentally.
[Top]
Social Ants
The
inspiration for applying individual based
modelling to describe the social life of the cells was an enthusiastic
description by Mike Holcombe of his work with Francis Ratnieks (Animal and Plant Sciences) on
the modelling of social insects. You can find out all about social
insects research on the Apiculture and
Social Insect Laboratory web site.
[Top]
X-machines
Mike
Holcombe, Marian
Gheorghe
Eilenberg
developed a general (Turing-complete) computational machine which he
called an
X-machine, and it was further developed by Holcombe.
The X-machine is similar to a Finite State Machine, but has two
important
differences: an underlying data set, and a set of functions which
define the
state transitions.
Computational
Systems
Biology GroupThis
research is all part of the Computational
Systems Biology Group in the Department
of Computer Science. Links to all the members of the group can be
found on the group's web page.
[Top]
