Biology Reference
In-Depth Information
interaction of the TCR with a cognate peptide
MHC
complex. Although the observed overlap in TCR between
cell subsets may be explained by independent expansion of
cell subsets due to antigenic challenge, it also suggests that
the determination of cell fate is a random process indepen-
dent of TCR affinity, which is in support of the stochastic
model. Ultimately, the outcome of having shared or closely
similar TCR repertoire between different cell subsets is yet
another layer of increased coordination between compart-
ments of the immune system.
between larger cell populations. In dissecting molecular and
cellular data (i.e., cell-focused and system-focused) we
remain far from a true understanding the integrated system.
One main problem is that it becomes difficult for humans to
maintain an integrated and correct picture of knowledge that
extends beyond a few interacting entities. Computational
modeling of immune systems aims to provide a solution to
this problem by incorporating different data types.
Ideally, any computational model of the immune system
would seek to include representation for each of the many
involved cell types and their intercellular communications.
A first choice when generating a computational model is to
define the exact phenomenon that will be modeled as well
as to decide the level of detail required. Model building
iterates between two steps: specification and simulation.
Specification can be broken into three parts: (1) defining
model assumptions, which may be as general as how two
entities (e.g., proteins) interact, and how unknown infor-
mation is treated; (2) explicit inputting of all known
information on the studied phenomenon, which is usually
a tedious task that requires collection of data from scientific
papers and measurements, and their translation into a well-
defined executable specification; and (3) setting of
boundary conditions, e.g., the interface between the studied
phenomenon and other phenomena external to it.
Immune system modeling has been around for several
decades, but the computational infrastructure capabilities
only recently allow us to go beyond toy and educational
models to models that provide research with novel insights
and hypotheses [58] . Many groups have used mathematical
and computational models to gain a better understanding of
immunological data. As the topic of this section is specifi-
cally building an integrated understanding of the immune
response, we focus on models that span a large number of
cells and in which the phenomena are being modeled at the
whole organ or system level, rather than a smaller cell
e
Limitations of a System-Focused Approach
Blood is a complex tissue, particularly with respect to its
many white blood cell subsets, each of which has its own
molecular signature. The system-focused approaches dis-
cussed here (both for human immune monitoring and for
antibody repertoire) use novel technologies to measure an
enormous number of immune system components. Yet with
few exceptions they do so while losing the cellular context
of the data gathered. This results in several limitations. For
example, the frequency of a given cell subset in the blood
can vary markedly (2
10-fold difference) between indi-
viduals or over time, and thus differences detected from
blood between disease and control may to a large extent
reflect the fluctuations in cell type subset frequency
between samples [54,55] , rather than true molecular-level
changes associated with disease.
For human immune monitoring-focused studies it is
currently impractical to capture whole genome information
on each and every cell type or single cell, and for most
diseases it is unclear whether there is a predominant cell
type that can be sorted and studied (i.e., in a cell-focused
manner as described above) associated with a disease.
Experimental designs that include time courses elevate this
problem to some extent, as an individual's samples are
compared to one another rather than across individuals, for
which there is usually much more variation. Recently, we
and others have developed computational solutions for
post-hoc extraction of cell type-specific information from
heterogeneous tissue data ( Box 25.3 ) [56,57] . Such tech-
niques often increase the signal-to-noise ratio in system-
wide studies by orders of magnitude and greatly facilitate
data interpretation, while also allowing the gathering of
both cell-specific and system-wide information, thereby
effectively creating a bridge between these two approaches.
e
cell
interaction level.Wewill describe here two levels of models:
cellular automata-based models and Statecharts-based
models. These two 'model specification languages' differ
greatly in their richness to express complicated constructs,
and from this stem large differences in the specification
efforts required by the researcher and ultimately the breadth
of the immune system that the model covers. Common to
both is that through simulation coupled to graphical visu-
alizations, both model classes are beginning to unravel the
complex cross-talk between cells, and are providing
a framework to study emergent system properties.
e
MULTI-SCALE-FOCUSED SYSTEMS
IMMUNOLOGY
Protective immunity is not the final outcome of any single
cell, but rather requires the execution of multiple processes
that draw on functionality elicited by many cell types
communicating between one another, and sometimes even
System-Wide Meso-Scale Cellular Automata
Models of Immunity
The simplest models rely on the ideas of cellular automata,
discrete models consisting of a grid of cells, each in one of
Search WWH ::




Custom Search