Information Technology Reference
In-Depth Information
Gene expression programs involved in the production of peripheral tolerance
have so far only been investigated using DCs derived from a myelomonocytic tumor
cell line, but also suggest that DC interaction with regulatory T cells can induce a
tolerogenic gene expression program including the induction of antiapoptotic genes
(Suciu-Foca Cortesini, Piazza, Ho, Ciubotariu, LeMaoult, Dalla-Favera, and
Cortesini 2001). More detailed analysis of the way in which various pathogens and
PAMPs affect transcriptional profiles leading to differential DC functions will help
uncover the contribution of DC plasticity in shaping the immune response.
7.4 Integrating Genomics Data: A System View
of the Immune Response
What becomes immediately clear with genome-scale surveys of transcriptional
programs and protein interaction maps is that focusing on single genes and gene
families will not provide an understanding of the complexity of the integrated
immune response. This understanding requires abstraction of models from existing
theories and high-dimensional genomics datasets. Modeling should not be an
anathema to biology although models are often treated with suspicion relative to
empirical data. This arises mainly through a misunderstanding of the purpose of
models. Biologists intuitively construct models all the time by forming hypotheses.
Such mental and verbal models concentrate on describing and integrating selected
aspects of their research, leaving aside certain facts as irrelevant. Good models make
planning the next experiment possible and allow a prediction of the results. If the
results differ from predicted, the model is adjusted. Modeling in systems biology is
simply a formalization of this process mathematically and extending the model to
large datasets.
7.4.1 Models of the Immune Response
To be able to produce such abstracted models from large genomic datasets it is
necessary to define the structure of interactions within a network of genes and
proteins, determine the dynamic relationships between the gene and protein
components, and determine the integrated network behavior. Insights come through
either data-driven or model-driven approaches to these aims. Network structures are
beginning to be compiled either by physically mapping protein—protein interactions
(Bouwmeester, Bauch, Ruffner, Angrand, Bergamini, Croughton, Cruciat, Eberhard,
Gagneur, Ghidelli, Hopf, Huhse, Mangano, Michon, Schirle, Schlegl, Schwab, Stein,
Bauer, Casari, Drewes, Gavin, Jackson, Joberty, Neubauer, Rick, Kuster, and
Superti-Furga 2004; Gavin, Bosche, Krause, Grandi, Marzioch, Bauer, Schultz, Rick,
Michon, Cruciat, Remor, Hofert, Brajenovic, Ruffner, Merino, Klein, Hudak,
Dickson, Rudi, Gnau, Bauch, Bastuck, Huhse, Leutwein, Heurtier, Copley,
Edelmann, Querfurth, Rybin, Drewes, Raida, Bouwmeester, Bork, Seraphin, Kuster,
Neubauer, and Superti-Furga 2002), by predicting network interactions from gene
expression data (Zhou, Kao, and Wong 2002), and by computing networks from
protein identification experiments using databases of known interactions (Yan, Lee,
