Biomedical Engineering Reference
In-Depth Information
5.3 Molecular Models
Many diseases have molecular causes. Therefore knowledge about the genetic
make-up of the human body, the molecular components on all levels along the line
of the flow of biological information, as well as their behavior and interrelations,
need to be integrated in consistent approaches to understand the underlying
mechanisms of diseases. As only a limited number of diseases are monocausal, the
challenge now is the understanding of the high number of multifactorial diseases.
Molecular models address individual pathways including information from
different layers of biological information such as metabolic and signal transduction
pathways and cell-cycle regulation among others. With the new technologies
already available for research purposes it is expected that in the near future these
technologies will also enter clinical practice and will be increasingly used for
patient health care. The genome is the basis, but depth and complexity will
increase tremendously with integration of new data from transcriptomics, pro-
teomics, metabolomics, and the other 'omics' analyses. Still there is a lack of
knowledge about how these components in the system interact and depend on each
other and how control and regulation mechanisms achieve stability and robustness
of the system. This still existing lack of knowledge is a limiting factor for full
implementation of molecular models in the medical field. But world-wide research
efforts in this area continuously complement existing knowledge. The integration
of Monte Carlo approaches in molecular models is currently established to fill the
existing gaps in knowledge and use simulations and predictions to account for
missing experimental data [
51
].
PyBios is a currently implemented model approach using an object-oriented
modeling environment (
pybios.molgen.mpg.de
). It is based on interacting
molecular models representing different cells or tissues integrating the patient's
personal genome and other data. Individual models describe different objects
representing the biological modules within the complex biological network acting
in the cell in interacting compartments. These objects might represent different
characteristics; for example, one object could describe the Ras protein representing
one set of functions, whereas an object representing the mutant Ras protein shows
altered functions resulting from the mutation. The PyBios system uses a Monte
Carlo approach to simulate and predict the elements that cannot be measured yet.
In a current pilot project (2010-2013) on melanoma patients in Germany
(TREAT20) the PyBios modeling approach is used for the first time in a clinical
setting [
52
]. The cancer model used in TREAT20 is based on the publications by
Hanahan and Weinberg 2001 (with an update in 2011) [
53
,
54
] representing the
signalling pathways relevant in cancer. Tumor tissue collected from each patient is
sequenced for the genome and transcriptome, respectively, exome data. These data
are the basis for establishing an individual patient model that is later combined
with a drug database to predict a possible drug response (Fig.
1
). In addition, with
this approach drugs that are unusual in clinical practice or even new drugs can be
tested for their possible efficiency with these 'virtual patients'. The results from the
