Biomedical Engineering Reference
In-Depth Information
minutes. Neural network systems typically provide a library of predefined models that the user can
incorporate in a neural network by connecting icons graphically instead of making extensive use of
mathematical equations. Like using a high-level programming language, there is no need to develop
or even fully understand low-level neuron model operation in these systems to create functional
classifiers.
Even if a general-purpose language is used to develop a simulation, there are numerous reasons for
going through the time and hassle of developing a model of a real-world system. Simulations allow
conditions in the real world to be evaluated in compressed or expanded time and under a variety of
conditions that would be too dangerous, too time-consuming, occur too infrequently, or that would
otherwise be impractical in the real world. Instead of taking days or weeks to set up and run a series
of biological experiments on the population dynamics of yeast under a variety of environmental
conditions, the effect of, for example, an increase in temperature, can be explored in a few minutes
through a simulation.
Common uses of modeling and simulation include predicting the course and results of certain actions,
and exploring the changes in outcome that result when actions are modified. Several bioinformatics
R&D groups are focused on developing simulation-based systems to determine, for example, if a
candidate molecule for a new drug will exhibit toxicity in patients before money is invested in actually
synthesizing the drug. In this regard, simulation is a means of identifying problem areas and
verifying that all variables are known before construction of the drug development facility is begun.
As an analysis tool, simulations help explain why certain events occur, where there are inefficiencies,
and whether specific modifications in the system will compensate for or remove these inefficiencies.
As listed in
Table 9-1
, the range of possible applications of modeling and simulation in bioinformatics
is extensive. These applications range from understanding basic metabolic pathways to exploring
genetic drift. One of the most promising application areas of modeling and simulation in
bioinformatics—and the most heavily funded—is as a facilitator of drug discovery, which in turn
depends on modeling and simulating protein structure and function. Given the exponentially
increasing rate at which models of proteins are being added to the Protein Data Bank (PDB),
modeling and simulation of proteins and their interaction with other molecules are the most
promising means in our lifetimes of linking protein sequence, structure, function, and expression,
with the clinical relevance of the proteome.
Table 9-1. A Sample of the Applications of Modelingand Simulation in
Bioinformatics.
Clinical What-If Analysis
Drug Discovery and Development
Experimental Toxicology
Exploring Genetic Drift
Exploring Molecular Mechanisms of Action
Personal Health Prediction
Drug Efficacy Prediction
Drug Side-Effects Prediction
Gene Expression Prediction
Protein Folding Prediction
Protein Function Prediction
Protein Structure Prediction
