Molecular Simulation and Systems Biology - Systems Biology: Networks, Models, and Applications - page 67

These coarse-grained particles interact with an effective potential
that cannot be derived directly from basic physical principles but
can be inferred by observing the results of more detailed simulations.
The use of atomistic molecular simulation to develop potentials
for coarse-grained models is a standard practice in the simulation of
nonbiological polymers [6]. Researchers are beginning to use coarse-
grained models for biological simulations of large systems, such as
lipid membranes [7].
The second way that molecular simulation methods can be used in
systems biology is to add more detail to a simpler model. Simulations
of wild-type and mutant forms of a protein can give a detailed,
atomistic picture of how the mutation affects the protein's biological
function. Simulations of a protein-protein binding event under dif-
ferent conditions (ionic strength, pH) may give insights into how the
process will occur in different cellular compartments or environments.
Comparative simulations of two homologous proteins can provide
insights into differences or similarities of function. In all of these cases,
the simulation results should always be verified by quantitative,
and ideally predictive, comparison with experimental data. With a
judicious choice of methods and their careful application, however,
molecular simulation is an important tool for gaining a complete
picture of biological processes—the central goal of systems biology.
THE ROLE OF HIGH-PERFORMANCE COMPUTING
Almost from the very beginning of electronic computing, advances
in the scale of molecular dynamics simulations have marched in step
with advances in computing performance (figure 3.2). At the same
time, both the methods and models used in carrying out the calcula-
tions have advanced as well. Often, the availability of increased
computational capability has meant that more realistic models can be
employed. For example, the first molecular dynamics simulations of
biomolecules were performed without inclusion of solvent, while now
simulations of proteins with explicit representation of solvent mole-
cules as part of the calculation are routine (to the point where most of
the computer time goes to simulating water rather than protein) (see
table 3.1, page 95).
Advances in high-performance computing capabilities can be char-
acterized by their ability to increase capability or capacity . Capability
refers to the ability to apply computational resources to a single prob-
lem. Large shared memory computers and clusters equipped with very
high performance interconnects are examples of systems that enable
capability calculations. There are many problems, particularly in the
life sciences, that involve performing many independent calculations,
such as large numbers of BLAST queries, or trying to screen a library