 |
|
|
|
Overview of Features
Typical events that occur during an MCell simulation include the
release of ligand molecules from a structure (e.g., a vesicle), de novo
creation or destruction of ligand molecules (e.g., synthesis, hydrolysis,
or redox reactions), ligand diffusion within spaces defined by arbitrary
surfaces (e.g., pre- and postsynaptic membranes, or a cell membrane with
attached patch clamp micropipette), and chemical reactions undergone by
ligand and "effector" (e.g., receptor or enzyme) molecules. Ligands, effectors,
reaction mechanisms, 3D surfaces, and other simulation components are specified
using a simple programming language, or Model Description
Language (MDL), that was designed with biologically-oriented users
in mind. When a simulation is run, one or more MDL
input files are interpreted (parsed) to create the simulation objects,
and then execution begins for a specified number of iterations. Each iteration
corresponds to one Monte Carlo time-step. A wide variety of numerical
and imaging results can be output from the run,
and, in addition, simulations can be stopped and subsequently restarted
from user-specified "checkpoints". Each
time that a simulation restarts, updated information can be read from the
input MDL file(s). Checkpointing is thus a powerful and general way
to change run-time parameters such as the time-step, reaction rate constants,
and surface positions, and can also be used to split long simulations into
segments that are run sequentially.
MCell's Monte Carlo algorithms
simulate ligand diffusion using 3D
random walk movements for individual molecules. The positions of surfaces
and effector sites are mapped in space, and
"encounters" with diffusing ligand molecules
are detected at points of intersection with the ligand's motion. The final
outcome of each encounter is decided by comparing the value of a random
number to the probability of each possible outcome. Different
possible outcomes depend on the properties of the surface. For example,
at the point of intersection, the surface may be reflective, transparent,
absorptive, or occupied by an effector site with an associated chemical
reaction mechanism. Random numbers are also used to decide between
all other possible reaction mechanism transitions that might occur during
each time-step. For example, bound ligand molecules may unbind, and
effector sites may change from one defined state to another, simulating
a protein conformational change.
The numerical
accuracy of MCell's algorithms has been rigorously tested, and
because of unique optimizations the time
required for simulations does not depend on the complexity of included
surfaces. In other words, simulation of a large-scale
tissue reconstruction (e.g. hundreds of thousands of polygons) requires
about the same amount of time as simulation of a highly simplified structure.
The effective speed-up introduced by our code optimizations thus amounts
to many orders of magnitude, and can literally save months of computer
time.
|
|
