Computational Structural Biology
Computer Science Division
|
|
|
The Problem
Proteins play a central role in cellular function and are ultimately
the mechanism through which many diseases have their effect.
Increasingly the design of drugs to treat such diseases is based on a
detailed understanding of protein structure and its interaction with
small drug molecules. Computational structural biology is concerned
with the modeling and computer simulation of structure, function and
dynamics of biological molecules. The NIH Research Resource in
Structural Biology brings together a multidisciplinary group of
researchers in Biophysics, Computer Science, Biochemistry and
Mathematics to develop and apply new modeling and simulation
techniques.
|
The Challenge
There are many computational challenges within the larger effort,
but computer simulation of molecular dynamics (behavior over time) is
one in which Computer Science is most closely involved. Such
simulations require a tremendous amount of computation because the
basic simulation timestep is very small compared to the time-scale of
the behaviors of interest. The challenge is to increase the
simulation rate, and to provide the possibility of "steering" the
simulations toward interesting phenomena by interacting with a
visualization of the running simulation.
|
The Approach
- To accelerate molecular dynamics simulations we are designing new
parallel algorithms and are concentrating on their high-performance
implementation on parallel computers with a potentially large number
of processors.
- To provide computational steering of molecular dynamics
simulations we
are constructing the SMD system which provides a graphical interface
to the dynamics simulation, through which we may introduce additional
restraints into the simulation to effect, for example,the extraction
of a ligand from a protein.
|
Project Highlights
- Analytical and experimental observations show that the performance of
parallel algorithms using spatial decomposition for truncated
interaction molecular dynamics simulations is more dependent on load
balance than communication efficiency on modern parallel computers.
A performance model was used to explore several decomposition methods
and to choose among them.
- The SMD system is being used to help find the exit pathway
through which Xenon can enter and exit a hydrophobic cavity
in the T4-Lysozyme.
-
Recent publications
|
Personnel
- Faculty
Jan Hermans (PI, Biochemistry and Biophysics),
Lars Nyland,
Jan Prins
- Current students: Geoff Mann (Biophysics),
Previous students: Jon Leech, James Chen, Michael Lappe
|
Funding
|
Links
|