Computational Biology Resource of the Center for Structural
Biology boasts a 64-processor Silicon Graphics Origin supercomputer.
The system is key to determining, analyzing, and modeling biomolecular
structure, says Jarrod Smith, the resources director.
aids structure analysis, modeling
by Mary Beth
Providing support for the specialized needs of the growing population
of structural biologists on campus, the Computational Biology Resource
is one of four core technological facets of Vanderbilts Center
for Structural Biology. The resource provides outreach to scientists
in other disciplines, as well, on projects that cross over into
the structural biology field. In fact, you could say that the mission
of the resource, in general, is to promote thinking on a molecular
level, says director Jarrod Smith.
Smith and Walter Chazin, director of the Center, have spent the
past two years building the Computational Biology Resource. Though
the resource resembles a traditional core facility, it differs in
a fundamental way. The facility provides the resources the
hardware, software, and expertise to use them but the investigator
needing the services provides the manpower to come and actually
do the work.
The resource assists with traditional bioinformatics needs, such
as deciding how to use information once it is collected. But the
bulk of what they do, according to Smith, falls more into one of
the fringe-definitions of bioinformatics.
Our piece of bioinformatics in structural biology would be
determining, analyzing, and modeling biomolecular structure,
To that end, the resource boasts an impressive collection of high-end
Currently, weve got a 64-processor Silicon Graphics
Origin supercomputer, as well as a 32-processor Linux cluster, for
a total of 96 CPUs dedicated to running biomolecular simulations,
biomolecular analysis tools, and ab-initio calculations for small
molecules, Smith says.
Even though the VAMPIRE computer cluster has more CPUs, the bandwidth
on this parallel system is greater, and the latency the amount
of time it takes for a packet of information to travel between two
processors is very low. These qualities are critical to many
algorithms used in computational biology.
Thats what makes a supercomputer to us, Smith
says. Its not only how fast the processors are, its
how tightly coupled they are and how you can use the thing as a
whole. The Origin has 32 gigabytes of shared memory; all the processors
see that memory and they all see it with the same bandwidth.
The facility also offers what Smith calls a visualization
lab. At this time, there are four high-end Silicon Graphics
Octane workstations, specifically designed for real-time, 3-D graphics
This is where people go to bring their molecules up into
the computer, rotate them around, and interact with them,
he says. We also have the capability to use stereoscopic glasses,
so you can actually see the object in three dimensions. Thats
an important tool for researchers doing modeling and drug design,
or for X-ray crystallographers needing to put atoms into electron
The software packages available through the resource are listed
on their Web page (http://structbio.vanderbilt.edu/ comp/), though
the list grows faster than they are able to update the site. Its
a good idea to ask if you dont see the one youre interested
in, Smith says. Each package comes with instructions on how to install
and begin, plus links to manuals and customized user hints. For
some of the packages, Smith has written tutorials for more advanced
interactions with the software.
As far as infrastructure goes, the resource has several servers
that handle Web, database, software, and file serving tasks, and
they have just installed a two terabyte RAID (redundant array of
inexpensive disks) array that dramatically increases storage capacity.
The RAID device does its work in the background, says
Smith. The advantage of that is you can get massive amounts
of storage in one, easy to maintain system. And there is a redundancy
built in, so you could lose a disk and nobody would know except
The two terabytes of space thats 2000 gigabytes
are necessary to handle the data collected at the NMR Center, the
simulations that are run on the supercomputers, and the data brought
back by the X-ray crystallographers from the synchrotron.
Smith and his staff, which currently consists of two system administrators,
are available as consultants to share their expertise in how to
get started and how to solve particular problems. Ultimately, they
plan to offer workshops in addition to one-on-one interactions.
The goal is that over time the required knowledge and skills
to take advantage of these technologies will start to percolate
through the community, Smith says.
Smith got his Ph.D. at the Scripps Research Institute, jointly
advised by Chazin, an NMR spectroscopist, and David Case, a computational
My background is in NMR structure determination, but most
of the work I did was in the computational methods for resolving
the structures, he says. This job really takes advantage
of my skills. Plus, I enjoy helping people solve problems, and thats
perhaps the most important aspect of the job.