Research Notes & Highlights, 2006

The National Resource for Biomedical Supercomputing

Twenty Years of Leadership Renewed for Five More Years


Spatially realistic cellular modeling centers on realistic three-dimensional cellular models to simulate the movements and reactions of molecules within and between cells, using MCell, DReAMM and PSC_DX software developed at the NRBSC.

In October, PSC received $8.5 million from the National Institutes of Health to renew its program in biomedical supercomputing, renamed last year as the National Resource for Biomedical Supercomputing. Through NRBSC, PSC scientists pursue research in the life sciences and foster exchange nationwide among experts in computational science and biomedicine.

Established in 1987, PSC’s biomedical supercomputing program was the first such program in the country external to NIH. Along with core research, NRBSC develops collaborations with biomedical researchers at many centers around the country and provides computational resources, outreach and training. The current award, from NIH’s National Center for Research Resources (NCRR), renews NRBSC for another five years.

VISUALIZATON: Structural Biology

Structural biology focuses on the development of computational tools used to determine the structure of proteins from their amino-acid sequence and also the development of quantum-mechanical simulation methods for biomolecules such as enzymes.

“This grant is part of NCRR’s ongoing commitment to bring together leading-edge computational resources and experts in computing with experts in biology and medicine,” said Ralph Roskies, PSC co-scientific director. “A great deal of important biomedical work over the last decade owes thanks to NIH support for this program,” said PSC senior scientist Joel Stiles, scientific director of NRBSC. “We’ve developed computational tools in simulation and visualization that are helping scientists nationwide.”

More information:

Core Research

The renewal award supports NRBSC’s research in three core areas: spatially realistic cellular modeling, large-scale volumetric visualization and analysis, and computational structural biology.

“Our long-term vision,” said Stiles, “includes integration of these areas to enable multiscale modeling of molecules, cells and tissues, with a substantial future impact on human health care.”

Training & Resources

VISUALIZATON: a protein from a small flowering plant named arabidopsis

Pittsburgh Supercomputing Center Workshops (2005-2006)
Introduction to the cray Xt3
Bioinformatics (for minority-serving institutions)
Developing bioinformatics programs
Computational methods for spatially-realistic
Microphysiological simulations
Nucleic Acid and protein sequence Analysis
Computational biophysics

A workshop underway in the PSC Computer Training Center, dedicated this year as the David W. Deerfield II Training Center, equipped with 30 “dual-boot” workstations and a projector for overhead display of the instructor’s desktop.

NRBSC training activities reach hundreds of scientists each year. Since its inception, NRBSC has provided access to computing resources for more than 1,200 biomedical research projects involving more than 3,500 researchers at 245 research institutions in 46 states and two territories. Among these are several projects featured in this booklet (p. 18 & 22).

NRBSC workshops on computational biology have trained more than 3,300 researchers in the use of highperformance computing for biomedical research, in such areas as spatially-realistic cell modeling, volumetric data visualization and analysis, protein and DNA structure, genome sequence analysis and biological fluid dynamics.

This year a Bioengineering & Bioinformatics Summer Institute program funded jointly by NSF and NIH was renewed for three more years. NRBSC participates with the University of Pittsburgh, Carnegie Mellon and Duquesne University in this 10-week summer intensive that trains promising college students for research in computational biology-related fields. NRBSC director Joel Stiles serves on the core faculty and other NRBSC scientists act as research mentors.

Networking the Future

One of the leading resources in the world for network know-how

PSC’s Advanced Networking group is one of the leading resources in the world for knowledge about networking. Through 3ROX (Three Rivers Optical Exchange), a high-speed network hub, they provide high-performance networking for research and education. Their research on network performance and analysis — in previous projects such as Web100 and current work with the NPAD diagnostic server — has created valuable tools for improving network performance nationally.

More information:

National Transit Rail

LOGO: National LambdaRail

Through National Lambda Rail, PSC is participating with four other NLR members in a new project, the National TransitRail Project, to provide an intelligently managed nationwide peering and transit program. Through NLR’s national network fiber infrastructure, the NTR project will work to reduce the number of “hops” required for data to get to its destination.

“NTR represents a new type of service,” says PSC network director Wendy Huntoon. “It is direct peering with content providers and internet service providers. The long-term benefit is to provide a better-performing and more cost-effective link to resources than current network connections.”

ILLUSTRATION: schematic of 3ROX network
3ROX Members Network Connections
Carnegie Mellon University, Norfolk State University, University of Pittsburgh, Pennsylvania State University, West Virginia University, Pittsburgh Public Schools, Woodland Hills School District, Intermediate Unit One
National Research Networks
Abilene—2.4 Gbps, TeraGrid Extensible Backplane Network—30 Gbps.
Government Laboratory
The National Energy Technology Laboratory
National Commodity Internet Networks
Global Crossing&8212;1Gbps; Sprint&8212;1Gbps. (peering) PITX
Comcast, Westinghouse Electric Co.
Pittsburgh Local Exchange Network
Comcast, TelCove
Computer Emergency Response Team
* NOTE: Gbps: a billion (Gtga) bits per second.


Through 3ROX, a high-speed network hub that serves Carnegie Mellon, Penn State, the University of Pittsburgh, West Virginia University, Norfolk State University, the Pittsburgh Public Schools and Woodland Hills School District, PSC provides advanced network resources for education and research. 3ROX connects the universities and PPS to Abilene, a high-performance network linking more than 250 U.S. universities and research organizations.

3ROX News
This year Norfolk State University became the newest 3ROX member, as 3ROX expanded its network aggregation in the mid-Atlantic region to include Pennsylvania, West Virginia and Virginia. “We’re pleased,” says PSC network director Wendy Huntoon, “to extend services to one of the historically black colleges and universities.” Norfolk State expects to work with other HBCU’s in their area to share their Abilene connection through 3ROX.

3ROX this year also added Intermediate Unit One and Woodland Hills School District in the east suburbs of Pittsburgh. As a result, any university-based resources available to the school district, such as distance learning or databases, will have better performance. Penn State expanded its network capacity by connecting to National Lambda Rail, offering up to 32 “lightpaths” — four are now active — via NLR’s infrastructure, increasing its overall bandwidth from 1.2 to 40 gigabytes per second. “With the success of this completion,” says Huntoon, “3ROX now maintains an advanced three-way network between Carnegie Mellon, the University of Pittsburgh and Penn State.”

Narrowing the Wizard Gap

The Internet is extremely robust because Internet protocols include recovery procedures that silently selfcorrect network failures. These procedures hide network problems at the cost of reducing performance. Within a local network, this slowdown is likely to be trivial, but can lead to unacceptably slow wide-area traffic — because the increased network round-trip time multiplies the delay. The self-correction also masks the flaws, making it very difficult to pinpoint the ultimate causes of the reduced performance. With the NPADd (Network Path and Application Diagnostics) Project, PSC network engineers and the National Center for Atmospheric Research have developed diagnostic tools — based on tools PSC developed in a project called Web100 — that analyze and determine the nature of any observed path failures.

The primary audience for NPAD is data-intensive scientific users. “The goal,” says PSC engineer Matt Mathis, “is to easily — and in many cases automatically — diagnose problems that impede performance at this scale.” The diagnostics (called pathdiag) are run through a web interface. With NPAD deployed on a local network, a user can with a couple clicks test the path to a client for flaws, determine whether it will support a long fast flow, and make suggestions how to fix any observed problems.

NPAD, says Mathis, can substantially narrow the “wizard gap” — the gap in available network performance between network wizards and typical users. “Several network researchers have demonstrated 40 gigabits per second under a variety of environments, but this does little to help typical domain scientists with observed median performances of only three megabits per second. The difference is a factor of 10,000, and most of the problems that cause half this gap — the first factor of 100 — can be completely and automatically diagnosed by pathdiag.”