News from the San Diego Supercomputer Center
March 2007
Friends and colleagues,
For college basketball fans, March is known as the month of “madness” as teams vie for the national championship. For SDSC and its partners in the petascale project, March provided an opening round of competition for a major national prize in the high-performance computing community. This important competition is being conducted by the National Science Foundation and supports the procurement of a leadership class supercomputer by 2011 that will sustain over 10^15 calculations per second on some of the world's most challenging problems. During a site review held this month, SDSC and its outstanding national partners were provided the opportunity to present the most salient features of our petascale proposal and answer questions for the review committee. Many thanks to our excellent team who have worked closely since last summer on a comprehensive plan for petascale computing, science, and engineering.
Also in March, Director Fran Berman attended a unique workshop offered by the Anita Borg Institute for Women and Technology (ABIWT) and targeted to women in positions of senior leadership in national agencies, laboratories, and centers. Leadership is critical to the smooth running of SDSC, and the workshop provided an opportunity for the participants to examine their personal leadership styles with a group of peers during a two-day session. The ABIWT staff and facilitator Sabina Nawaz were excellent and the workshop was both productive and useful for the participants.
Finally, in March a meeting of the National Science Foundations's Large Resource Allocation Committee (LRAC) was held in San Diego, hosted by SDSC. The LRAC receives proposals for allocations on the largest NSF machines and assigns available cycles after substantive discussion and evaluation. We congratulate all the LRAC allocation recipients. Valued SDSC partner, the Southern California Earthquake Center (SCEC), received the largest NSF compute allocation ever--a whopping 15 million CPU hours on TeraGrid resources. The SCEC community will use this large allocation to support simulations of the realistic models required to predict the impacts of massive earthquakes. The SCEC simulations not only are compute-intensive, but data-intensive, and the SCEC community currently stores more than 150 Terabytes (150 X 10^12 bytes) of data at SDSC.
A sampling of March news and activities at SDSC follows below. We welcome your visits when you are in San Diego.
Until next month,
Fran Berman and Vijay Samalam
| SDSC Simulations Point to Causes of Parkinson’s, Alzheimer’s Diseases | |
 |  Using the power of supercomputer simulations on SDSC’s Blue Gene Data system and a Blue Gene machine at ANL, researchers are zeroing in on the causes of Parkinson’s disease, Alzheimer’s disease, rheumatoid arthritis and other diseases. In a study published in the April Federation of European Biochemical Societies (FEBS) Journal and featured on the cover, lead author Igor Tsigelny of SDSC and colleagues describe for the first time how a protein known as alpha-synuclein aggregates to form harmful ring-like pores in human membranes, the kind of damage found in Parkinson’s and Alzheimer’s patients. The researchers also found that the destructive properties of alpha-synuclein can be blocked by beta-synuclein – a finding that could lead to treatments for many debilitating diseases. |
| Lifetime Achievement Award to UCSD Librarian, SDSC Collaborator | |
 | Brian E. C. Schottlaender, University Librarian at UC San Diego and key SDSC partner in projects involving the management, stewardship and preservation of digital data, has been awarded the inaugural 2007 Ross Atkinson Lifetime Achievement Award from the Association for Library Collections and Technical Services (ALCTS). The award recognizes Schottlaender’s contributions to the profession and to ALCTS, a division of the American Library Association, where he has filled several leadership roles, including president of ALCTS. Working in close collaboration, the UCSD Libraries and SDSC are national leaders in the area of digital management, stewardship, and preservation. The many risks to digital information pose a significant threat to future generations of research, education and societal applications. To address this problem, SDSC and the UCSD Libraries are participating in a major project, Chronopolis™, which will serve as a national-scale digital preservation repository model. |
| TiVo for Your Data: SDSC Hosts Open Source Data Turbine for Streaming Data | |
 | A widely used middleware tool for gathering remote sensor data, Data Turbine, originally developed by Creare, is now freely downloadable as open source for noncommercial use. Based on extensive development work in the Network for Earthquake Engineering Simulation Cyberinfrastructure Center (NEESit) at SDSC, the middleware is now being hosted in NEESforge at SDSC. Data Turbine has been described as a shared TiVo for your data. Both local and remote users can simultaneously browse data, both slower and faster than real time as needed. Data Turbine is especially good at aggregating dissimilar data sources scattered across a network. By hiding from users the details of camera and other device drivers and network protocols, it gives a clean way for researchers to work together with multiple kinds of data. |
| Global Community Attends OSG Consortium Meeting at SDSC | |
 | More than 120 attendees from five continents gathered March 5-7 at the San Diego Supercomputer Center at UC San Diego for the Open Science Grid (OSG) Consortium All-Hands Meeting. The meeting included technical and applications sessions, as well as opportunities to help develop OSG's future plans. Founded in 2004, the OSG emerged to help the high-energy physics community deal with the avalanche of data and computing requirements anticipated when the Large Hadron Collider comes online this year. The OSG has since expanded to include many disciplines and is now a national grid computing infrastructure operated by a consortium of 14 U.S. universities and four Department of Energy (DOE) national laboratories. The TeraGrid and OSG are working to increase the interoperability of their resources and harmonize their underlying middleware, helping users use resources from both infrastructures seamlessly. TeraGrid and OSG are also collaborating in education and training. |
| Update: Rosetta Code Protein Structure Prediction Leads CASP7 Result | |
 | In the 7th annual Critical Assessment of Structure Prediction (CASP) experiment, organized to encourage improvements in computational methods, the Rosetta code prediction for a specific target protein led more than 100 entries from other groups who used a variety of different methods. To power this demanding computation, SDSC researchers participated in the largest-ever run of the Rosetta code, developed by Professor David Baker of the University of Washington. After preparing the code on SDSC’s Blue Gene Data system, the researchers were able to carry out a complete, from-scratch computation in just 3 hours on the 114 teraflops IBM Blue Gene Watson machine, something that normally takes weeks. Improving predictions of the 3-D structure of proteins is a pivotal step for the rational design of new drugs for tomorrow’s medicine chest. More information is online at the Baker Lab website. |
| SDSC Develops Grade Level Assessment Service for NSDL Web Resources | |
 | The National Science Digital Library (NSDL) offers a wide array of online educational resources. To help educators efficiently find age-appropriate resources, SDSC researchers have developed a Grade Level Assessment Service (GLAS), featured in the recent NSDL Whiteboard Report. The new value-added GLAS service associates a grade level with each word used in an NSDL online resource. SDSC researchers compiled a dictionary for grade level analysis using an iterative approach. The current assessment uses a dictionary containing 71,095 words associated with grade levels K-12. A larger extended dictionary with 4 million words is being developed which includes common misspellings and technical terms. The computationally demanding task of analyzing the NSDL content – some 50 million files totaling 7 terabytes -- uses SDSC's TeraGrid cluster. Grade level results are then published through the GLAS service on the NSDL site. The standards-based GLAS service builds on SDSC's collaboration with NSDL since 2002 to develop a Library "memory" through long-term archiving services using the SDSC Storage Resource Broker (SRB). For more information see the NSDL Whiteboard Report. |
