A Shared University Research (SUR) award from IBM will help researchers at Johns Hopkins University add a finer-grained level of detail to a decade-old computational modeling project. The university’s new Center for Cardiovascular Bioinformatics and Modeling will use IBM hardware and software technology to merge gene expression and proteomics data with its existing portfolio of cell-, tissue-, and organ-level simulations of diseased and healthy hearts.
Raimond Winslow, director of the new center, said this is the second IBM SUR award related to the heart-modeling project. The first, awarded in 1999, provided an RS/6000 SP supercomputer for the computationally intense initiative. With terabytes of additional gene expression and proteomics data in the pipeline from the expanded effort, Winslow said the new round of equipment is sorely needed.
The IBM architecture includes an eServer p690 with 16 microprocessors and 64 gigabytes of memory; a seven-node eServer xSeries Linux cluster; a fibre-channel storage area network powered by a FastT Storage Server to house data in a central site that can be accessed by many computers; and a robotic Virtual Tape Server subsystem with 28 terabytes of storage capacity. Hopkins will also use IBM’s DB2 database, its DiscoveryLink data integration technology, and WebSphere internet infrastructure software.
Over the past decade, Winslow said, his lab’s work within the department of biomedical engineering has had great success in achieving “increasingly detailed levels” of cardiovascular simulation, from very large-scale models of the whole heart based on biophysical properties, down to the cellular level. Now his group is teaming up with the university’s division of cardiology to go sub-cellular.
Working on the single-cell level, researchers in the new project will measure changes in gene expression and protein expression in both diseased and healthy heart samples, and then integrate that information into large-scale computational cardiovascular models.
“The intent,” Winslow said, “is to create a very diverse and comprehensive knowledgebase of primary data and models that the entire research community can access.”
The initial 28 terabytes of storage “is enough to get us going,” but Winslow noted that as the team accumulates more and more data, particularly 2D gel images from proteomics experiments, “our storage requirements will grow rapidly.” The top short-term infrastructure priority is the development of a MAGE-OM-compliant database architecture to store the gene expression information, a project IBM is pitching in on. While other databases like ArrayExpress offer a MAGE-compliant means of depositing data, “you can’t get that software and implement in house,” Winslow said. “Our intent is to create software that the community can use.” In the meantime, the microarray data files are piling up, so the Hopkins researchers and IBM are working “frenetically” to get the database ready as soon as possible.
The new center is funded through a combination of six separate research grants from the National Heart, Lung, and Blood Institute of the NIH, Winslow said.
IBM said it supports around 50 SUR awards annually worldwide. Its last bioinformatics-related SUR award was to the Rochester Institute of Technology in January [BioInform 01-13-03].