Contact: Lauren Ward
412-268-7761
|
For immediate release:
October 8, 2003
|
Carnegie Mellon Researchers Receive $2.5 Million to Develop Next-Generation, Integrated Systems for Bio-Molecular Imaging and Information Discovery
PITTSBURGH—Carnegie Mellon University scientist Robert Murphy has received $2.5 million from the National Science Foundation (NSF) as part of a five-year, $9.4 million multi-institutional grant headquartered at the University of California, Santa Barbara. This grant for “Next-Generation Bio-Molecular Imaging and Information Discovery” was one of eight large grants made this year by NSF’s Information Technology Research Program (see www.engineering.ucsb.edu).
 |
 |
| Examples of the patterns displayed by different proteins within cells. These images were collected using a confocal fluorescence microscope, which can collect three-dimensional images of cells. The colors in the images represent staining for DNA (red), total protein (blue), and a specific protein (green). The name of the protein or organelle that contains this specific protein is shown above each panel. A fully automated computer system developed by Dr. Murphy's group can recognize the differences between all of these patterns with greater than 97% accuracy. Even more importantly, it can discriminate the patterns of similar proteins, such as those for the proteins giantin and gpp130, that human observers cannot distinguish. |
The ultimate goal of the project, which includes researchers from the University of California, Santa Barbara; the University of California, Berkeley; and Massachusetts Institute of Technology, is to develop new information processing technologies that will enable researchers to extract detailed information from images that depict the distribution of biological molecules within cells.
Central to the Carnegie Mellon research component is the first-ever software framework for automatically analyzing high-resolution digital images from fluorescence microscopes. Developed by Murphy, the software is able to discriminate patterns within cells that cannot be distinguished by the human eye.
“Over the past 10 years, the increased availability of sophisticated light microscope imaging systems has led to an explosion in the acquisition of digital images by scientists,” said Murphy, professor of biological sciences and biomedical engineering, and a member of the Center for Automated Learning and Discovery (CALD) in the School of Computer Science.
Inundated with such large amounts of data, scientists have recognized that automated approaches to categorizing and comparing these images are urgently needed. Murphy’s work will pave the way to fully automate the extraction of information from fluorescent images and to construct statistically sound models of the biological processes the images depict.
Understanding the location and function of proteins and other complex molecules within a cell is critical to understanding the many biological processes that cells carry out to survive. Scientists must capture and analyze images of cells to extend their understanding of subcellular structure and function.
“The key is that these images must be taken at the very highest magnification possible in order to maximize our ability to distinguish similar proteins and subtle changes in their patterns,” Murphy said.
Murphy’s software enables a new branch of proteomics, location proteomics, which describes and relates the location of proteins within cells. This work should provide a more thorough understanding of cellular processes that underlie certain diseases, such as cancer and neurodegenerative diseases. Findings made using this sophisticated acquisition and interpretation technology thus should improve methods to diagnose and intervene earlier in these disease processes.
Already, Murphy has collected and analyzed the fluorescence patterns from many proteins within a cell to create standard subcellular location features (SLFs). These SLF “fingerprints,” which form the basis of Murphy’s software, are capable of determining the subcellular location of proteins from previously unseen images.
Working with the extensive collection of SLFs, Murphy and Carnegie Mellon co-principal investigators Tom Mitchell, director of CALD; Christos Faloutsos, professor in the Department of Computer Science; and Jelena Kovacevic, professor of biomedical engineering, will investigate different subcellular features to characterize a larger number of fluorescence images for future use and comparison to newly acquired images. To accomplish this task, a portion of the grant will be used to further develop methods in data mining and pattern recognition, techniques used to discover patterns in data and to classify those patterns.
The work will be integrated with efforts at the collaborating institutions to provide a comprehensive set of tools for biological researchers. The grant also will be used to train and educate graduate students in the emerging field of computational biology, Murphy said.
###
Links:
 Robert Murphy
|
