|
Cover Story Sidebar |
|
|
Finding the Needle in a Haystack
Researcher is probing gigabytes of data by JANE M. SANDERS | |
CURRENT IMAGING TECHNOLOGIES and biopsy analyses – though they have advanced significantly – cannot slow the cancer death rate without some additional diagnostic tools because cancer is just too complex, says May D. Wang, an assistant professor focusing on bioinformatics and bioimaging research in the Wallace H. Coulter Department of Biomedical Engineering at Georgia Tech and Emory University.
|
Imaging is only useful in some cancers when the tumor is at least a half centimeter in size. And biopsy analysis is subjective regarding cancer subtype and stage. What physicians need in addition to these traditional tools is a cancer-specific molecular profile, Wang says. Such information would assist in early detection, prognosis decisions and the development of more effective therapies.
The task is daunting, but Wang is determined and confident. “With 33,000 human genes and up to one million proteins to study, that’s where my computer science and engineering training come into play,” she says. “We need to find a cure for cancer, and engineering skills fit well with this goal.”
Wang’s research team is designing a novel framework of analytical algorithms to analyze gene and proteomic data obtained from high-throughput technologies, such as DNA microarray analysis and mass spectrometry, to find cancer-specific molecular profiles and to reconstruct and study functional pathways. She is also developing a new bioinformatics system to archive, retrieve and process the molecular imaging data. And Wang is creating a real-time, human anatomy simulation tool to predict biological behavior, even at the molecular level.
“This is a long-term research program,” Wang notes. “It will be some time before we have a complete archive of all the population’s molecular profiles and clinical records, so we work at the current stage of technology using a limited sample of data.”
Wang is encouraged, however, by her team’s progress in algorithm development for identifying disease-specific molecular markers and then correlating those markers with clinical outcomes. “We’re seeing the light at the end of the tunnel,” she says. “Already, we have made important discoveries like finding reliable genetic markers for cancer.”
In one experiment, Wang and her team analyzed data from a published study of 70 prognosis-related genetic signatures derived from 25,000 genes present in 78 cancer patients. Wang used her algorithm to reduce the number of signatures to 13, effectively yielding better predictions.
Georgia Tech researchers are developing a user-friendly computing environment for real-time analysis of human anatomical 3-D image data from the Visible Human Project at the National Institutes of Health.
(300-dpi JPEG version - 814k)
Once the molecular profiles are found, the next challenge is to reconstruct the underlying biological functional pathways. Wang’s group has proposed a new modeling framework to study metabolic and signaling pathways. Wang has filed provisional patents on the molecular profiling and functional pathway modeling.
She has also developed a biological function mining system called GOMiner and a powerful Web-enhanced version called EGOMiner. The system simultaneously searches 17 databases to identify the function of individual genes or proteins. GOMiner – called such because it outlines gene ontology – can also group genes based on their functional category and indicate which genes are regulated up or down, or remain unchanged. Further, it performs statistical analyses and is integrated with other bioinformatics tools to determine cell signaling pathways and molecular structure and find related published studies.
“This is an all-in-one tool to automate the research process,” Wang explains. “We think it’s having a significant impact on the research community.”
GOMiner, developed initially for the National Cancer Institute (NCI), took only five minutes to yield the genetic information that it previously took NCI scientists six months to reconstruct, Wang notes. GOMiner, which is available to scientists on the Web at www.miblab.gatech.edu, has more than 5,000 users. EGOMiner has more capabilities and will be deployed from the MIBLab Website soon.
In a related research effort, Wang and her team are developing a user-friendly computing environment for 3-D, real-time navigation of human anatomical image data from the Visible Human Project at the National Library of Medicine, part of the National Institutes of Health (NIH). The system will also navigate through 3-D cell and molecular imaging data acquired by bionanotechnology techniques. In addition, Wang’s system will simulate biological behavior from gene, protein, and lipid-level profiling.
The project is benefiting from a recent $400,000-plus donation from Hewlett Packard of a powerful 68-CPU computing system and eight digital projectors. The system is powerful enough to handle the 55 gigabytes of Visible Human data and an exponentially growing number of gigabytes of molecular imaging and traditional medical imaging data.
“We’re creating this interactive visual representation for the new generation of medical professionals who grew up with computers, as well as some older professionals who are frustrated by computers,” she explains.
Wang’s team is working to solve several technical issues, including real-time data manipulation. Right now, one move of the 3-D tracker takes a few minutes to display, she explains. Another issue is the ability to correlate the NIH anatomical data with a real patient’s medical imaging data containing molecular-level information so doctors can reconstruct a realistic simulation useful for diagnosis and treatment. And a third issue is the goal of correlating that anatomical reconstruction with molecular-level imaging data to see what’s happening at the microscale.
Eventually, Wang hopes to make it possible for clinicians to touch and feel a patient’s internal systems from organ to cell and molecular levels in real time without invasive exploratory surgery.
To support her work, Wang recently received a Georgia Cancer Coalition Distinguished Cancer Scholar award for translational cancer bioinformatics and bioimaging research.
photo by Gary Meek
For more information, contact May Wang at 404-385-2954 or maywang@bme.gatech.edu.
Contents
Research Horizons
GT Research News
GTRI
Georgia Tech
Send questions and comments regarding these pages to
webadmin@edi.gatech.edu
Last updated: July 7, 2004
