New Program Interrogates Gene Pathways

November 13, 2003

CAMBRIDGE, Mass. — Any criminologist will tell you that witnesses, even the best intentioned, don’t always get it right. Confusion, trauma—even a smattering of white lies—inevitably distort the story to some degree, requiring multiple witnesses to corroborate each other’s testimonies. New software developed by researchers at Massachusetts Institute of Technology and Whitehead Institute for Biomedical Research promises to do for biology what any criminal investigator would do at a crime scene: cross-examine witnesses until a single, coherent account of the event emerges.

The new publicly available program, called GRAM (Genetic Regulatory Modules), incorporates two different data sets, each of which tells scientists something about gene regulatory pathways, to reveal the cell’s internal wiring system. The first data set identifies molecules called regulators that bind to genes to switch them on, the second describes the levels of the products of these genes at any given time.

“Each data set independently really isn’t sufficient to give us the exact causal relationship between the two,” says MIT computer scientist David Gifford. “But here, with this program, we’re bringing the cellular network into focus by having different perspectives of the network and then putting them together.”

Typically, researchers take the “one witness” approach when trying to map regulatory pathways in cells: They examine only the levels of a gene product called messenger RNA—the molecule that a gene produces to transport its encoded protein recipe to the cell’s protein-production machinery. By measuring these levels, researchers hope they can reverse-engineer regulatory pathways and eventually identify the regulators that originally caused the genes to produce them.

According to Whitehead scientist Richard Young, this one-data-set approach is akin to studying only a subset of the information available at a crime scene in order to determine the criminal’s original intent. The more information available to forensic experts, however, the more likely the crime will be solved.

In a paper published in the November issue of the journal Nature Biotechnology, Gifford and Young published findings in which they used GRAM to discover the effect rapamycin, a drug given to post-operative transplant patients, has on gene regulatory pathways in yeast. After treating the yeast cells with the drug, the team first identified the locations of regulators on the genome, then brought in data sets with measured mRNA levels. When they processed these two data sets together through GRAM, the program in effect corroborated the testimonies of both these “witnesses” and determined which regulators switched on which genes.

“With this model we’re able to say, ‘this regulator bound to this gene at this time in the cell, and turned the gene on to produce this message, or mRNA,’” Young says. “And we can do that for most every gene across the genome.”

Because rapamycin simulates starvation, Gifford and Young were able to see what occurs in a cell when it is deprived of nutrients. Says Gifford, “We could see how the cell reprograms various aspects of metabolism that deal with starvation. We could also see how the expression of key genes changes.”

In the near future, pharmaceutical companies may start developing therapeutics that target an entire network pathway, rather than just targeting specific molecules. Mapping the pathways, the scientists say, is the first step toward that type of drug discovery.

The next step in the study, Gifford and Young say, is to incorporate a variety of other data sets into the program to further expand its potential, an effort already under way in their labs.

Written by David Cameron.

This research was supported by the Burroughs Welcome Fund Interfaces, a National Defense Engineering and Science graduate fellowship and the National Institutes of Health.

Full citation for print versions
© Nature Biotechnology, Vol 21 No 11, pp 1337-1342
“ Computational discovery of gene modules and regulatory networks”
Authors: Ziv Bar-Joseph (1), Georg K. Gerber (1), Tong Ihn Lee (2), Nicola J. Rinaldi (2,3), Jane Y. Yoo (2), Francois Robert (2), D. Benjamin Gordon (2), Ernest Fraenkel (2), Tommi S. Jaakkola (1), Richard Young (2,3), and David Gifford (1)

(1) MIT Computer Science and Artificial Intelligence Laboratory, Cambridge, Massachusetts 02142, USA
(2) Whitehead Institute for Biomedical Research, Cambridge, Massachusetts 02142, USA
(3) MIT Department of Biology, Cambridge, Massachusetts 02142, USA


Communications and Public Affairs
Phone: 617-258-6851


Scientists Work to Break Cellular Code

Scientists Produce the Script for Life

Whitehead Institute is a world-renowned non-profit research institution dedicated to improving human health through basic biomedical research.
Wholly independent in its governance, finances, and research programs, Whitehead shares a close affiliation with Massachusetts Institute of Technology
through its faculty, who hold joint MIT appointments.

© Whitehead Institute for Biomedical Research              455 Main Street          Cambridge, MA 02142