The scientists wanted to eliminate any preconceived biases in selecting the gene pairs to study, and they wanted a more nuanced picture of how particular combinations of genes affect cell function. "It's one thing to say that a double mutant is dead—dead is very clear," Weissman says. Harder to analyze but equally important are the subtler cases, where double mutations sicken a yeast colony, but less so than a single mutation would, or where they actually make a colony healthier.

Because the database currently covers only a small percentage of all yeast genes, says Harvard's Struhl, its usefulness is limited to researchers who happen to be interested in those genes. "A whole-genome analysis would be a great boon," he says.

Weissman and colleagues are working on expanding the method to the rest of the yeast genome—and then to other organisms, including humans. Meanwhile, the scientists have made the atlas freely available on the Web so that other researchers can mine the database for gene-gene interactions.

		Researchers had not suspected Rtt109's protective role because it bears no structural or chemical resemblance to the group of proteins believed to do that work, explains Jonathan Weissman. It turns out that the enzyme marks chromosomal components called histones to signal that a particular stretch of DNA has already been copied. This chemical cross-checking is essential to preventing DNA damage during replication. About the same time Weissman and colleagues published their results in early 2007, three other groups, using more conventional approaches, also reported Rtt109's role in protecting DNA. But to do so, Nevan Krogan observes, took laborious screening through the entire yeast genome. In contrast, the E-MAP database allowed the UCSF group to narrow the possibilities to just a handful of promising proteins. "We came to the same answer," says Krogan, "but in their case, it took years of work. In ours, it took months." —S.C.

	PAGE 2 OF 2	Back