Connecting the Dots
In many ways, biology is becoming a numbers game. The human genome contains more than three billion DNA letters, representing some 40,000 genes, which actually encode untold billions of proteins, thanks to a complicated system of enzymes that slice, dice, and otherwise modify proteins as they’re made. Detailed information on all these players, and on their counterparts in other organisms, is filling gargantuan databases around the world. The job of the Institute for Systems Biology is to draw connections between the data its researchers accumulate and all of the information they can scrounge from these databases and the scientific literature. It is a breathtakingly ambitious mission, marked by huge challenges in the gathering, storing, and crunching of data, so the institute has coupled its state-of-the-art computers via extremely high-bandwidth connections to machines at a supercomputing center in Fairbanks, AK, that has the capacity to store more than 300 terabytes of information.
One homemade software program called Cytoscape helps the researchers make sense of the data. Developed collaboratively by the Institute for Systems Biology, the Whitehead Institute for Biomedical Research in Cambridge, MA, and New York’s Memorial Sloan-Kettering Cancer Center, Cytoscape creates visual representations of systems. To the untrained eye, the program’s collection of circles connected by lines to other circles looks like some hugely complicated engineering chart that spells out the production process at a manufacturing plant. But Aderem emphasizes that without Cytoscape, the researchers would be lost. “Humans can extract huge amounts of data, but no human can juggle more than 20 parameters,” says Aderem. “By visualization, though, they can do 100,000 parameters or more.”
Sitting at his computer, Aderem opens a Cytoscape representation of yeast metabolism. Although yeast metabolism offers an exceedingly simplified model of human metabolism, the same rules that control a single-celled fungus inform how the trillions of cells in a person operate. Each circle in Cytoscape, known as a node, represents a gene or protein. “If you perturb the nodes, that will result in large changes,” Aderem notes. For humans, he says, “these are obvious drug targets.” The visualization of the system also allows scientists to predict a medicine’s side effects: if a drug interferes with a specific protein, Cytoscape shows researchers how that might have a negative effect on a connected pathway that controls such critical functions as respiration or metabolism of sugar. “This would take 15 years and billions of dollars to see in the terms of standard drug development,” says Aderem.
Indeed, researchers typically spend years studying a drug in laboratory and animal experiments before moving it into cumbersome, expensive human trials, which often fail because surprising side effects suddenly surface. Such failures can cost pharmaceutical companies hundreds of millions of dollars and patients their lives. But with a detailed map of the systems that go haywire during bad drug reactions, drug companies might one day be able to substitute a quick computer analysis for many of those costly experiments.