Finding patterns
Shirley Liu, PhD, is helping DFCI researchers understand gene activity.
In another area, Cheng Li, PhD, is giving a boost to microarray technology, which samples the activity of thousands of genes in normal or diseased cells. Thanks to Li's efforts, scientists now have a handy way to make sense of the complicated grids produced by such technology. Li and his former advisor, Wing Hung Wong, recently completed work on the DNA-Chip Analyzer (dChip), a software system for biologists and physician-researchers with limited training in computation or statistics. Available on the Web, the system enables them to find patterns of gene activity that distinguish cancer cells from normal ones and help identify different types of tumors.
For all their exploratory power, microarray techniques cover just 1—2 percent of the human genome, the library of genes in people's cells. They focus on the "coding" portions of genes — segments that spell out the ingredients for building proteins. Left out of this survey are the "controlling" regions that determine when, and under what conditions, genes are activated and deactivated. If the coding region holds the recipes for proteins, the controlling region is the order slip that prompts the cell to make them. Because the latter are spread out along chromosomes, finding and studying them poses a considerable challenge.
"Our goal is to give researchers a tool for finding the sites where key proteins bind, delivering a message to switch on or off".
— Xiaole (Shirley) Liu, PhD
Enter the work of Xiaole (Shirley) Liu, PhD, who has devised an algorithm for locating them. "Our goal is to give researchers a tool for finding the sites where key proteins bind, delivering a message to switch on or off," Liu says. As these projects suggest, biostatistics promises to be woven ever more tightly into biomedical research as data grows in volume and complexity. Though faster computers will inevitably be required, their presence will render even more crucial the skill of people trained in unlocking patterns of meaning from seemingly impregnable walls of data.
"Biostatistics will, if anything, become more conceptually demanding as science advances and new areas of cell life come under investigation," Harrington says. "In this department, our tradition is to anticipate, rather than play catch-up with, the direction of cancer science. We'll be looking to support that kind of thinking in the years ahead."
