The technique could be applicable for studying a wide range of genes. Libraries of β-galactosidase genes attached to other genes are widely available; interested biologists could make bacteria or yeast cells with the enzyme's gene attached to their gene and study its expression using Xie's system.

But Xie's interest in getting a closer look at real-time gene expression goes beyond looking at genes with low activity. He studies the small but significant differences in the genetic activities of identical cells. Using conventional techniques, Xie says, "You just get average behavior." This wouldn't be a problem if genetically identical cells in identical environments behaved identically -- but, strangely, they don't. At the single-cell, single-molecule level, genetic activity is governed by randomness.

"Normally you think about gene expression or chemical reactions in general as very smooth phenomena, but it turns out they're very noisy, almost unpredictable," says van Oudenaarden, who in 2002 was one of the first scientists to provide quantitative evidence for the randomness of gene expression. "It's very apparent that there is a lot of fluctuation and variability in gene expression."

With Xie pushing the limits of resolution to single protein molecules, scientists can now do more quantitative studies of how individual cells can behave differently.