Then researchers could work backward to figure out which genes direct progenitor cells to differentiate into particular types of neurons, says Stephen Crews at the University of North Carolina at Chapel Hill. Crews has created a two-dimensional atlas of gene expression for more than a hundred genes involved in the development of Drosophila midline CNS. Because transcription factors are largely conserved throughout evolution, Crews expects the three-dimensional fly atlas will help in understanding development of the nervous systems of higher animals. "Having a searchable atlas of hundreds of genes that all labs could have access to would really be a huge step forward," he says.

To create the first draft of the atlas, Doe and Layden used fluorescent antibodies to label neuron-specific transcription factors in late-developmental Drosophila embryos. Then, using confocal microscopy, they produced two-dimensional images—optical "slices"—of the resulting three-dimensional patterns of fluorescence. Doe and Layden are creating stacks of these slices for each of the more than 200 transcription-factor genes identified to date.

Myers and his team take these stacks and produce a computer-generated model for each gene that resembles a constellation—glowing points of color in a black three-dimensional space. The last step in building the atlas involves overlaying these patterns on one another. To do this, Myers' team is writing software that aids in the recognition of the patterns created by the glowing nuclei of each neuron. "We're learning to identify these nuclei the way astronomers identify stars in the sky," Myers says. "How do you find the North Star? You look for the pattern created by the stars that make up the Big Dipper. In the same way, we are looking for invariant patterns."

Making the atlas a searchable one will be the next challenge, as current search technology is limited to one dimension. "We can easily do a Google search using the name of a person," says Doe. "What we want to do would be like searching Google for the shape of a person's head." According to Myers, no one has yet accomplished this next-level search technology. "We'll learn a lot of biology by doing this," he says, "but we will also learn a lot about exploring and understanding three-dimensional space."

Photos: John Clark

	PAGE 2 OF 2	Back