|Mountains of data:
A three-dimensional C. elegans gene expression map
Edward R. Winstead
September 28, 2001
The sequencing of the soil worm C. elegans three years ago yielded some 19,000 genes, most of which had no known function. Using DNA microarrays, researchers have predicted functions for some sequences based on similar expression patterns between known and unknown genes. Taking this strategy to another level, researchers have now created a three-dimensional C. elegans gene expression map by assembling and analyzing data from hundreds of microarray experiments.
The map has 44 'gene mountains,' or clusters of co-regulated genes. Some clusters were formed of genes expressed in similar tissue, such as muscle or neuronal; others were formed of genes that carry out similar cellular functions, such as ribosomal genes or collagens. The complete map is a mountain range showing gene expression clusters for 17,661 C. elegans genes (93 percent of the genome).
A team led by Stuart K. Kim, of Stanford University Medical School in California, developed the map. They assembled and analyzed results from 553 microarray experiments by 30 different laboratories, creating the first large compendium of C. elegans expression data. The data were generated for normal and mutant worms during a variety of developmental stages and growth conditions.
The map is intended to be a tool for discovering gene function on a genomic scale. In their initial experiments, the researchers used the map to identify genes that are CO-regulated with known sets of genes, which allowed them to make predictions about the worm's biology.
"We studied the genes in each mountain to find patterns suggesting the underlying biological property for that group of genes," Kim and colleagues write in a recent issue of Science. "Overall, we were able to infer a potential physiological importance for 30 of the 44 mountains by showing that specific mountains were enriched for particular sets of genes."
As expected, the map revealed more detailed information than individual microarray experiments did. For example, expression studies had identified a set of 756 genes with apparently similar functions; when arranged on the map, these genes were subdivided into four mountains with distinct biological roles.
In addition, the researchers found that the position of a gene within a mountain could reveal information about its function: "We frequently observed that genes with similar function were placed close to each other in a section of one mountain."
Kim's team used a computer program called VxInsight to visualize the spatial distribution of the genes. The software placed genes with highly similar expression profiles in close proximity on a two-dimensional scatter plot. To add a further visual cue, the scatter plot was converted into a three-dimensional terrain map, with the z-axis denoting the density of genes within an area.
. . .