Dealing with reams of data

Scientists work toward unraveling gene expression in the brain

Using Web-based tools they developed to sift through reams of data, scientists from the Kennedy Krieger Institute and Johns Hopkins hope to unravel the genetics of neurological problems associated with Down syndrome, autism and lead poisoning.

Their search starts with microarrays, or so-called “gene chips,” which measure the activity of tens of thousands of genes all at once. By analyzing the pattern of gene activity in brain tissue, the scientists hope to find genes that are more or less active than normal and that may, therefore, be involved in causing problems.

On Feb. 16 at the annual meeting of the American Association for the Advancement of Science, Jonathan Pevsner, Ph.D., will demonstrate how two tools they developed, called SNOMAD and DRAGON, can be used to find the needle in the haystack of microarray data. As an example, Pevsner applied the programs to microarray data from Down syndrome.

“In some conditions, like autism, the biological cause is still unclear, but even in Down syndrome, which we know is the result of having an extra copy of chromosome 21, we don’t know exactly what genes or processes lead to the neurological changes,” says Pevsner, associate professor of neurology at Kennedy Krieger and an associate professor of neuroscience at the Johns Hopkins School of Medicine.

While it makes sense that all chromosome 21 genes would be more active than normal in Down syndrome, no one has ever proved it. In his presentation and in an upcoming issue of the journal Genomics, Pevsner will report that using microarrays (and DRAGON) showed that, indeed, as a group, chromosome 21 genes are dramatically overexpressed.

“There’s no smoking gun on chromosome 21 in our initial analysis, but further investigation might reveal specific genes that influence the severity of the condition,” says Pevsner.

In addition to dealing with the complexity that comes with receiving a mountain of data from a microarray experiment, in many cases scientists interested in answering the biological question — which genes are expressed differently — may not have the mathematical or computational expertise to analyze and interpret the results to get an answer with meaning, notes Pevsner.

“To use microarrays effectively, you have to do both the biology and the math correctly,” he says. “SNOMAD and DRAGON supplement other available analysis tools to help researchers make sense of their results. The bottom line, however, is that any result must be confirmed.”

SNOMAD, or Standardization and Normalization of Microarray Data, was developed in 2001 by a graduate student in Pevsner’s lab, in conjunction with Scott Zeger, Ph.D., chair of biostatistics at the Johns Hopkins Bloomberg School of Public Health. The online computer program processes researchers’ microarray data to search for “signal” within the “noise” of normal variation in gene expression levels, says Pevsner.

DRAGON, or Database Referencing of Array Genes Online, ties the results of an individual microarray experiment to other available information. For example, DRAGON cross-references over- and under-expressed genes in a researcher’s microarray to five online databases, identifying the genes and pulling together what is already known about their functions and roles in disease. The program can also produce visual displays of the results — graphs, charts, drawings — that the researcher can manipulate to see — really see — how the results fit together.

“Microarrays are really an exploration, and at the end of the analysis we have to decide if we believe it or not,” says Pevsner. “But even with the complexities inherent in the brain, we think microarrays can help improve understanding of neurological disorders.”