By Laura Buenning

Consider identical twins, born with precisely the same DNA and raised in the same environment. As teenagers, one develops allergic rhinitis, or hay fever; the other does not.

A partial explanation can be found in how our immune systems evolve as we age. Scientists have long observed that early exposure to unsanitary, microbe-rich environments strengthens our immune systems.

“But it’s not just a matter of exposure,” says Mike Langston, a UT Knoxville biomathematician and computer scientist who builds powerful computational tools to decipher meaning from biological data.

“Mammalian immune systems are subject to a degree of randomization as we grow older. So the allergic responses of monozygotic (identical) twins will frequently diverge as they mature, which is a good thing if you want at least one to survive.”

Langston and a team of geneticists from Sweden, Norway, and elsewhere in Europe have the ideal data set for studying the biology behind developing immunity. Scandinavia is well known for tracking the progress of identical twins and for its specific and well-timed rhinitis allergens, most notably grass and birch pollen. In many allergic diseases, the biochemical pathways to recognizing allergens—pollen, dust mites, mold—are fairly well understood and seem to work well.

“Recognizing what is friend or foe is often not the issue. Instead, the feedback mechanisms that shut down a reaction might be broken. So we just keep on producing histamine, and inflammation gets worse and worse,” Langston says.

Even when they know the genes and gene products key to the relevant biochemical pathway, the presence of multiple coding sequences for a single amino acid can make it difficult to halt the assault.

“But there is plenty of room for optimism,” Langston says. “If we know of a particular problem protein, we might be able to target it, knock out the gene that produces it, and eliminate the problem.”

Radiation and Immune System Repair

Langston’s work with geneticists at Oak Ridge National Laboratory studying the health effects of low-dose radiation takes a different tack. Here, instead of a broken immune response, the focus is on immune system reaction and repair. Langston recently participated in a meeting on the subject in Finland, a country that invests substantial resources to study the potential effects of low-dose radiation on human health.

“When the Chernobyl accident occurred (April 26, 1986), prevailing winds made northern Finland one of the first places to detect that something was wrong and to tell the world about it. In relative terms, many Laplanders received huge exposures,” Langston says. “After all, snowmelt, from snow exposed to fallout, provides drinking water. And reindeer, feeding on lichen exposed to fallout, provide meat.

“We understand the risks of high-dose exposures fairly well. For low doses, we’re not so sure. Some believe all doses carry risk; others believe low-dose exposures may even have a positive effect on living organisms.

To study the problem, Langston and his ORNL colleagues irradiate genetically identical mice raised in a sterile environment, then extract spleens and other tissues and collect data on gene activity from these and a control group of mice receiving no radiation.

“Like the liver, the spleen filters out damaged cells. It can quickly show that an organism has been challenged,” Langston says.

Identifying gene activity can pinpoint which genes are turned on or turned off before and after the radiation insult—no small task, considering the number of genes and possible combinations involved.

Biological Pathways

This is where Langston’s team and its computing tools come into play.

“Suppose we have 30,000 genes in an organism,” Langston says. “Whether we measure activity by sampling across strains, stimuli, or even at different times and levels of pathogen exposures, for each measurement we can look at every possible pair of genes—some 900 million—and ask whether these two genes react in similar ways.

“Speaking simply, if we have a pair of genes that turn on and off many times and are always on or off together, then we begin to suspect that these genes are somehow connected in the same biological pathway. From a computational standpoint, we would say they are highly correlated.”
Of course, Langston says, “30,000 genes can’t all be communicating with each other at the same time. Nevertheless, depending on many factors including tissue type and stimulus, some surely are, and that’s where we try to extract biological meaning.”

Despite the sizes involved, determining gene-to-gene correlations is actually easy, relative to some other steps in Langston’s repertoire. Each tool is one step in a pipeline. Some of the most important require novel mathematical methods; many are incredibly difficult, requiring a lot of time even on our best supercomputers.

Scientist as Middleman

“The biologist (or chemist or physicist) generates huge volumes of raw data. We step in and run them through our graph algorithms pipeline to reduce the data’s dimensions. Then we deliver small and highly distilled data sets that appear to possess biological relevance, back to the biologist, who studies them to see if they make sense,” Langston says.

“Using the previous example with genes, we may present the biologist with only tens of genes to examine closely, not the original 30,000.”

Of course these sets are merely products of computation; depending on data quality, they may be nothing more than noise.

“If the results do not make sense, we go back and start over. It is not uncommon to find that some previously unrecognized factor biased the original data. But no one realized that there was a problem until we ran the data through our pipeline,” he says.

Personalized Medicine

Piece by piece, genetic studies refine our understanding of healthy cell communication. Immune system response is a good example. Langston envisions a future in which health care for each of us could be as simple as one cheek swab, followed by automated analysis. By the time we walk back to the doctor’s office, he or she will have prepared both a diagnosis and our own personal treatment regimen.

Tags: Computing • DNA • Evolution • Mike Langston