Christoph Lange, assistant professor of biostatistics, and Nan Laird, professor of biostatistics, have developed a method for finding the genetic underpinnings of all kinds of diseases characterized by variation in quantitative traits, and have created an accompanying software package called PBAT. The package builds upon an earlier program called FBAT, developed by Laird and Xin Xu, assistant professor of genetic epidemiology in the Department of Environmental Health.

When sorting through reams of genetic data, researchers have a number of analysis tools from which they can choose. One approach is called family-based association studies, which compare the genes of close relatives. The goal is to see what gene variations the family members have in common and which ones they don’t share, and then to determine if the presence or absence of a gene is linked to a specific disease.

The PBAT package provides tools to plan and analyze data from family-based association studies with quantitative traits, as well as tools to screen large numbers of genetic markers for associations to disease.

Lange and Laird "are among the world leaders in novel statistical methods for genetic association testing," said Scott Weiss, director of respiratory, environmental and genetic epidemiology at the Channing Laboratory at Brigham and Women’s Hospital, HMS professor of medicine, and HSPH Professor in the Department of Environmental Health. Weiss and his colleagues have been collaborating with Lange to root out genetic factors contributing to childhood asthma and to early-onset chronic obstructive pulmonary disease.

A particular strength of PBAT is a two-step method of identifying and testing the most promising genes in a way that overcomes what statisticians call the "multiple-comparison problem," where the sheer amount of data points yields too many possible candidates. In a nutshell, PBAT shrinks down the number of viable candidates better than other statistical methods.

Lange said that 7,000 people have downloaded PBAT in the two years it has been widely available. He and his colleagues have taught the methods in workshops around the world, including Sweden, Mexico, France, and, most recently, in a three-day short course at HSPH. The FBAT and PBAT software is freely available at http://www.biostat.harvard.edu/~fbat/default.html and at http://www.biostat.harvard.edu/~clange/default.htm. A users group will be forming soon. The PBAT software may be incorporated into two popular commercial statistics packages.

In addition to its ability to whittle down gene candidates for a disease, another appealing aspect of FBAT and PBAT is that it can aid family-based association studies without needing information about parents. For example, scientists exploring Alzheimer’s disease cannot ask the parents of a patient for a DNA sample because the disease usually manifests in patients late in life and their parents are often already deceased. FBAT and PBAT are flexible enough to use siblings as a proxy for parents. "The ideal study is to have every pair of parents and the offspring," Laird said, "but we can do pretty well with an affected child and an unaffected sibling, especially with common disorders."

Even after PBAT navigates the daunting statistical challenges, researchers have a long road ahead. After all, identifying genetic variations associated with disease is only the first step. Researchers must work out the biological mechanisms in test tubes and animal models, determine how much risk a particular version of a gene confers on people who carry it, establish how different associated genes interact to cause disease, look for other contributing factors, evaluate whether the gene variants predict the effectiveness of different treatments, and ultimately develop new therapies.

--CCM