Statistical method makes finding rare genetic variants easier
By Ruth SoRelle, M.P.H.
Combining two forms of analysis into a new statistical method should make it easier to identify rare genetic variations that increase the risk of developing common diseases, said Baylor College of Medicine in Houston researchers in a report that appeared in a recent issue of the American Journal of Human Genetics.
These rare genetic variations are now believed to be the source of increased risk of a host of common health problems, including heart disease and diabetes. The search for rare variants such as these is one of the rationales for the 1,000 Genomes Project, which seeks to completely sequence the genomes of at least 1,000 people from populations around the world. BCM's Human Genome Sequencing Center is a major center of the effort.
Focus on rare variants
"Up until now in studies such as those being done genome-wide, we have concentrated on finding common variants for common diseases," said Suzanne Leal, Ph.D., professor of molecular and human genetics at BCM and the paper's senior author. "We have not been able to study the involvement of rare variants in common diseases because we did not have the statistical tools or the data."
Now she and BCM graduate student Bingshan Li have developed a method called Combined Multivariate and Collapsing Method that helps identify those rare variations that contribute to increasing the risk of developing a disease.
The genome contains many such variations in the way the genetic code is "spelled." In many instances, common variants contribute to increasing risk but only in subtle amounts. Some of the more rare variants may have an even greater effect, but identifying them in whole genome is only now taking off with the development of new sequencing technologies.
Improved analysis
This new method, which combines multivariate and collapsing analysis, gives researchers a better analysis tool, said Leal.
"This is the first paper that specifically addresses the analysis of rare variants for common diseases," said Leal.
The task is important, she said.
"Although they are individually rare, these rare variants that cause common disease are numerous and may have a stronger genetic effect than common variants," she said. "They will be more important in predicting risk of disease than common variants and may be important in drug development."
Right now, the technique can be applied in studying candidates genes for association with disease, said Li. But in the future, when the sequence data become available, it can be applied to entire genomes.
Prediciting risk
As its name indicates, the new method combines a multivariate test with a collapsing method. Both tests involve statistical analysis of multiple variables and their association with disease. The combination provides a powerful test that is robust to misclassification of functional status of the variants.
"Rare variants are important and suitable for personalized genomics," said Li. "If a person has specific rare variants, we are better able to predict risk."
Funding for this work came from the National Institutes of Health.
View the full paper..
