A Manhattan plot shows strongly associated common SNP genotypes in 4 genes—APOA5, LPL, GCKR and APOB—in patients with hypertriglyceridemia, which were found using microarrays. (Click to enlarge)

In August, Robert Hegele—director of the London Regional Genomics Center and professor of medicine and biochemistry at the University of Western Ontario—published a GWA study on hypertriglyceridemia (1), a lipid disorder characterized by high levels of blood fat. Merging technology, his team used next-generation sequencing to find rare variants that would have been missed using microarrays. After analyzing the genomes of 500 people, the researchers found that the majority of variation was not caused by common variants (directly matched to the phenotype). The researchers identified four genes with most relevance as determined in the initial study, and resequenced these genes in the same patients using next-generation methods. What they found was that these genes containing common variants contained rare variants as well.

Next-generation sequencing systems are now being used to sequence large numbers of human genomes, and researchers hope that these efforts will lead to identifying more rare variants within the population. The 1000 Genomes Project, launched in 2008, plans to create an improved map of the human genetic variation by sequencing the genomes of 1200 individuals from around the globe. The project has already released preliminary data. And dropping costs means better access to next-generation sequencing tools, which will enable even more data on rare variants to emerge.

Data: Not Enough, or Too Much?

But even with increasing sequencing capability, Brett Abrahams, assistant professor of genetics and neuroscience at Albert Einstein College of Medicine, says that there are still issues with the technology's scope. “Because rare variants are often observed in only one patient,” he says, “it becomes a challenge to be able to conclude that the variant plays a role in disease, without having any context to effectively compare it to, as one would with a common variant.”

The inherent problem with rare variants is that they are, in fact, rare. “Depending on the population, a typical GWA study is unable to look at rare variation effectively because the sample size isn't large enough,” says Abrahams. “So if you see something that looks deleterious like a truncating mutation, and you never see it again in another affected individual, on a statistical level it is impossible to draw any inferences.”

Stephen Scherer, director of the Center for Applied Genomics at Toronto's Hospital for Sick Children and professor of molecular genetics at the University of Toronto. (Click to enlarge)

According to Hegele, to adequately study rare variation would require as many as 2000–3000 participants. “What we consider to be a big GWA study using microarray technology is very small for studying rare variants,” says Hegele. “We need to have multiple people with the rare variant involved in a study to feel confident about drawing conclusions.”

This will require more efficient and cost-effective genome analysis. “I suspect there will need to be a third-generation–type breakthrough where the DNA sequence and structure is being assayed directly from the chromosome strand,” says Scherer. “I'm not complaining about the latest technologies—they are great—but there will need to be another leap forward in technology.” Such technology would lower the cost and time of sequencing to the point where studies of 2000 or more participants becomes financially feasible. But even if such capability is developed and a new library of rare variants constructed, the research community still must consider the greater question: how these variants impact disease.

Data from the Hospital for Sick Children's database showing the genome-wide distribution of structural variants. (Click to enlarge)

Clinical Relevance

“The issue isn't how we can identify these rare variants, but what can we conclude once we have determined them to be present,” says Abrahams. “The problem is compounded even further because some rare variants may actually exert effects through interaction with other variants.”

Writing recently in Cell, Mary-Claire King, a University of Washington professor of genome science, noted that there are indeed current strategies to identify rare variants in individuals (2). But she cautioned that researchers should be careful before claiming that any variant is associated with a specific disease. “Biological relevance,” she warns, “must be established before a mutation can be causally linked to a disorder.”

According to Hegele, the major research focus in the future will be to understand rare variants’ function in disease pathogenesis in order to predict responses to treatments. “If you know the pathway that is being affected, and trace it back to the genomic level, that might help researchers and physicians come to the right treatment for specific patients faster.”

The disease architecture may be such that the genetic underpinnings vary based on individuals, says Abrahams; then researchers will have no choice but to look at each case as its own entity. “But what people are hoping, is that despite the actual disease contributors being different among individuals, the pathways that are impacted will be shared, allowing for therapeutics to be broadly effective.”

Goldstein foresees third-generation sequencing greatly influencing the hunt for rare variants. “In less than five years, we will have long since had our genomes sequenced and interpreted,” he says. “We'll be able to sequence large sections of genomes quickly, so most of the genes that carry variants will be identified. What happens with that information is the only question that remains.”