The completion of the Human Genome Project triggered a gold rush to mine the genes behind common human traits. But despite successes such as finding genes linked to Crohn's disease and age-related macular degeneration, gene miners have returned with slim pickings in most cases. Even for highly heritable traits like height, gene hunters have dug up only a paltry 5% of responsible genes.
This has led to a debate among the miners: One camp says there are no more DNA nuggets of big effect left to find. The other says that such nuggets exist but cannot be sifted out using today's tools. Instead, whole-genome sequencing of selected people is needed to find rare genes with big effects.
A novel statistical analysis from an Australian group now weighs heavily into the first camp. The team, led by quantitative geneticist Peter Visscher of the Queensland Institute of Medical Research in Australia, argues that the tools are working just fine. The problem is that the genes have been undetectable because their individual effects are tiny. The best way to recover this genetic gold dust, the researchers say, is to stay with existing tools but increase the size of the gene-mining operation.
In an article published online 20 June in Nature Genetics, Visscher and colleagues report that this should be doable. They used an approach called statistical estimation, which helps assess whether common gene markers known as single-nucleotide polymorphisms (SNPs)—single letters in the DNA alphabet—could account for height variation in a British population. Like previous gene miners, the researchers carried out a genomewide association study, which examines the correlation between SNPs and individuals' height. But unlike past studies, they did not test each SNP individually—an approach that has seen very few SNPs clear the high hurdle of statistical significance. Rather, they tested 300,000 SNPs simultaneously in each member of the population.
The team found that when tested in aggregate, the common SNPs could account for 45% of the height variation in their population. Given that height is 80% heritable, this represents more than half of the heritable component. "So we can exclude the notion that rare, untagged genes of large effect explain the variance," says Visscher.
Visscher says his work "implies that most 'missing' variation is not going to be found by sequencing lots of people" but by scaling up current gene chip-based studies in larger populations. The approach, he adds, could help researchers identify a stronger genetic basis for diabetes, heart disease, and schizophrenia.
Kenneth Weiss, a geneticist at Penn State University, University Park, is not surprised by the findings. "The heritability has not been missing at all; it's been right out there for all to see for a century or more," he says. "It's an elegant statistical approach based on real data rather than theoretical models that should help lead to well-grounded conclusions," adds computational geneticist Shaun Purcell of Massachusetts General Hospital in Boston, who, as part of the International Schizophrenia Consortium, headed a so-called genomewide association study to search for genes associated with schizophrenia.
Other geneticists are more circumspect. "I suspect that this paper on its own will not completely settle in many people's minds how much 'missing heritability' is explained by common variation," says Joel Hirschhorn, a geneticist at the Broad Institute of the Massachusetts Institute of Technology and Harvard University who carried out one of the studies of height. "Nevertheless, it's an interesting and significant step forward, and it will be important to see whether larger genomewide association studies confirm the predictions of this paper."