A Height Gene? One for Smarts? Don’t Bet On It; Any given gene is likely to be unimportant, large-scale studies have found

A Height Gene? One for Smarts? Don’t Bet On It

Jan. 31, 2014 9:19 p.m. ET

As I skimmed my emails one morning—mostly Viagra ads and news of the three lotteries that I had won during the night—one stopped me in my tracks. “Is overeating in your genes? Take an online test.” I was curious—not about whether my genes prompted me to pig out, but about how the company’s test was supposed to determine that.

As it turned out, all the company offered was a questionnaire asking if you ate when you weren’t hungry. I marveled at its chutzpah but also felt vaguely relieved that at least the company hadn’t made even more grandiose claims about genes. You sure can’t find out about them with a few questions; you almost certainly have to do some lab work.

In the 11 years since the sequencing of the human genome was announced, new techniques have radically changed the approach to identifying genes related to particular traits. In the past, one would examine a handful of “candidate” genes, whose functions were already known to relate to the trait. Now “genome-wide” surveys examine the links between a trait and all of the 20,000 or so human genes.

This has been an extraordinarily powerful approach. For the most part, what has emerged from these studies is evidence of the minimal extent to which some trait is “in your genes” and of how relatively unimportant any given gene is.

Let’s consider an acclaimed genome-wide study, published in the journal Nature in 2010, which looked for genes related to human height. These are insanely difficult studies to do. The work required an international consortium of more than 280 scientists and involved examining more than 180,000 human genomes.

And the results? The scientists implicated hundreds of genetic variants in regulating height. A handful of those genes were already known to be involved in skeletal growth; others had been shown only to have functions unrelated to growth, and scads of the genes were previously functional mysteries—terra incognita.

The single genetic variant that was most powerfully associated with growth explained just 0.4% of the variation in growth, and all the hundreds of identified variants put together explained only about 10% of the variation—which is not a lot of explanatory power.

An equally acclaimed study used similar techniques to study body-mass index (BMI). After research on nearly 250,000 subjects, the genetic variant most clearly identified with BMI accounted for only 0.3% of the variation.

Turning to the genetics of behavior, a superb recent study, published in the journal Science, examined the genetic variants associated with educational attainment, as measured by years of schooling (which, it should be noted, is not remotely equivalent to IQ). There were the usual over-the-top numbers testifying to the gargantuan nature of such studies—126,559 subjects, about 200 authors of the research paper.

The results? The most predictive single genetic variant accounted for 0.02% of the variation between individuals. Putting together all of the identified genetic variants explained only about 2% of the total variation. In a commentary accompanying the paper, the authors concluded with an understatement equally applicable to height or BMI: “In short, educational attainment looks to be a very polygenic trait” (i.e., one involving many genes).

In other words, a large number of genes and their variants influence these traits, with any given gene being of minuscule significance. And even our knowledge of the collective effect of the hundreds of genes involved explains just a tiny percentage of human variation.

So the next time I’m tempted to read an email pitch about some trait supposedly “in my genes,” I’m going to stay on more solid ground and answer the email from the widow of the Nigerian general with a financial proposition for me.