By comparing sets of genes from tens of thousands of people, researchers have found some that the body can’t seem to live without.

By comparing sets of genes from tens of thousands of people, researchers have found some that the body can’t seem to live without.

Jane Ades/NHGRI

The 3230 genes you can’t do without

Fiddle just a little bit with any one of about 3200 genes in the human body and you could be toast. That’s the conclusion of a new study, which finds that about 15% of our 20,000 genes are so critical to our livelihood that certain changes can kill us before we’re born. The findings should help researchers better track down the genes that cause human disease.

Given how useful any insights into gene function are for understanding the genetic basis of disease, the data are “priceless,” says Kári Stephánsson, a geneticist at deCODE in Reykjavík.  

The new study, which was recently posted to a preprint repository but has not yet been published in a peer-reviewed publication, was the result of researchers comparing the parts of the genome known as exomes, which code for proteins, from 60,000 people—10 times more than had ever been attempted. Researchers led by Daniel MacArthur, a geneticist at the Broad Institute of the Massachussetts Institute of Technology and Harvard University in Cambridge, Massachusetts, achieved this feat by reaching out to teams around the world who had collected their own sets of exomes. Among all those genes, the researchers found 10 million variants—places within genes that varied from person to person.

MacArthur’s team calculated how many variants each gene should have if those changes arise by chance. Then they compared that to the number of variants actually found for each gene. The result: 3230 genes that had either no observed variation, or compared to what was expected, much, much less of the kind of changes that could lead to a malfunction of the gene.

Such data suggests that, whenever one of these genes mutates, the embryo usually dies or the person is too sick to reproduce—so the variation disappears. ”Genes that displayed no variants should be essential or play crucial biological functions,” says Ping Xu, a molecular microbiologist at Virginia Commonwealth University in Richmond.

Ping and others have similarly found such essential genes in bacteria or in mice. And many of the highlighted human genes are associated with the same critical cellular operations, such as the cell’s protein-building factories, as in those species, MacArthur’s group reports. About 20% of the human genes uncovered by the analysis are already associated with diseases, but many are not—yet. They are the first places David Goldstein, a geneticist at Columbia University in New York City who was not involved with the work, says he will look for connections to medical conditions. And because they seem vital “these genes can be specifically monitored during drug development for potential side effects and cytotoxicity,” Xu says.

The key to the success of MacArthur’s team, say other researchers, was gathering so much DNA data. (MacArthur declined to comment since the paper isn’t officially published yet.) “They do a really nice job of showing that you get a lot more information by studying an order of magnitude more people,” says Joshua Akey, a population geneticist at the University of Washington in Seattle.

Even so, Goldstein is quick to point out that 3230 is not the complete set of essential genes in the human body and that only by studying more exomes will researchers be able to refine that number. Furthermore, exomes don’t cover DNA in between genes, which help regulate gene activity, and variation there can also be important. Finally, what researchers really want to know is what specific part of each gene is essential. Bottom line, Goldstein says: “This is another step on a very long road that we are on to understand what does and doesn’t happen in the human genome.”

*Correction, 12 November, 3:54 p.m.: The introduction of this story has been changed to note that changes to the 3230 genes are not necessarily minor.