Stephen Friend never thought he’d go looking for superheroes.

“The idea came from frustration and unfulfilled quests,” says Friend, a physician at the Icahn School of Medicine at Mount Sinai. For decades, he had watched geneticists trying to find the genes that underlie diseases as diverse as cystic fibrosis, Alzheimer’s disease, and schizophrenia. Such studies have been undoubtedly successful, but the growing list of culprit genes have rarely led to treatments for their respective disorders.

“I had an a-ha moment,” says Friend. “If you want to find a way of preventing disease, you shouldn’t be looking at people with the disease. “You should look at people who should have been sick but aren’t.”

These people, unbeknownst to them, carry genes that all but guarantee that they’ll get fatal diseases. And yet, somehow, they’re completely healthy. They might carry other genes that mitigate their risk. Or perhaps, some aspect of their diet, lifestyle, or environment shields them from their harmful inheritance. Either way, Friend reasoned that if he could find these “genetic superheroes,” and work out the secrets of their powers, he could find ways of helping others to beat the odds.

It was a reasonable concept, with an encouraging precedent. In 2006, by studying an African-American woman with extremely low cholesterol, scientists identified mutations that dramatically reduce the risk of heart disease by disabling a gene called PCSK9. That discovery led to a new generation of heart drugs. Elsewhere, a Labrador puppy called Ringo was bred with a mutation that causes severe muscular dystrophy, but lived in perfect health thanks to a protective mutation in a different gene. Perhaps Ringo’s resilience will lead to cures for the otherwise untreatable disorder.

Friend’s collaborator Eric Schadt has met a couple of genetic superheroes himself, as Aimee Swartz detailed in The Atlantic two years ago:

“Schadt points to two patients who have come through Mt. Sinai: a woman in her mid-50’s with a mutation in the CFTR gene—the gene understood to cause 100 percent of all cases of cystic fibrosis—who has never had more than mild respiratory issues. And a 45-year-old man who learned during an unrelated medical procedure, that he had Louis–Bar syndrome—a rare neurological disorder caused by a gene mutation—though he had never exhibited any symptoms of the usually fatal disease. “Both carry the code for an inherited childhood disease but do not bear the symptoms of the disease,” he said.

Rather than stumbling across such superheroes, Friend and Schadt wanted to search for them in a systematic way. So they launched an ambitious initiative called the Resilience Project, in which they analyzed the genomes of almost 590,000 people who had taken part in a dozen earlier studies.

They focused on mutations in 874 possible genes that have been linked to a severe childhood disease, and have been described as “completely penetrant”—that is, if you have the mutation, you will inevitably develop the disease. Friend’s team was looking for healthy adults who had somehow dodged these inescapable bullets.

At first, they identified almost 16,000 possible superheroes, who collectively had 300 mutations between them. But in some cases, they weren’t confident that the mutations had been accurately sequenced. In other cases, the mutation turned out to be quite common in the general population, which meant that it couldn’t possibly be the cause of a rare, severe disease. After excluding these uncertain cases, the initial list of 16,000 suspects dropped to just 303.

The team convened an expert panel to carefully review each candidate, and rule out any who might not actually be resilient. For example, many people had mutations that weren’t completely penetrant after all: they sometimes cause disease later in life, or lead to mild symptoms that slip under the radar. One woman had a mutation for Gaucher’s disease and had never been diagnosed. But when the team examined her medical records, they clearly saw mild signs of the condition, including a long history of easy bruising and bleeding.

After filtering these ambiguous cases, the team were left with a final list of just 13 potential superheroes. Three should have had cystic fibrosis, but didn’t. Three others should have had atelosteogenesis, a bone and cartilage disorder that kills most people before they’re even born. Who are these people?

No one knows.

The team couldn’t get in touch with any of their final 13 because they had all taken part in studies whose consent forms didn’t include any kind of re-contact clause. That’s a huge problem. It means that the team can’t answer the crucial question: What makes these people resilient?

“I’m really frustrated,” says Friend.  

Without re-contacting, “they can’t truly nail the claim of resilience” at all, says Chris Gunter from Emory University School of Medicine. The individuals might have subtle signs of disease, like that woman with undiagnosed Gaucher’s. They might have been diagnosed after their medical records were last filed. They might even be genetic mosaics—people who carry distinct genomes in different parts of their bodies. If their disease mutations are only found in organs that aren’t affected by those diseases, that would explain their supposed ‘resilience.’

“The authors did a tremendous amount of work but I’m still a bit skeptical of the remaining candidates,” says Deanna Church from 10x Genomics. “If these individuals really have pathogenic variants in these genes and don't have the disease, it would be really interesting, but I think we need more information to really know if that is the case.”

Friend isn’t giving up. Rather than analyzing data from existing studies, his team has been trying to recruit over a million volunteers for a fresh study. This time, they won’t just be looking at severe childhood disorders, but also more common ones like Parkinson’s or Alzheimer’s. And this time, the consent forms will include some frickin’ re-contact clauses. As Friend said in his TED talk, “We need a swab of DNA and a willingness to say, ‘What's inside me? I'm willing to be re-contacted.’”

If they find genuine superheroes, they can run experiments to work out the secrets behind their powers. For example, they could create stem cells from samples of skin, use those cells to grow laboratory facsimiles of various organs, and study those “organoids” for clues to their owners’ resilience. They could then test their predictions by using gene-editing tools like CRISPR to remove, add, or tweak different genes from the organoids. The technology is there; it’s just a question of finding the right people.

“On this point the results are rather sobering,” writes geneticist Daniel Macarthur in a commentary on Friend’s paper. In their first study, the team identified just 13 candidates after analyzing 590,000 people. At that rate, it’s very unlikely that even with a million properly consented volunteers, they’ll find enough genetic superheroes to then detect protective genes.

The concept still makes sense, but Macarthur argues that it’s beyond the scope of any single initiative. Instead, researchers will need to combine their efforts to populate a database that spans hundreds of millions of people. He writes: “Finding genetic superheroes will require other kinds of heroism—a willingness of participants to donate their genomic and clinical data, and a commitment by researchers and regulators to overcome the daunting obstacles to data sharing on a global scale.”

Or, as Gunter says, “The more we sequence, the more we need to sequence.”