Bring us your genes!

A Viking scientist’s quest to conquer disease.

By Adam Piore
Illustration by John Hendrix July 2, 2015

In the ninth century there was a Norwegian Viking named Kveldulf, so big and strong that no man could defeat him. He sailed the seas in a long-ship and raided and plundered towns and homesteads of distant lands for many years. He settled down to farm, a very wealthy man.

Kveldulf had two sons who grew up to become mighty warriors. One joined the service of King Harald Tangle Hair. But in time the King grew fearful of the son’s growing power and had him murdered. Kveldulf vowed revenge. With his surviving son and allies, Kveldulf caught up with the killers, and wielding a double-bladed ax, slew 50 men. He sent the paltriest survivors back to the king to recount his deed and fled toward the newly settled realm of Iceland. Kveldulf died on the journey. But his remaining son Skallagrim landed on Iceland’s west coast, prospered, and had children.

Skallagrim’s children had children. Those children had children. And the blood and genes of Kveldulf the Viking and Skallagrim his son were passed down the ages. Then, in 1949, in the capital of Reykjavik, a descendent named Kari Stefansson was born.

Like Kveldulf, Stefansson would grow to be a giant, 6’5”, with piercing eyes and a beard. As a young man, he set out for the distant lands of the universities of Chicago and Harvard in search of intellectual bounty. But at the dawn of modern genetics in the 1990s, Stefansson, a neurologist, was lured back to his homeland by an unlikely enticement—the very genes that he and his 300,000-plus countrymen had inherited from Kveldulf and the tiny band of settlers who gave birth to Iceland.

Stefansson had a bold vision. He would create a library of DNA from every single living descendent of his nation’s early inhabitants. This library, coupled with Iceland’s rich trove of genealogical data and meticulous medical records, would constitute an unparalleled resource that could reveal the causes—and point to cures—for human diseases.

In 1996, Stefansson founded a company called Decode, and thrust his tiny island nation into the center of the burgeoning field of gene hunting. “Our genetic heritage is a natural resource,” Stefansson declared after returning to Iceland. “Like fish and hot pools.”

Stefansson set sail on an epic journey. He and his crew collected DNA from 150,000 of their fellow countrymen (half the population) and constructed a genealogical chart that accounts for the family tree of virtually every member of the small island nation. Next they succeeded in reading the entire 3-billion nucleotide genetic sequences of more than 11,000 Icelanders. They could now infer the individual genomes of the entire Icelandic population. Then they embarked on a massive treasure hunt for individual Icelanders with missing segments of DNA carried by the rest the population. They matched these “knocked out” genes with their impact on the individuals carrying them. That quest has only just begun. But already it has resulted in a rich bounty.

In 2015, Stefansson and his team reported they had found rare mutations that increase the risk of Alzheimer’s disease, gallstones, atrial fibrillation, and thyroid disease. If researchers can find the chemical pathways the mutations affect, they will have unprecedented insights into the causes—and possible cures—of some of humanity’s worst diseases.

In the Middle Ages, scribes set down the legendary deeds of Kveldulf the Viking and the nation’s legendary explorers and families in the Sagas of the Icelanders. In the depth of winter not long ago, I ventured to Iceland to write my saga of the Icelander who would conquer the genetic code.

Iceland in December is a cold and forbidding place. The sun doesn’t rise until 11 a.m. By 4, it’s gone. The wind whips down glacier-capped mountains, across barren, volcanic plains. On the dark morning I set out from my waterfront hotel to meet Stefansson, a bitter chill was in the air.

Decode’s well-lit, modern headquarters, with its cheerful potted plants and hardwood floors, is perched on the outskirts of Reykjavik like a warm medieval castle on the edge of a forest. But as I approached, I was apprehensive. Famed for a mercurial temperament, Stefansson has been known to get up and leave when irritated by an interviewer—or simply drop his forehead to the edge of his desk and loudly sigh. Due to a misunderstanding, I was half an hour late.

Outside Stefansson’s office, Decode’s press handler sat next to me seething. He warned that his mercurial boss’s reaction was impossible to predict. Would he still meet with me at all? “We’ll see,” he muttered.

Finally, Stefansson emerged. Trim, tall, with a beard, a shock of white hair, wearing sneakers and a white striped hoodie, he extended his hand. I apologized.

“Americans have a tendency to be arrogant, devious, and stupid,” Stefansson replied, smiling. Then he led me into his office.

Stefansson often adopts a subversive, Viking-like defiance, and enjoys playing the provocateur. The son of a famous Icelandic radio journalist, Stefansson insists he remains an artist at heart, who grew up “absolutely determined” to become a writer and a poet—a path that seemed very likely until fate intervened. One night just as Stefansson was completing high school, he got drunk with a classmate, and stayed out wandering around after the bars closed.

“In the morning, when we were somewhere between drunk and hungover, we decided to walk to the university and apply for medical school,” Stefansson said. “I had no interest in it. He wanted to go to medical school. And then, all of a sudden, I was in medical school.

“I found medical school extraordinarily boring,” Stefansson added. “I don’t know why I stayed.”

There was at least one reason to stay. By the time Stefansson left home, he had developed a personal understanding of the ravages of unchecked diseases. Stefansson’s beloved older brother—a gifted athlete, who had “forced good literature on me” and “was my hero growing up”—had a psychotic break. Stefansson was struck by the power of his brother’s delusions—at one point, he called up Stefansson and apologized for attacking him with an ax, an incident that never occurred—and the devastating impact the little-understood disease wrought on his family.

“He was completely off the wall,” Stefansson said. “Schizophrenia is a disease of thoughts and emotions. And what defines you as an individual are your thoughts and emotions. So once you are hit by a disease like that, you are changed. You become a different creature to those who interact with you.”

Stefansson dove into the psychiatric literature of the time searching for answers, and for a time flirted with the possibility of devoting his life to psychiatry. But he found psychiatrists to be “poorly educated” with “an uninteresting approach to the problem.” Instead he focused on the biological study of diseases of the nervous system, neurology, and pathology.

Icelanders had been living in geographic isolation since the days of Skallagrim, incubating an unusually homogenous gene pool of blonde-haired, blue-eyed virtual cousins.

In 1977, he landed at the University of Chicago as a postdoc and, eventually, a member of the faculty. It was there that he fell in love with science and surrendered to his true calling. In Chicago, Stefansson distinguished himself for his work on multiple sclerosis (MS), the crippling disease where the body’s own immune system attacks the lipid brain coatings, known as myelin sheaths, crucial to the transmission of the electrical impulses that run the nervous system.

As a young professor and researcher in the 1980s, Stefansson and a graduate student named Jeffrey Gulcher spent months hunting for individual proteins expressed in the brain that might be causing the body to attack the sheaths. If they could identify them, they reasoned, perhaps drugs could neutralize them. But the approach was frustrating. Even when they succeeded in isolating a protein that appeared to be associated with the onset of an immune attack, there was no way to know for sure it was actually the cause of the disease, and not part of the body’s response to it.

Stefansson recognized genetics might provide a far better solution. Every one of us carries somewhere in the neighborhood of about 20,000 different genes in our cells, each compromised of anywhere from 27,000 to 2.4 million pairs of DNA’s core building blocks, nitrogen compounds known as nucleotides that contain one of four bases, cytosine, guanine, adenine, and thymine. The sequences of these bases, referred to by their first letters—“A,” “C,” “G,” and “T”—encode the molecular level instructions for every single protein our bodies produce. Those proteins in turn determine everything from our hair color to our temperament to the shape of our toes.

But sometimes our genomes contain typos—mistakes in the sequence of base pairs that can cause the body’s instructions to go awry. It is the combination of lifestyle and these mutations in the DNA sequence that lies at the root of most human diseases, most scientists believe.

It seemed obvious to Stefansson that if you could find differences in the genes coding for proteins unique to individuals afflicted with MS, there was a high likelihood that protein played a role in causing the disease and would serve as a promising drug target.

SPLENDID ISOLATION: Icelanders had been living in geographic isolation since the days of Vikings, incubating an unusually homogenous gene pool of blonde-haired, blue-eyed virtual cousins. The insular population makes it relatively easy for geneticists to spot errant disease-causing mutations.Lindsay Blatt

In the 1980s, identifying genetic differences was a Herculean chore. During their initial exploration of proteins involved in MS, Stefansson and Gulcher sequenced a single 7,500-unit strand of DNA that coded for a specific protein using the best tools science had to offer. It took three years.

By the early 1990s, Stefansson had become convinced that things were about to change. Scientists had launched the Human Genome Project, a $3 billion international effort to identify all the genes in human DNA, and determine the sequence of the estimated 3 billion chemical bases thought to comprise those genes. In 1993, Stefansson and Gulcher were recruited by Harvard Medical School, and the new technology would be just what they needed to study MS.

At Harvard, Stefansson had a breakthrough idea. He would conduct his MS study in Iceland. In diverse populations like the United States, disease-causing genes are difficult to distinguish from other variations because there are a dizzying number of individual variations to sort through. Any given American might be carrying DNA reflecting mutations unique to Africa, Eastern Europe, Brazil, Russia, and the North Pole. Finding which one of those mutations accounts for a given disease is a little like searching for a typo by comparing a book in German to its counterpart in French.

But Icelanders had been living in geographic isolation since the days of Skallagrim and his fellow settlers, incubating an unusually homogenous gene pool of blonde-haired, blue-eyed virtual cousins. The insular population would make it far easier to spot errant disease-causing mutations, and gain new insights into the causes of a wide array of human ailments. Stefansson asked an Icelandic neurologist to recruit MS patients and instruct them to bring along a close, healthy relative for comparison. Many arrived instead trailing large entourages of extended family, all eager to help.

Gulcher and Stefansson returned to Boston with DNA from 200 MS patients, just 15 siblings, but hundreds of more distant kin. Most geneticists relied on sibling pairs for studies. But Stefansson had the insight that his distant relatives would prove even more useful. Distant relatives inherit far fewer common genes than two individuals who have the same parent (who share roughly 50 percent of the same DNA). While that made it harder to find two distant relatives who carried the same disease-causing mutation—once you located them, there would be fewer additional commonalities to eliminate as the cause of the disease.

A Harvard biologist compared Stefansson unfavorably to his Viking ancestors—“at least they made no pretense that their raids were in the public interest.”

A confident Stefansson decided to expand his lab and hunt for shared DNA sequences that would narrow the location of the gene or genes related to MS. When the pair applied to the U.S. National Institutes for Health for funding, however, they received a bitter dose of reality. The reviewers were unimpressed by his sample of Icelandic relatives. Why, they asked, weren’t there more sibling pairs? Siblings were the accepted protocol for genetics studies, they reminded Stefansson and Gulcher, because they share DNA—entirely overlooking the fact that distant relatives carry shared DNA as well.

Gulcher was furious. But Stefansson was undeterred. Even if they had received the grant, he did not believe it would have been sufficient to finance the vision he had begun to develop since returning from Iceland. Over morning breaks with Gulcher at a Starbucks down the street from his Harvard lab, Stefansson argued for something far larger in scope. Iceland had an obsession with genealogy, a tradition of centralized medical record keeping, and an altruistic populace. The nation had a massive tissue bank that had been preserving samples from autopsies since 1915, and had recorded every single case of cancer diagnosed since 1952.

If Stefansson and Gulcher could open an institute, or sell venture capital firms on a private company, they could be the ones to harness Iceland’s vast genetic resources to glean insights, not just about MS—but about all human diseases. If they could raise the money, import big biotechnology to Iceland, reverse the brain drain—bring back all those Harvard and Stanford Ph.D.s who’d gone abroad for want of local opportunity—they could rely on widespread support. There was no telling what they might be able to do.

The pair began approaching venture capital firms. Within a couple of months, Stefansson raised $12 million. He was headed back to Iceland to set up shop.

As for his coveted job at America’s most famous college? “Harvard is a lousy university,” Stefansson said. “It’s a loose confederacy of independent institutions and I was eager to get away from there. It was a boring place.”

One of the first things Stefansson did after launching Decode in 1996 was gain access to Iceland’s largest genealogy database, created by a computer programmer on his off time for family-tree enthusiasts. It contained 400,000 records, dating to the settlement of Iceland, including calfskin manuscripts, church records, and the census of 1703, allowing virtually all Icelanders to gain at least some insights into their family origins. But the database was not as comprehensive as it could have been, and so Stefansson sent teams to comb through parish records and subsequent censuses to beef it up.

Icelandic political leaders were hungry to build a strong biotech sector. In 1998, Stefansson began lobbying for a vast, new database that would include the national health insurance records of all living Icelanders, and could be cross-referenced with genealogy and DNA samples. Though much of the data had already been computerized, it was spaced out in hospitals and doctor’s offices across the country, in a wide array of disparate systems that would have to be standardized and unified.

The government unveiled a bill to grant Decode the exclusive right to build the database and market it abroad to researchers. The legislation created a firestorm. It contained a provision that assumed Icelanders granted Decode universal consent to use their information, kept anonymous, unless they submitted forms opting out. Critics labeled it an Orwellian power grab. Harvard molecular biologist Richard Lewontin compared Stefansson unfavorably to his Viking ancestors—“at least they made no pretense that their raids were in the public interest”—and suggested a scientific boycott of Iceland. A Dutch Data Protection Commissioner warned Iceland risked conviction by a European Court. After months of vitriol, the legislation passed Parliament. But it was so freighted with privacy protections—including a requirement that Decode fund a watchdog—that Stefansson abandoned the effort to partner with the government to create a national database. “It would have driven us into bankruptcy overnight,” he said.

It proved a minor setback. Decode continued to collect medical records and DNA from volunteers—along with the permission 
to reuse them for future studies approved by government bioethics and privacy committees. Each 
volunteer filled out detailed questionnaires on characteristics ranging from hair color to cholesterol 
levels to smoking habits, adding up to 2,500 traits. By 2001, Decode reported it discovered or localized genes with small but significant links to preeclampsia, osteoporosis, schizophrenia, Alzheimer’s, stroke, and heart attack. Its database was seen as an information goldmine.

That year, Decode launched an initial public offering on NASDAQ and was valued north of $1 billion. By the mid 2000s, Stefansson headed a huge enterprise with hundreds of scientists, geneticists, computer engineers, investor relations personnel—even anthropologists. They were housed in a gleaming new headquarters with an airy atrium lunchroom ringed by several floors of glass-walled offices and laboratories, and topped with a polished glass ceiling that bathed the blond wood floors and paneled walls in sunlight almost 24 hours a day during Iceland’s luxurious summers.

In 2000, the Human Genome Project unveiled a draft sequence, revealing that 99.9 percent of the 3 billion bases carried by each individual are identical. Many geneticists argued that meant the causes of common diseases were likely to lie in the remaining 0.1 percent, and most variations were in single bases attached to nucleotides, called “single nucleotide polymorphisms” (SNPs).

Scientists discovered that most genetic material is passed from each parent to child in a remarkably structured architecture that consists of a predictable number of genes chunked together and kept largely unchanged, in a grouping called a haplotype. These haplotypes were believed to have developed very early in human evolution, which meant there were only a small number of common flavors possibly carried by every single person.

In 2002, an international consortium of geneticists set out to find and catalogue individual mutations that were thought to be distinct to every single known individual haplotype, in order to build such a reference guide. By the fall of 2005, scientists had catalogued almost 1.5 million points of DNA, or SNPs, associated with specific haplotypes—which could be quickly identified in blood samples using ingenious chips, called SNP chips. These SNPs could then serve as genetic “signatures” for specific haplotypes, or regions of haplotypes. If you could find SNPs unique to individuals with a disease, it was thought, you could home in on that area of the haplotype and find the mutation that caused that disease.

When the first SNP chips arrived in Decode headquarters in 2006, Stefansson met with his lieutenants to strategize. One day, his computer and statistician whiz, Danial Gudbjardsson, an MIT-trained electrical engineer, burst into Stefansson’s office “absolutely fuming with enthusiasm,” Stefansson recalled. Earlier that day Decode had received word from the Icelandic Government’s Data Protection Commission and Ethics committee granting it permission to do a study on atrial fibrillation (AF), the most common form of irregular heartbeat. Now Gudbjardsson had arrived to report that his team had homed in on the locations of two genetic variants that conferred about a 70 and 40 percent increase in the likelihood of developing AF.

Those with two copies of the more powerful variant—one from each parent—had a significant increase in the likelihood of developing the disease. AF was a big fish—it can cause palpitations, congestive heart failure, and increases the risk of stroke four to six times. And they had already done a study on strokes. All they’d had to do was pull out the DNA they already had in their genetic library and run a SNP chip searching for the genetic mutations they found that were associated with AF.

Finding genetic links to the disease using the new methods had taken three hours. In the following months, Decode found common variants of genes associated with glaucoma, sleep disorder, breast cancer, heart attack, osteoporosis, obesity. “After the SNP chips, we had almost a factory of discoveries,” Stefansson said. “We left the rest of the world behind in the dust.”

BANK ON IT: Decode, which houses blood samples from thousands of Icelanders, has constructed a genealogical chart that accounts for virtually every member of the small island nation.Decode Genetics

Even so, within just a couple of years, skepticism about the power of the new techniques began to grow. David Goldstein, director of the Institute for Genomic Medicine at Columbia University, had collaborated with Decode on at least two studies. But after a 2007 study coordinated by a research institute of the University of Oxford, he emerged as a critic. The two-year study, which drew on data from 50 research groups, screening the genes of 17,000 people, was described as the biggest study of common disease every completed. It linked 24 genes to seven of the most common diseases, including heart disease, Type 1 and 2 diabetes, bipolar disorder, and Crohn’s disease.

But Goldstein was struck by the fact that the study had found only one gene in the case of bipolar disorder—a disease known to be influenced by inherited genes—and the number of bipolar cases correlated with the gene was “barely significant.” In fact, the sum of all the genes associated with the various diseases in the study—even if you added in Decode’s findings on some of those diseases—didn’t seem to come close to accounting for all of the cases one would have expected based on how often these diseases ran in families.

“There were many individuals in the study, so the sample size was reasonable, there should have been more there if common variation was the cause,” he said. “There just wasn’t much there.” Scientists began to refer to this confounding outcome as the “mystery of missing heritability.” Perhaps, Goldstein and others argued, most diseases weren’t as influenced by common variants as many had assumed, but rather were more influenced by a wide variety of rare mutations—too rare to assume they could be reliably found on haplotypes.

Stefansson continued to argue that even if common variants only explained part of the story, they could still be useful for treating diseases. If they revealed insights into the underlying chemical pathways involved in diseases, those could be used to develop drugs.

But he had bigger problems. By January of 2009, Decode had just $3.7 million left. The credit markets were frozen, and the Icelandic government could do nothing to help—the nation itself was teetering on the verge of insolvency. In November, Decode declared bankruptcy.

Stefansson was not yet willing to concede defeat, and he was pretty certain he knew at least a couple financiers who might be willing to help buy his company out of bankruptcy. When Decode had first gone public in 2000, two venture capital funds—Arch Venture Partners, in Seattle, and Massachusetts-based Polaris Venture Partners—had both cashed out to the tune of about a 700 percent return on investment. A persuasive Stefansson talked the two venture capital companies back into the fray; for significant shares in Decode, they ponied up a total of $40 million. “It seemed to me the company was hitting its stride,” said Terry McGuire, a partner at Polaris.

Indeed, just as Decode emerged from bankruptcy in January 2010, a new innovation was hitting the market that would prove more transformative than SNP chips—techniques for sequencing an entire human genome faster and cheaper than ever. If you could sequence an entire human genome cheaply and compare it with others, you didn’t need to look at common haplotype markers. You could compare the genetic code at every spot and theoretically, with enough computing power, pick out those rare mutations. The Human Genome Project unveiled a draft of the first fully sequenced genetic code in 2000 for about $500,000. For that same amount of money in 2010, Decode could acquire a revolutionary new machine created by the biotechnology company Illumina that was capable of fully sequencing three DNA samples simultaneously every 10 days. One of the first things Stefansson did was strike a deal to acquire four of the machines.

The invention and commercialization of whole-genome sequencing machines was a breakthrough. Previously, scientists had assumed the answer to understanding disease lay in finding common mutations passed down from long ago and identifiable using SNP chips. But this approach had what many would come to see as a major limitation. Common haplotypes were old. If a mutation caused a disease wouldn’t it reduce an individual’s chance of survival? Wouldn’t that then reduce that individual’s chances of passing on his DNA to his offspring? Wouldn’t natural selection weed out the mutation?

Now, with whole genome sequencing, it became possible to look for extremely rare, relatively new mutations—mutations so powerful they were likely to be quickly weeded out of the population by natural selection. They were so deadly, in other words, they reduced the likelihood a person would live to pass on the mutation to offspring. You might not find these mutations on the ancient, shared haplotypes. But precisely because they were 
so deadly, they were more likely to reveal something significant about a disease pathway.

These “disasters of evolution,” Stefansson said, “disrupt physiological function rather than placing you on a normal distribution curve. But they give you a biochemical pathway—they give you clear indication as to what biochemistry and what biology is being perturbed to cause the disease.”

As Decode emerged from bankruptcy in 2010, a new excitement crept 
into the offices. Several times a week, Stefansson and his top lieutenants would file into a small, glass-walled conference room. Stefansson would emerge, stroll to the front of the room, and take a seat at the head of a big table, facing his brain trust of Ph.Ds.—“the kids”—arrayed in front of him across the other side “so that I can see them, and they are very nice to me.”

Sitting around that table, Stefansson and his team discussed how they might most effectively use their new whole genome sequencing machines. They would
 do so by “double-dipping”—first finding 
a large group of Icelanders with sufficiently variant family trees to capture the disparate genomes present in every branch—then winnowing down those samples even further by selecting individuals from within them who also happened to have rare diseases of particular interest.

Decode announced it had sequenced 2,635 Icelanders, a number that allowed it to infer the genomes of more than 100,000 Icelanders.

As a young neuropathologist, Stefansson had spent many days slicing up the brain tissue from patients with Alzheimer’s disease, contemplating as so many other researchers had before him the telltale plaques and tangles that scientists have long known were associated with the disease. Way back in the winter of 1997, Stefansson had hit the road with his wife and one of his venture capitalist backers, piloting a car through Iceland’s picturesque west coast, visiting nursing homes and collecting blood samples which still remain in a basement bio-bank available for sequencing.

One afternoon in 2011, Thorlakur Jonsson, who headed the Alzheimer’s project, scanned a report and noticed that it tagged a strong association on a gene known to code for Amyloid Precursor Protein (APP). The finding wasn’t particularly surprising. APP was known to be a primary component of amyloid plaques, misfolded proteins that clogged the brains of those stricken with Alzheimer’s disease, plaques that many believed may even cause the disease itself. Mutations in the genes coding for APP had long before been shown to increase the likelihood of developing Alzheimer’s disease.

But when Jonsson looked closer at the report, he spotted something unexpected—the computer’s algorithms had flagged the APP mutation not because it was associated with patients who had the disease. It had flagged the mutation because those who had it were less likely to develop Alzheimer’s disease. Could this be an anomaly?

The finding presented Stefansson with an opportunity to explore a question that had long intrigued him. In the many hours he’d spent hunched over a microscope, he’d always been struck by the fact that many of the features associated with Alzheimer’s—most notably its amyloid plaques and the neurofibrillary tangles—were also present in elderly patients who did not have the disease. It was just that the density was much higher.

Stefansson knew that cognitive decline was associated with aging. Young people did not have cognitive decline, or plaques and tangles. It seemed likely to him that the plaques and tangles associated with Alzheimer’s might also be the cause of cognitive decline. Stefansson wondered if the mutation that protected against Alzheimer’s also protected against regular cognitive decline. Proving that point would “give an extraordinarily strong argument in support of the notion that Alzheimer’s is just an accelerated form of aging in the brain,” he said.

To test the hypothesis, the Decode team sequenced the genes of a control group—a focused population of elderly patients of similar age and living under similar circumstances. When they compared the sequences from elderly patients with and without cognitive decline, they proved this point—the elderly patients they sequenced who had the mutation did not exhibit signs of cognitive decline. Those without the mutation—even those without Alzheimer’s—often exhibited cognitive decline.

When Decode published its paper on the new gene, it made worldwide headlines. If drug makers could find a way to replicate the gene, they would have a blockbuster drug that might cure Alzheimer’s disease and other forms of senility long associated with aging.

The company was on a new roll. In 2012, Augustine Kong, the company’s vice president of statistics, had begun to dig into the numbers associated with the fully sequenced schizophrenic and autistic samples, paying special attention to what are called “de novo” mutations—spontaneous mutations appearing in a child that were not present in either of the parents. After identifying de novo mutations in the schizophrenic, autistic, and other samples by comparing the DNA of individuals with the DNA of their parents, Kong began to look for any common characteristic in the parents that appeared to be associated with the mutations.

One variable in particular stuck out—in well over 90 percent of the cases, the increased number of mutations appeared to be associated with the age of the father. Stefansson was astounded. He and Kong went back through Decode’s genealogy database, browsing hundreds of years into the past, and analyzing how the ages of new fathers had changed over the years. To their surprise, the mean age of fathers at conception of a child was far higher in the 19th century than in much of the 20th century. Then in the 1980s, the mean age of the father at the time of conception rose again—which, when you analyzed the correlations, appeared to go hand-in-hand with the rate of autism.

In a paper published in a 2012 issue of Nature, Stefansson, Kong, and collaborators presented evidence from 219 samples culled from 78 trios of two parents and a child with autism or schizophrenia. In addition, they sequenced the genomes of high-risk families of schizophrenia and autism to see if they could find any inherited genes that might predispose individuals to the disease. The finding jibed with others coming out at the time that had found a link between older fathers and higher rates of schizophrenia and autism. Decode’s paper generated worldwide headlines. In fact, Stefansson argued, the finding showed not only what might be the cause of schizophrenia and autism, but it explains nearly all the difference in the rate of new mutations introduced into populations around the world. The older the age of the fathers at the time of conception—the more variation you are likely to see in the population.

Itsik Pe’er, an associate professor of computer science at Columbia University who studies computational genetics, said the link between paternal age and mutation rate is now widely accepted, as is the idea that such mutations increase the likelihood of autism and schizophrenia. However, he noted, “mutations occur in nature in all the time. So there are mutations that occur before the father is older, and these mutations are not affected by age.”

In 2012, as Stefansson and his company were unveiling their new studies, Stefansson learned Decode had an admirer in Amgen, the world’s largest biotechnology company. Two of the most promising drugs in the Amgen pipeline had grown out of the discovery of rare, but powerful mutations of exactly the sort Decode was now unearthing. Within a month, Amgen struck a deal to acquire Decode for $415 million. The acquisition revived the criticism among a vocal group of Icelanders that Decode was profiting on the DNA and medical records of Iceland’s citizens, who saw little in return. One writer and blogger mocked Decode for giving citizens a T-shirt when they signed the consent forms.

In 2015, Stefansson told me Amgen “has left us completely alone,” and insisted that he is dedicated to his country’s citizenry. “The people have been so generous in contributing to our research,” he said. “What is our obligation or duty toward them?” His answer is to work out rules with the nation’s health care and ethics authorities to inform Icelanders when they are carriers of “dangerous mutations,” like the gene associated with breast cancer. “I think it is basically criminal not to do it, and not in keeping with the general philosophy in our society,” he said. “We should help those who are at serious risk and the current minister of health is very enthusiastic about finding ways to take advantage of this.”

Decode published four papers in Nature in March of 2015. It announced it had sequenced 2,635 Icelanders, a number that allowed it to infer (using family tree data and what it knew about the mutation rate) the genomes of more than 100,000 Icelanders whose genomes have not yet been fully sequenced. It has now sequenced 11,000 full genomes. Using the same statistical methods, Decode could infer the likely presence of mutations in virtually the entire Icelandic population, as they know the mutation rate, how most of the population is related, and how much DNA everyone shares.

In the trove of data, Decode found rare mutations that dramatically increase the risk for Alzheimer’s, gallstones, atrial fibrillation, and liver and thyroid diseases—mutations that appear to be the result of “knocked out” or missing pieces of DNA. Perhaps most intriguing is the detective work that lies ahead. All told, Decode identified 1,171 knocked out genes, present in nearly 8 percent of the 104,000 people studied. The next step is to work backward—in the opposite direction one normally goes in genetics research—and cross-reference these knockouts with medical records and phenotypical data and try to determine the impact of these mistakes in nature.

Daniel MacArthur, a geneticist at Massachusetts General Hospital, and assistant professor at Harvard Medical School, who was not involved in the Decode study, called its reach astonishing. “It’s a pretty amazing set of results,” he said. “Decode over the last decade has assembled an astonishing collection of genetic data—without question the deepest genetic characterization of any national population in the world.” He pointed out, however, that in a “bottleneck” population like Iceland’s, “you can only find a relatively small subset of [knockout] genes.” Decode can identify mutations that cause disease in Iceland, but it will be up to other groups studying other populations elsewhere to identify the vast mutations that remain in the human genome.

In Stefansson’s office on that winter day, the Viking scientist concluded our interview with a nod toward his brother, his hero, who ingrained a love of literature in him. “When I think about what I have been doing in science, I am convinced that this all has to do with a story,” he said. “If you cannot tell a story with your science, your science isn’t worth a shit. I’m 63 and I think back on what is important in my life. I wish I would be a better man. I wish I had been a better husband and a better citizen. But I have told a certain story with my life, and, probably the most interesting aspect of that story are the discoveries that I’ve made in the science lab—the contribution I have made to the understanding of how man is put together.

“If you think about what is most important when it comes to life on this planet, it is this information that lies in DNA. If you take Walt Whitman and a leaf of grass, the difference between the two lies in the difference in the sequence of their A’s, C’s, G’s, and T’s. DNA is the holy grail of life.”