Home ancestry tests may tell us more, and less, than we wanted to know.
THE YEAR HAS BEEN A JUMBLE FOR ME. Long story short: I was Portuguese, then I wasn’t, then I was again.
It all started after my wife and I spat in vials and mailed the samples off to a laboratory, where our DNA was extracted from the skin cells that had sloughed off into our saliva. Many thousands of DNA segments were read and analyzed, and the results returned via email.
This is both commonplace and remarkable. Not two decades have passed since the human genome was fully sequenced at a cost of $2.7 billion. Today anyone with $60 to spare can get their DNA analyzed—not their entire genome, but a meaningful subset—by a host of companies including Ancestry and 23andMe, the two largest.
By early 2019, some 26 million people worldwide had submitted their DNA to the four largest testing companies and the figure is expected to rise sharply in the coming years, especially over the holidays, as spit kits have become increasingly popular stocking stuffers. But what these gifts promise and what they deliver may be two different things. And what we find may be more interesting than what we set out to look for.
THE GROWING ENTHUSIASM FOR what some have called “recreational genomics” stands in sharp contrast to the “Bring Your Genes to Cal” controversy of 2010. That year, as part of the “On the Same Page” program, which is meant to give freshmen and transfer students a unifying cultural experience, all incoming Berkeley students were invited to submit saliva samples. These would be tested for three common gene variants associated with the metabolism of folic acid, lactose, and alcohol.
The intent was educational. One of the project’s main architects, Berkeley Professor Jasper Rine, explained in the journal Nature that, as DNA testing was poised to become more accessible, more people would soon have knowledge of their gene variants. The problem was that the “gulf between that knowledge and an understanding of what the genetic variations mean” would remain. “Bring Your Genes to Cal” could at least initiate a campus-wide discussion of these issues. All “great universities,” Rine told the journal, have a duty “to confront major societal challenges such as this, so that our students can exercise good judgment, informed by facts.”
Whatever the intention, the backlash was swift and widespread. Bioethicists expressed concern that students would not receive genetic counseling with their results, while others worried about privacy issues. Berkeley economist Brad DeLong, for example, wrote a blog post with the wry title “Bring your genes to your life insurance sales representative.” Eventually the state public health department stepped in and, under pressure, the project was modified, the data only presented in the aggregate.
In some ways, the episode was echoed by 23andMe, the Google-backed gene testing company, which in 2013 was ordered by the FDA to “immediately discontinue” selling its kits to consumers until it could prove that its health-related tests were accurate and that consumers understood the results. The company subsequently shifted its focus to less controversial results showing ancestry and personal traits (like whether you have wet, sticky earwax or the dry, flaky kind). But in 2015, the agency reversed itself, and 23andMe once again tests for genetic variants that indicate susceptibility to certain diseases, including Alzheimer’s and Parkinson’s.
FROM THE OUTSET, I WAS MORE INTERESTED in ancestry tests although, honestly, I wasn’t expecting to learn much. I already knew—or thought I knew—my basic heritage. Like most Americans, I was a mutt, the dominant strains in my case being Irish and Portuguese, with some additional English and German blood on my mother’s side.
Riffing off my name, my paternal grandfather liked to introduce me around town as “Patrick Joseph, the only Irish-Portugee in Falmouth.” That would be Falmouth, Massachusetts, on Cape Cod, where the Joseph surname is widely understood to be Portuguese.
Grandpa Joe had never been to Portugal and didn’t speak the language. His grandparents had emigrated from the Azores, the Portuguese islands in the middle of the Atlantic. His parents, too, were of Azorean descent. He himself married an Irish girl: Marie Halligan, my grandmother. This was in the days when some Southern Europeans were still discriminated against. The Irish side, I learned, boycotted the wedding.
As a young kid on the playground, I chose simplicity over a hyphenated heritage. When asked, “What are you?” I identified myself as Portuguese, mostly, I think, due to my fondness for my grandfather but also to what you might call cultural DNA, the kind that expressed itself at the dinner table—linguiça and chouriço, kale soup, Portuguese sweet bread, etc.
So perhaps you can imagine my surprise when the results arrived from Ancestry, and my “ethnicity estimate” showed no Portuguese whatsoever.
Meanwhile, my wife, who is Peruvian, had 19 percent Portuguese in hers—something she’d never been aware of. Our kids, who had grown up eating my “Portuguese fried bread” for breakfast, found this amusing, as did my in-laws. Myself, I struggled to accommodate the new information. I don’t want to exaggerate: It wasn’t earth-shattering, but it was—what’s the word—unsettling. All these years I had claimed to be Portuguese. If I wasn’t—not even a little—then who was I really?
AS IT HAPPENS, I GOT MY RESULTS around the same time Massachusetts senator and presidential candidate Elizabeth Warren released her own DNA test to the public—a move made in response to long-standing questions about her claimed Native American heritage and relentless taunting by President Trump, who insisted on calling her “Pocahontas.” While Warren claimed to be vindicated by the results, they in fact showed only a distant relationship, between six and ten generations removed. Many Native Americans were angry that Warren released the results at all.
Kim TallBear, a former postdoctoral fellow and assistant professor at Cal, has been a particularly vocal critic of Warren’s decision, which she said privileged nonindigenous definitions of what it means to be indigenous. First Peoples, TallBear told a reporter, “construct belonging and citizenship in ways that do not consider these genetic ancestry tests. So it’s not just a matter of what you claim, but it’s a matter of who claims you.”
Before my own results arrived, I might have scoffed at Warren’s predicament. Now I felt a certain kinship. At least she had some evidence, however scant and problematic, of the heritage she claimed. Suddenly, I did not.
THE ABSENCE RAISED TROUBLING QUESTIONS. Chief among them: Was I my father’s son? Dad isn’t always the guy on the birth certificate, after all; some estimates put the rate of “misattributed paternity” as high as 12 percent (although the real rate is probably closer to 3 percent or less). When I FaceTimed my parents to break the news, I kept a close eye on my mother’s reaction. To my relief, she looked bemused but utterly blameless; not a glimmer of guilt flitted across her countenance. Of course, that didn’t rule out some upstream infidelity. Was my father his father’s son? A Coast Guard commander, Grandpa Joe had spent long periods at sea. Had Grandma Joe been with another man in his absence? Neither were alive to ask, which was just as well. I didn’t really want to know.
In the age of the spit kit, many a well-hid skeleton has come tumbling from the closet. Secret families have been brought to light, not to mention rapists and murderers. Sperm donors have been identified by their disparate offspring, and biological parents found by the children they long ago gave up for adoption.
Buried deep in its terms and conditions, Ancestry cautions: “You may discover unanticipated facts about yourself or your family when using our services that you may not have the ability to change.” The unstated assumption is that the facts are correct, but with so many tests and so much data, it seemed possible, even probable, that mistakes might creep in. Indeed, a 2018 study published in Nature found that 40 percent of gene variants detected in “direct-to-consumer” tests were false positives.
To be fair, it also seemed possible that I was not fully understanding the results. Below the “Ethnicity Estimates” section of the Ancestry report was another category called “Additional Communities.” It suggested my ancestors were part of Portuguese migrations from the Azores (to California, not New England) but did not explain how that could be if I wasn’t Portuguese. I read the site’s white paper on the subject but found it opaque. (Ancestry did not respond to my queries for this story.)
Given my confusion, I did what seemed the only sensible thing: I ordered more kits. Soon, a brightly colored package arrived from 23andMe with the words “Welcome to You” in big, friendly letters on the front. With some trepidation, I once again drew on my salivary glands and filled the vial to the dotted line. For the sake of scientific rigor, I also ordered another kit from Ancestry, curious to see whether the results would be reproduced exactly.
Another vial, more spit, this time registered under my work address and a nickname.
WHILE I AWAITED THE NEW RESULTS, I engaged in a more old-fashioned genealogical activity: building my family tree.
Genealogy, I realize, is not everyone’s cup o’ tea. Ralph Waldo Emerson, for one, compared talking with a genealogist to sitting up “with a corpse.” The transcendentalist judged the exercise too backward-looking, too mired in the Old World to be a worthy American pursuit. It has in fact become an American obsession, reportedly the second most popular hobby, after gardening.
Ancestry, based in Lehi, Utah, allows subscribers to tap into the vast resources of the Mormon Church, which has meticulously assembled the world’s largest genealogical archive. The Granite Mountain Records Vault is housed in a climate-controlled, nuke-proof bunker dug into the mountains outside Salt Lake City. Why all that effort? The Latter-day Saints believe their ancestors, once identified, can be retroactively whisked into the Celestial Kingdom. Thankfully, even non-Mormons and heathens like me can access the database, which now reportedly contains more than 5 billion names and 90 million family trees. (Pro tip: While Ancestry.com charges for access, most of the same resources are free at the church’s nonprofit portal, familysearch.org.) Aided by that body of work, I was able, in the space of two afternoons, to populate my tree going back more than 400 years. The Joseph line quickly dried up, but my mother’s side branched like a river delta, channels reaching all the way back to the Massachusetts Bay Colony.
This, too, was a surprise. No one had ever bragged of any Mayflower or Plymouth connections. I’d always thought of us as more swamp-Yankee than DAR. I had to accommodate this new understanding as I recorded distant grandparents with wonderfully exotic names like Peleg, Meletiah, Ebenezer, and Remember. In a curious way, they reminded me of my friend Proverb Jacobs, the former Cal football player and Oakland Raider who was once dubbed the “Last of the Biblical Tackles.” Before he died in 2016, Proverb, who returned to Cal to get his Ph.D. in education, had followed his lineage back through the bottomland parishes of Louisiana to West Africa. His tree, too, was filled with names to conjure by: Ransom Venson, Creecy Griffin, Jaco Congo. In learning Proverb’s history for a story I was writing, I’d secretly taken comfort in the knowledge that my ancestors had never owned slaves.
Now I had to wonder. The first enslaved Africans arrived in New England in 1638, aboard a ship called Desire.
AS I FILLED IN THE NAMES AND BIRTHDATES of close relatives, it occurred to me that I had now given Ancestry not just significant personal information but my DNA as well. Never mind the Celestial Kingdom; in the age of Big Data, genetic testing companies are sitting on an earthly gold mine. 23andMe has already struck deals with Big Pharma. Last year, drug giant GlaxoSmithKline poured $300 million into the company. For its part, Ancestry partnered for three years with Calico, one of Google’s Alphabet companies, to study the role of genetics in longevity. That may sound innocuous, but given Silicon Valley’s well-known obsession with life extension technology (Calico’s raison d’être), news of the project brought to mind eerie sci-fi scenarios like the one portrayed in the movie Gattaca, where genetically modified humans are the superior race and the natural-born are classified as “Invalid.”
Did I want my data contributing to that kind of eugenic dystopia? No, I did not. But when the second round of results arrived from Ancestry, I pushed all such concerns aside, immediately opening the email and scanning my new profile. To my amusement, I saw that I now appeared twice in the database—as Patrick and as Pato. Twins. Or near-twins; Pato’s ethnicity estimate was slightly less French (12 as opposed to 14 percent) and slightly more Norwegian and Ashkenazi (at 1 percent apiece).
Still no Portuguese.
A week later, the results from 23andMe arrived and suddenly there it was: “Patrick, your DNA suggests that 21.5% of your ancestry is Spanish & Portuguese.” Furthermore, the evidence pointed most strongly to recent ancestry in Portugal, particularly the Azores. This was an even more granular analysis than I’d expected and, by email, I asked a 23andMe representative how it was accomplished. She pointed out that the company’s “reference populations for recent ancestor locations are comprised of over 330,000 customers.” These are grouped into more than 1,500 geographical regions, including the Azores. “We look for identical pieces of DNA that you have in common with individuals of known ancestry from around the world.”
From what I can gather, Ancestry arrives at its ethnicity estimates in a similar fashion, using its own reference population and methodology. As of this writing, however, it identifies far fewer—just over 500—geographical regions. (That number doubled at press time.) Perhaps, I thought, the Azores were a blind spot for the company, either due to a lack of classification or a lack of Azoreans in their reference data, something Priya Moorjani, assistant professor of genetics, genomics, and development at Cal, calls bias. “The bias gets introduced when you don’t have reference data for every population that potentially could be part of the ancestral mix.”
In any case, both tests couldn’t be right. Since the 23andMe results matched my own notion of who I was, I put more credence in them, but at the same time I puzzled over the precision of the new results. “We currently provide ancestry estimates down to the 0.1%,” the 23andMe spokesperson wrote, “but as our database grows and we continue to add reference datasets, breakdowns will continue to become more granular.”
The more I thought about that, the more skeptical I became. More granular than a tenth of one-percent? Was I really 21 and a half percent Portuguese? Or was it more like 21.6 percent? Or 21.4? Paradoxically, the exactitude made the report seem less reliable, not more.
Moorjani acknowledged that ancestry tests are becoming more sensitive and the analyses more precise, but also cautioned, “across all those numbers there is obviously some uncertainty.”
Finally, I began to wonder: Was Portuguese-ness—this thing I’d gone looking for—even measurable? Was any ethnicity? Writing in Scientific American, geneticist Adam Rutherford, author of A Brief History of Everyone Who Ever Lived, pointed out that, while DNA kits are good at finding close relatives, “For deeper family roots, these tests do not really tell you where your ancestors came from. They say where DNA like yours can be found on Earth today. By inference, we are to assume that significant proportions of our deep family came from those places. But to say that you are 20 percent Irish, 4 percent Native American or 12 percent Scandinavian is fun, trivial and has very little scientific meaning.”
TIRED OF ABSTRACTIONS, I decided to reach out to a flesh-and-blood cousin—one I’d never met but learned about through 23andMe. This is another feature of ancestry kits: They can show you relatives in their database, some of whom are open to being contacted. Of the 1,210 people 23andMe identified as relatives of mine, Donald Wilson of El Cerrito was the closest, geographically.
Biologically, he is quite distant. According to the site, we are likely fourth cousins (meaning we share a set of great-great-great-grandparents), or perhaps third cousins once removed (that is, sharing great-great-grandparents, but a generation apart), or even just half-cousins (meaning we have only one ancestor in common). From the information he shared in his profile, I could see he also had Azorean heritage with familiar surnames, such as Correia and Pereira (which in my family was anglicized to Peters). Our tests showed we had just 0.29 percent of DNA in common—one half-identical segment, 22 centimorgans long, situated on the 12th chromosome. If you really want to know.
Of course, the truth is that all of us share the vast majority—roughly 99.9 percent—of our genetic material. (For that matter, we share 60 percent of our DNA with fruit flies, nearly half with a banana.) But given that the human genome is made up of 3 billion base pairs, that still leaves some 3 million that set us apart. Don and I had just a teensy bit more DNA in common than anyone I was likely to meet walking down the street.
We agreed to meet for coffee one morning at an Italian bakery. It might have been an awkward encounter, but there were some immediate, small-world connections to ease the conversation. The aforementioned Proverb Jacobs, for example, had been a student-teacher at Don’s high school. Don himself was a Cal alum (Class of ’72) and a retired football coach. Like me, he had taken tests from both Ancestry and 23andMe. “In the Ancestry results,” he said, “my dad was missing.” He meant there wasn’t enough Irish in his estimate. It was a mirror version of my own experience.
I asked if he’d grown up eating linguiça (he had) and fried bread (he hadn’t). And we compared notes on our Azorean-immigrant forebears. Mine were mostly seafaring folk. His worked in factories. His grandmother was a local labor leader—a “Portuguese Mother Jones,” he called her. All in all, it was a very normal and pleasant conversation. We talked about family, we talked about football, we talked about the weather.
Before I left, we shook hands and snapped a selfie together. I thought we looked like cousins.
ONCE, MANY YEARS AGO, I TOOK A REPORTING TRIP to an Indian reservation in Washington state where I spent the morning interviewing a tribal elder with the last name Charles. In the afternoon, walking along the river, I met another Charles, who, spotting me for a stranger, asked if I was lost. When I inquired whether he was any relation to my interviewee, he responded in the distinctively uninflected manner I associate with native people in the American West. “It’s a tribe,” he said. “We’re all related.”
I’ve thought about that response a lot while researching this story, for the single most profound thing our ancestry tells us is this: Ultimately, we’re all related. I do not mean that in any mystical sense but rather in a concrete, mathematical one. Given the exponential nature of family trees (our number of ancestors doubles every generation), our pedigree quickly grows impossibly large, until branches overlap in a phenomenon known to genealogists as pedigree collapse. It’s something Graham Coop, a UC Davis population geneticist and former fellow at Berkeley’s Simons Center for the Theory of Computing, has written extensively about on his blog. While the math is a bit involved, the conclusion it leads to is both intuitive and profound. As Coop explains, “Six hundred years ago (roughly 20 generations back), I’ll have just over a million ancestors alive, a thousand years back I potentially have over a billion ancestors alive. There simply aren’t that many people alive in Europe back then, and so I’m a descendant of everyone who lived then as long as they left descendants (and vast numbers did).”
He means he’s a genealogical descendant of all those ancient ancestors, mind you, not necessarily a genetic one. If you’re of European stock, Charlemagne is somewhere in your family tree—really more like a family thicket—but the chances of being a genetic descendant are scarce, less than 1/100 million by Coop’s calculation. That’s because we don’t get DNA from everyone in our family tree. True, you get 50 percent of your genes from each of your parents, but not the same mix as your sister, unless you’re twins. You and she both got fewer genes from your grandparents, still fewer from your great-great-grandparents, of which, don’t forget, you have 16. “Forty generations back,” Coop writes, “most of your genome traces back to a random subset of around twenty-six hundred individuals out of all your millions of ancestors.”
AS I WAS PUTTING THE FINISHING TOUCHES on this essay, Ancestry updated my ethnicity estimates—both mine and Pato’s. Now it had us down as 7 percent Portuguese, with the Azores—in particular, the island of Faial—as the most likely ancestral home. There was, however, a margin of error attached to the figure. The actual percentage, I saw, could range anywhere from 0 to 30 percent.
I could only smile.
Editor in Chief Pat Joseph is the only Irish-Portuguese in the office.
From the Winter 2019 issue of California.