Putting the Gene Back in Genealogy

Your DNA holds the secrets of your ancestry, and at least a dozen companies offer to crack the code. But there's more than a bit of hype here.

Illustration by Polly Becker

Writer Rebecca Skloot (center) turns to DNA to track down a broken branch in her family Tree.Illustration by Polly Becker

It all started, as so many quests do, on the Internet. Just a random link on a random Web site, but something made me click it. And before I knew it, I was scrolling through what felt like an online genealogic red-light district, a sensory overload of temptation and promises of fantasies fulfilled: "Use DNA to illuminate the past and bring ancient ancestry to life" ... click ... "The only way to know where you come from is by reading your DNA" ...
click ... "Are you adopted and curious about your heritage? Recent advances in genetic testing have put the answer to questions like these at your fingertips" ... click ... "See if you have a long-lost relative" ... click ... "Identify your ethnic and geographic origins, both recent and far distant."

For anywhere from $219 to $1,200, depending in part on the level of accuracy I required, companies such as Oxford Ancestors and Genelex were offering to trace my family back thousands of years to a remote prehistoric mother, to tell me precisely which ethnic groups I belonged to and in what concentrations -- whether, for example, the blood of Native Americans, Jews or Vikings ran through my veins. One company promised -- for no additional charge -- to use my DNA to put me in touch with "Genetic Cousins" I never even knew I had.

As fantastical as these come-ons sounded, for some reason
I kept clicking on one link after another. What can I say:
I am -- at least as much as any other American mutt -- a
sucker for any information that might patch a hole or two in my family history. And as it happened, I'd just learned of one intriguing personal genealogical mystery I was interested in solving. So I kept on exploring.

On the other side of those links I found companies whose advisory boards were packed with seriously credentialed scientists -- population geneticists, molecular anthropologists, you name it -- from prestigious universities. To back up their claims, the companies linked to research published in
highly respected peer-reviewed journals like Nature and
Science and media reports from The New York Times and Nova. And I thought, Huh, maybe there's some legitimate science here.

So I picked up the phone.

Hey, Mom, I said, will you send me some of your DNA?

Until the dawn of the internet age, a person trying to create a family heritage map, fill gaps in a family tree, or even answer a single question about an ancestor generally followed the model of my distant cousin Les Hickenbottom. He has spent much of his retirement at genealogical gatherings -- surrounded by people whose RVs sport "I Brake for Graveyards" bumper stickers -- or in various courthouses and city halls around the country, sifting through birth, marriage and death certificates, wills, land deeds, and court records, panning for any information he can find about our family. This approach is known as paper-trail genealogy, and the Internet has made that trail (in digital form) dramatically easier to follow: The National Archives put 50 million documents online in April 2003, and when Ellis Island posted its immigration files on the Web in April 2001, the site received eight million hits in the first 6 hours.

But you can't digitize what doesn't exist, and sooner or later most genealogical hobbyists hit an unbreachable gap in the historical record, a place where the paper trail simply vanishes: Last names change, people move, adopt, disappear. Suddenly, there are no more documents to follow -- history seems to stop dead in its tracks. It's what genealogists call a brick wall in the paper trail.

My own brick wall started to materialize about three years ago, when my then 87-year-old maternal grandfather, Robert Lee, decided it was time to pass on a piece of family lore. It was a tale he'd kept to himself since one night in 1942, when he sat with his aunts as they rocked slowly in their chairs after supper, spinning family stories. Grandma Elenor Hickenbottom, one of them whispered, she was a black woman. This was the early 1940s in southern Illinois -- a time and place when people whispered about things like mixed race, if they talked about it at all. My grandfather searched out and found Elenor's grave -- a small stone hidden in the woods behind the church -- and visited it each year without fail, but he didn't speak a word of her existence for nearly 60 years. "It's something I've wanted to talk about
for a long time, but I couldn't," he told my mother and her brothers. "My brother Ronald and I were the only ones who admitted the truth; now he's dead, so I just have to tell you, even if it makes you mad."

Of course, it didn't make us mad -- just curious. Were we truly descended from an African-American woman? Where had she come from? Had she been a slave? Unfortunately, the paper trail was useless in this case -- in part because no one knows Elenor's maiden name; whenever she appears
in historical documents, it's by first name only. As circumstantial evidence in support of my grandfather's story,
this omission is actually compelling: Many African-Americans who ended up in Illinois were born in the South
with no birth certificate, fled north to freedom, and proceeded through life without documentation. Circumstantial
evidence, though, hardly constitutes proof.

But what if the proof resided elsewhere? What if it resided within my family's cells?

I settled on two companies to run the tests. I chose the first -- Florida-based DNA Print Genomics -- because it developed and is the main supplier of the one test that analyzes a person's DNA for evidence of ethnic heritage. The company would tell me what percentage of my DNA (if any) was Native American, Indo-European or African. The second company -- Family Tree DNA -- offered the largest selection of tests and claimed the greatest accuracy by testing more genetic markers than anyone else. Family Tree DNA was also the only company offering to find real flesh-and-blood "long-lost relatives" for me -- not exactly the object of my inquiry, but intriguing nonetheless.

To understand the tests these companies and their competitors offer, it's important to first ground yourself in some basics. Different tests rely on different kinds of DNA. One, nuclear DNA, is what most people think of when they imagine DNA -- it's passed from mother and father to child and influences everything from eye color to height. Each strand of nuclear DNA is a sequence of more than 3.2 billion letters (all either A, T, G or C). Nearly 99.9 percent of that sequence is identical in all humans, but 0.1 percent varies in specific locations on the genome -- a T changed to a C or a G to a C. These differences account for the vastness of human diversity. Most of these mutations are single nucleotide polymorphisms, or SNPs ("snips") -- generally harmless changes that occur spontaneously and are passed from generation to generation. Some populations, Africans among them, evolved in particularly isolated groups, passing unique SNPs on to each generation. Over time, this created what scientists call "ancestry-informative markers," particular SNPs where, say, West Africans are more likely to have a C and Europeans are more likely to have a T. These ancestry-informative markers are what DNA Print Genomics uses to determine what percentage of four ethnic groups each customer descended from.

Then there's the DNA on the Y chromosome -- a small sex chromosome found only in men. Of the 23 pairs of chromosomes in the human body, the Y chromosome is the most
stable. Other chromosomes swap DNA regularly, but the Y chromosome is passed almost entirely intact from father to son like a sealed time capsule. Because of this, it's a window into male heredity -- when you look at a man's Y chromosome, you're also looking at the Y chromosome of his father, his father's father and so on through thousands of generations. You're looking into ancient history, to an ancestor who lived 100,000 years ago or more. Like SNPs, specific Y DNA sequences are associated with certain groups that evolved in isolated populations, passing certain mutations from one generation to the next until they were prevalent -- creating, for example, the so-called Native American and Viking Y chromosomes.

And finally there's mitochondrial DNA -- the female
ancestral equivalent of the Y chromosome. Every person, male or female, inherits mitochondrial DNA (mtDNA) from his or her mother through cellular organelles called mitochondria, which produce energy for cell functioning. It's a packet of DNA that, like the Y chromosome, remains essentially intact. So when scientists look at a person's mtDNA, they're looking at that person's mother's mother's mother and so on, back to the Ice Age. All mtDNA sequences found in humans fall within 36 different types called haplogroups, and Bryan Sykes, an evolutionary geneticist at Oxford University, has traced these sequences back to 36 hypothetical "mothers" who originated in different parts of the world. (He used this DNA research as the basis for his best-selling book The Seven Daughters of Eve.) So, depending on which mtDNA sequence you have, companies that run this test can tell you whether you descended from Ursula, who lived 45,000 years ago in Greece; Helena, who lived 20,000 years ago in the northeast Pyrenean foothills; or 34 others from around the world. And once Family Tree DNA has identified your mtDNA sequence, it will enter that information into a database full of sequences from more than 3,000 people to put you in touch with your Genetic Cousins -- a term they coined (and trademarked) for people with similar Y chromosome or mtDNA sequences.

By the time I developed this level of understanding about the various tests, it became clear that in two out of the three cases -- Y chromosome and mtDNA tests -- the results were far more likely to illuminate what might be called genetic anthropology (a genetic snapshot of, and connection to, people who lived thousands of years ago) than the classic genealogy of family trees and ancestors with only a handful of "great"s in front of their names. Still, according to Martin Richards, an evolutionary geneticist at the University of Huddersfield in the U.K. who did some of the mitochondrial DNA research that led to the mtDNA test, African ancestry should be among the easiest things to test for. "There are a lot of specific genetic differences between Africans and non-Africans," he told me, "which means if you have any African ancestry, it stands out." Similar differences exist in most ethnic groups. They're just more prevalent in Africans because of the isolation in which the continent's population evolved. (Like most scientists I talked to, Richards is quick to point out that these differences have nothing to do with superiority or inferiority.) I asked the folks at DNA Print Genomics whether they could look at my DNA and tell me if, say, my grandfather's great-great-grandmother was a black woman. From your DNA, it's doubtful, they said, but from your grandfather's DNA, quite possibly. When I asked the founder of Family Tree DNA the same question, he didn't hesitate: "Yes," he said, "absolutely."

So I ordered the tests: mitochondrial and Y chromosome tests for my maternal grandfather from Family Tree DNA, to learn about the ancient men and women he came from, and to take a shot at finding long-lost Genetic Cousins. The same for my father, to get a full picture of my genetic history from both sides. Then from DNA Print Genomics: nuclear DNA tests all around -- one for me, my brother, father, mother and her parents -- to see what ethnic groups we came from.

The companies express-mailed DNA collection kits around the country to my family members. The next morning, all six of us rubbed the insides of our cheeks with something that looked like the spawn of a Q-Tip and a toothbrush. (After 30 seconds of such cheek brushing, the instructions said, these devices would be swarming with hundreds of thousands of cells.) Then we carefully sealed the swabs in envelopes and tossed them in FedEx drop boxes thousands of miles apart.

The craziness started a few weeks later with a phone call: "Rebecca? Phil Brooks here with DNA Print Genomics. Did you know you're Native American?" Um, no. "Well you are -- 10 percent Native American, 90 percent European." Ten minutes later: "Rebecca? Phil Brooks with DNA Print Genomics again. I need to talk to you about your brother." Then he lowered the phone, and I could hear him whispering to someone in the background. "Do you want to tell her or should I?" Then commotion, several voices, one man saying, "No, just tell her what we found," and the lab manager, Matt Thomas, suddenly on the line launching into a lecture about basic genetics. "Children are supposed to have a combination of genes from two specific people," he said. Suddenly it hit me: I forgot to tell them that my brother and I have different fathers. I wasn't trying to trick them, I just forgot. My brother was four when I was born; he's always just been my brother. "Are you trying to tell me my brother isn't my brother?" I asked after several minutes of listening to him stumble. Thomas inhaled sharply. "Um ... if I were, would that be true?" Once I'd explained, he lowered the phone again and yelled: "They have different fathers!" The room erupted in cheers. "We had a little bet going," he said sheepishly. "I just lost."

As they cheered, an automated e-mail from Family Tree DNA flashed on my screen: "You have a match!" it said, pointing me to a Web site where I could get contact information for my new Genetic Cousins. Then another e-mail: this time from a woman named Patricia Baker who had gotten the same automated e-mail; she was welcoming me to the family. Patricia was one of Family Tree DNA's first customers; she had
tested herself hoping to learn about (and, ideally, find) her father, whom she'd never known and whom her now-deceased mother had refused to discuss. Patricia, I learned, gets excited with the arrival of every new e-mail from Family Tree because she hopes they've found someone who can help find her father. Apparently the company didn't clarify one crucial point: The test she had done -- a mitochondrial test -- only provides information about her mother's side of the family. The test Family Tree offers that could shed light on Patricia's father is a Y chromosome test, which Patricia can't do, because she has no Y chromosome and no known male relative.

My next e-mail was from Bennett Greenspan, founder and president of Family Tree DNA. "I'll tell you exactly which of your relatives were Jewish," he wrote, "and exactly where they came from. How's that for confidence?!" Impressive indeed, particularly since I had never let on that I came from Jewish stock. Within a few hours, he called. "You've got beautiful DNA," he whispered into the phone. Then: "These results are going to be more than you can handle." Turns out, he said, I have Jewish ancestors on my father's side from Eastern Europe (which I do); my maternal grandfather's mitochondrial DNA is the most common European DNA out there, and my grandfather has the same kind of Y chromosome as my father, which sure sounded like he was saying my mother and father were somehow related.

DNA Print's analysis soon followed, and showed that my family is European, Native American and East Asian -- everything on the list except African. My grandfather registered
as 100 percent European -- no trace of Elenor. Amazingly enough, they told me, my mother and grandmother came back with identical genetic balances: 87 percent European; 13
percent Native American.

At this point, I'm thinking: If these tests are right, my
mother is a clone of my grandmother, I'm Native American enough to qualify for a few scholarships, there's no trace of my African ancestor, and since my mother's father and my father share the same Y chromosome, I'm the offspring of some strange sort of inbreeding. Things could hardly get weirder. Or so I thought.

Enter the Cousins: Within days of talking to Patricia Baker, I found myself on various
e-mail lists with many of the 81 other total strangers Family Tree says I'm related to. Our families came from different parts of the world and never shared a single surname, but the excited Cousins filled my inbox with family trees, personal stories, genetic histories, anything that might help find our family connection. These Cousins were German, Irish, English, Dutch, Austrian, French, Native American, Greek, Ukrainian and nearly everything else, and most of their messages were variations on a theme: "My mother was a foster child" ... "I guess we are cousins on our mother's side somewhere" ... "I (took these) tests as part of my search for family identity." One man signed himself "your
DEFINITE cousin." Sometimes Genetic Cousins have family reunions. When they meet each other, they say things like, Wow, you look just like Aunt So-and-so. And why not? According to Family Tree, their DNA says there's a 99.9 percent chance they're related.

"That's just ludicrous!" yelled Martin Richards when I read him the e-mails from Family Tree introducing me to my Genetic Cousins. "It's not meaningful to say you're related to those people any more than you're related to anyone in America, just because you have a common ancestor in the Ice Age. If you trolled the world that way you'd find more relatives than you'd know what to do with."

This doesn't mean the company is lying when it says there's a 99.9 percent chance I'm related to those Genetic Cousins, although it may be using a broader definition of "related" than its many prospective customers might assume. With each generation, the number of ancestors a person has increases by a
factor of 2 to the power of the generation. Go back five generations, adding each generation together, and you've already got 62 direct ancestors -- not counting siblings, cousins, aunts or uncles. At Family Tree DNA, the best-case scenario gives you a 50 percent chance that you and your Genetic Cousins share a common ancestor within seven generations -- a pool of 254 relatives -- and a 90 percent chance you share a relative in 23 generations, or 16,777,214 ancestors. That is to say, if you gathered enough people to populate the state of Florida today, then added about 778,000 more, you'd likely (but not surely) find a distant relative in common with your Genetic Cousin. "If you just took any two random people and looked at their DNA," says University of London geneticist David Goldstein, "you would find some part of their genomes that was derived from a common ancestor fairly recently." And by "fairly recently," he's talking less than 23 generations back. "The question is, Does it actually say much about the similarity of people if they share DNA 15 to 20 thousand years old? No, it absolutely, emphatically doesn't."

This is not the first time Goldstein has encountered a gap between the way these tests are marketed and their underlying science. In 2000 he found that within one small island population, people of Viking descent had a higher frequency of a certain genetic sequence than the non-Viking population. One company now tests for that sequence, saying people with it descended from Vikings. "We simply never said that," says Goldstein. "People sometimes write me saying, I have the Viking gene, what does that mean? I say, I haven't the slightest idea. Outside that tiny island, we have no clue what that sequence means."

It's tempting to shrug off misrepresentations and dramatic claims as the sort of harmless hype that often floods to market when a truly significant new technology or scientific discipline debuts. Goldstein, though, isn't so quick to shrug. "I really worry about a backlash," he says. "Because there is something quite serious we need to do with genetics and health care in the very near term." Goldstein envisions a time in the near future when doctors will prescribe medications based on patients' genetic profiles. "In order for us to use genetics in personalizing medicines," he says, "it's quite important that the public be able to trust the way genetic information is used. If we've oversold dramatically what we can tell people about their heritage using genetics, then there's naturally going to be skepticism about what we can tell them about their responses to drugs."

When pressed, Greenspan concedes that mitochondrial DNA isn't much use for traditional genealogy. "It's projecting back to a group that may have formed 20,000 to 30,000 years ago," he told me. "So that's more of an anthropology test." As for its use in genealogy research: If two women came to him saying, Are we related? he could give them an answer with traditional DNA mapping techniques. "Other than that," he said, "if you're looking to find your family through a female DNA test, you may as well be playing the lottery."

When I told my grandfather about our family's results, I asked if he was disappointed to see no trace of African blood. "No," he said, after sitting silently for a long moment. "It
doesn't make me feel anything special, except that I think there was something wrong with that test."

Was there? "She might have been within the range for the test to pick her up," said Mark Shriver, a consultant for DNA Print and assistant professor of anthropology at Penn State, when I asked him how to interpret the results in light of family lore. "Unless she was a light-skinned African. If she was 100 percent African, that would give your grandfather 6.25 percent; if she was light-skinned, that would decrease his number, but he should still have some. It could be that by chance he didn't inherit her African genes, or maybe she's right outside the detectable range."

Or maybe it's a result of limitations in DNA Print's database. "It's possible to learn something about a person's ancestry with autosomal (nuclear) DNA testing," says Noah Rosenberg, a genetic anthropologist at the University of Southern California. "But you have to be very careful. When they say you're 90 percent European and 10 percent Native American -- that's based on samples they collected from people they decided were representative of Europeans and Native Americans. So really what they've told you isn't what ethnicity you are, but how similar you are to the people in that database." If you change the database, you change the results.

DNA Print built its database from 3,000 SNPs. To ensure purity, they didn't use, say, African-Americans, they used native Africans. But, as Mark Shriver says, there's no way to know for sure who's purely African, European or anything else. "You just try to collect the best samples you can," he says. "It is important to realize that we're not measuring anything essential or absolute about the DNA, we're just referencing it to samples that we know more about."

Most of the geneticists I talked with look forward to the day when there is a worldwide DNA database large enough
and diverse enough to legitimately represent different populations. With such a database, they say, providing precisely accurate information about a person's genetic heritage could be possible. Meanwhile, geneticists make do with various public and private databases; the largest -- an NIH database called dbSNP -- contains four million SNPs. The frequencies of some of them are being measured in populations including East Asian and African-Amercan. But scientists involved emphasize that this isn't intended for personal genealogic testing. "We're collecting this data for medical research," says Harvard geneticist David Altshuler. "What these companies are doing is, in my opinion, an unfortunate consequence of that research." The NIH's dbSNP may someday be a valuable resource for, say, studying population migration on a global scale. But it's nowhere near the size required for applying it to an individual.

"In no way am I saying their product is totally bogus," says Ranajit Chakraborty, a population geneticist at the
University of Cincinnati. "It has some scientific basis." If a person has many ancestry-informative markers that are common to Africa, there's a good chance that person is of African descent to some degree. But even that can't be stated with absolute conviction. "You can say one group has a certain population-
specific variation, but it doesn't mean that mutation is not found in other
populations as well," Chakraborty says. "So you cannot be very precise. If I take your DNA and type it and say you're 100 percent European, that does not mean you don't have African origin genes somewhere in your genome."

Ultimately, every expert I talked to -- even those affiliated with the ancestral DNA companies -- agreed that no DNA test can make definitive statements about ancestry (such as "Rebecca Skloot is 90 percent European and 10 percent Native American"). But nuclear DNA testing can tell me I have some genes that are more likely to be Native American than something else. That information may be worth a few hundred dollars to some people, because it just might help scale a few brick walls. "Beliefs tend to drive traditional genealogy research," says DNA Print's founder, Tony Frudakis. "If you think your ancestors were European, you'll only look in European archives. These tests can open new paths to follow in genealogy research."

But one thing is certain: To date, no ancestry test can rule out part of a person's history simply because it doesn't detect it. When I told Noah Rosenberg that my DNA tests found no evidence of Elenor Hickenbottom, he chuckled. "Let's put it this way," he said. "I wouldn't discount your family story just yet."

Contributing editor Rebecca Skloot's first book, The Immortal Life of Henrietta Lacks, is forthcoming from Crown.