Such wholesale disclosure of personal information could end up being dangerous for participants, Church acknowledges. They could find it hard to get health insurance or suffer other discrimination if, for instance, their genes reveal susceptibility to illness. "Some of them know they're going to get hurt, like astronauts and mountain climbers. But if enough of them see a benefit to themselves, their families, and society, then it will keep growing," says Church.
Share alike. The GeneTree website that Critchfield is using, which debuted in October, links genetic genealogy to paper genealogical records. The company is an offshoot of the Utah-based Sorenson Molecular Genealogy Foundation, which has collected DNA samples from people around the world to build a database for genealogical research. Critchfield found about 300 names to add to her genealogy on the GeneTree site, and hopes to get more from people she has E-mailed.
The notion of sharing genes on the Web debuted last year, when both Craig Venter, a leader in efforts to sequence the human genome, and James Watson, who shared a Nobel Prize for figuring out the structure of DNA, posted their genomes online. That quickly turned into a lesson in how genes can prove to be an embarrassment. In October, Watson made remarks disparaging the intelligence of Africans. In December, Stefánsson analyzed Watson's public genome and showed that 16 percent of the scientist's DNA came from a recent African ancestor, the equivalent of having an African great-grandfather.
Those kinds of surprises are not uncommon on the rough frontier of retail genetics. Eight years ago, Henry Louis Gates Jr., a professor of African-American studies at Harvard University, was among the first African-Americans to have his DNA tested in an effort to fill in the historical record of the slave trade. That test said his mother's family had Egyptian roots. But when he sent his DNA to a second company in 2005, Gates was told his maternal lineage was not African at all but European. It turned out that because the first company's DNA database had samples almost solely from Africa, it didn't pick up on the big surprise. Since then, Gates has turned to traditional genealogical research to make sense of the conflicting genetic data. "Most likely," he says, "I'm descended from an indentured servant who slept with a slave in the 17th century." In his new PBS series African American Lives 2, which will air in early February, he'll recount his search for that woman.
Gates recently joined with Family Tree DNA of Houston to launch AfricanDNA.com, which will augment DNA analysis with legwork by historians and anthropologists. "I want African-Americans to identify with Africa," he says. "I just don't want them to misidentify with Africa." He is philosophical about the glitches in his own genetic search. "I want people to be amused by this, intrigued by this, rather than defined by the process. It's a process still in its infancy."
Genetic analyses can be wrong for many reasons. DNA samples are easily contaminated; technicians can be sloppy; the technology is new and constantly changing; and the machines can generate "phantom mutations" (mistakes). Decodeme's Stefánsson points out that the startling discovery of African DNA in Watson's genome could be wrong because the original scan, done by another company, contains errors. (Some parts show Watson having two X chromosomes, impossible for a man.) Stefánsson says: "Genome sequencing, although it has taken enormous steps forward, is not yet there."
Tests for ancestry compare a person's DNA with that of others, so the larger and more accurate a genetic genealogy database, the more precise its results. (The accuracy of genome-based disease risk tests also depends on database size, as well as the accuracy of research that links certain mutations to specific diseases.) The National Geographic Society's Genographic Project, which has been collecting DNA from volunteers since 2005, has about 225,000 samples of mitochondrial DNA, which traces maternal lineage. Africandna uses 58,000 samples of mitochondrial DNA; GeneTree uses 53,000 mitochondrial DNA test results from the Sorenson database; 23andme tests against 1,000 samples. Companies don't always detail the size of their databases in promotional literature, and many customers don't know to ask. Hank Greely, a law professor at Stanford University and an expert on ethics and bioscience, says the industry all too often has the trappings of science without the substance. "Scientists tell you the limitations; they tell you what you're getting and what you're not," he says. "These guys don't."