DNA Data Interpretation

Raw data from direct-to-consumer genetic tests can provide limited but occasionally useful information.

By Laura Beil

Illustration by Jamie Jones

Starr Gibson wasn’t interested in using her DNA for health guidance when she sent off her saliva sample in 2013. As a genealogy blogger, she cared more about what her genes had to say about the past than the future.

Curiosity, and the fact that she had to pony up only $5, led her to upload the genetic information she obtained from’s AncestryDNA test to, a site that connects users to medical literature about genes and disease. Some of the information is quirky — is your earwax wet or dry? — while other information explains how some genetic variants can influence the risk of disease.

“I wanted to see what the [service] could do, and whether I should recommend it to family members,” says Gibson, 37, who lives in La Porte, Indiana. So what did Promethease say? Her profile predicted blue eyes and curly hair. Interesting, she thought, given that she has brown eyes and straight hair.

But the report also noted variants that predispose her to Crohn’s disease and inflammatory bowel syndrome, neither of which she has, but might have something to do with her lifelong sensitive stomach. The whole experience was mostly a day’s worth of cheap entertainment, but she also came away realizing it could really feed into her occasional tendencies toward hypochondria. “It takes a lot of willpower to go through it and not imagine you have every disease on there,” she jokes.

Interpreting the Data

Promethease is just one of several services that offer insights — however tentative — into how some of your gene variants might relate to disease, appearance, and metabolism. The service uses raw data customers obtain from direct-to-consumer (DTC) tests such as, FamilyTreeDNA, 23andMe, and other sites.

23andMe, which catalogs a small portion of the genome, once offered its customers health risk profiles based on their DNA. In 2013, the Food and Drug Administration (FDA) shut down that portion of 23andMe’s business, saying in part that analyzing genes and subsequently calculating disease risks constituted a medical test. (The company now offers genetic tests in the health arena, but mostly to identify carriers of inherited childhood conditions such as cystic fibrosis; it has the FDA’s blessing to do so.)

Promethease is able to operate without FDA approval because it does not calculate disease risk or recommend actions. It simply connects consumers to its companion site, called SNPedia, a Wikipedia-like, crowdsourced repository of information on known variants and their relationships to various traits and diseases. On a busy day, some 500 people may get a Promethease report.

As Gibson discovered, the report is a straight catalog of medical literature, and that’s intentional. After all, 23andMe got in trouble for trying to go further than that.

“We’ve created this funny situation where the FDA wouldn’t allow a partial contextualization of genetic information,” says Robert Green, a medical geneticist at Brigham and Women’s Hospital. “So, you can either go to your doctor, which involves more time and expense, or you can go to Promethease, and there isn’t any context.”

Indeed, people reach out for more information and search for the context Green describes by posting on website chat rooms and forums. Some of those postings are from consumers who are happy about their results; others are from people overwhelmed by the volume of information. So, before you use a service like Promethease, know what you’re getting (and not getting) and how you might react to seeing a partial glimpse of your genome.

Operators of sites such as Promethease say they are offering a public service. They contend that connecting the public with available scientific knowledge about their genome is just one more way for people to learn about themselves, and there’s no reason such information shouldn’t be easily obtainable. That said, some genetic variants have hundreds of papers devoted to them, says Green. “No regular person is going to read through [all of] those in some way that makes sense.” Still, even if the sheer complexity of the literature swallows you whole, that’s not necessarily a bad thing. “I think learning about genetics, even the complexity of genetics, is useful.”

The services are not without doubters, who say the current state of knowledge is too preliminary and vague to be helpful. Skeptics say that scientists don’t yet understand enough about the human genome to know its role in complex diseases such as diabetes, cancer, or cardiovascular disease.

“The data that they give you are, by and large, worthless for understanding your health and guiding your health care,” says James Evans, a professor of genetics and medicine at the University of North Carolina in Chapel Hill and the editor of the journal Genetics in Medicine. Evans, a longtime skeptic of DTC genetic testing, worries about companies that may overpromise about the value of their results.

No More of a Diagnosis than Google

Promethease doesn’t calculate risk with a tidy score. Instead it offers unvarnished medical literature. Consumers can browse the information themselves on SNPedia. Gibson, who had her genome analyzed, found this feature particularly helpful. “If it’s really bothering you, you can look at the research behind it to see how good it is,” she says. “You can see that one has a lot of research behind it, and a lot of money thrown at it. But another one doesn’t.”

That’s the kind of thoughtful evaluation the founders of Promethease hope consumers will do. The site was established 10 years ago by Greg Lennon, a geneticist, and Michael Cariaso, a computer programmer, who worked together at genomics start-up GeneLogic in the late 1990s and early 2000s. “We don’t believe there is generally a single, simple answer like ‘Your risk of disease X is Y,’” they wrote in an email. “Instead, we connect you to the scientific literature.”

And they agree that the quality of studies varies by design and rigor, and users should take that variability into account. “We take your variants and query our database to select the subset of literature relevant to you,” Lennon and Cariaso wrote. “Our reports provide features to make it easier to search, sift, sort, and filter those results, but they don’t make medical diagnoses any more than a Google search does.”

Scientists are still studying the ways in which knowing about your genome influences your choices. Green, who is also affiliated with Harvard Medical School, has long researched the connection between genetic knowledge and personal behavior. As is often the case in science, he can’t give a definitive answer. The strength of the connection between knowledge and action varies. “How much do people understand? What do they take away? At what level of concern are they prompted to make a change? How long does that change last? These are all questions we’re asking,” Green says.

We’ve created this funny situation where the FDA wouldn’t allow a partial contextualization of genetic information. So, you can either go to your doctor, which involves more time and expense, or you can go to Promethease, and there isn’t any context.

Experts like Mike Watson, the executive director of the American College of Medical Genetics and Genomics, aren’t telling people to stay away. Rather, they caution against considering services like Promethease as a kind of medical record. “If you’ve got the money and the interest, knock yourself out,” he says. “But don’t act on it. If you are going to make clinical decisions based on the information, you need to find someone who is trained and experienced in interpreting genomic information.” For example, you may learn of variants that suggest your body metabolizes a certain drug faster or slower than average. This interplay between genetics and medication, called pharmacogenomics, is a promising field that is “still in its infancy, but has enormous possibilities,” said a 2016 review in Clinics in Laboratory Medicine.

In an article published in June 2014 in the journal Clinical Pharmacology and Therapeutics, doctors from Harvard Medical School cautioned that pharmacogenomics should be guided by clinicians; they described a patient who was misled by a genetic finding from a DTC testing company that claimed the patient had a reduced ability to process certain drugs used to treat autoimmune disorders. Such examples, they say, “illustrate risks associated with [direct-to-consumer] testing without professional interpretation.”

The problem, in James Evans’s view, is that even many primary care physicians aren’t always sure what to make of a genome analysis. “It could divert attention from more important matters at hand,” he says. “It could lead to downstream testing that costs money. Since we’re all in this together, I have a vested interest in making sure you don’t get tests that you don’t need.”

So far, there isn’t evidence that these tests have led to a population of worried well, says Lawrence Brody, the director of the Division of Genomics and Society at the National Human Genome Research Institute. And another fear — that someone with a “healthy” profile will feel so bulletproof he or she will, as Brody puts it, “smoke for breakfast, eat pork rinds for lunch, and [drink] two martinis for dinner” — also has not come true. In one study he and his colleagues carried out, about 80 percent of study volunteers understood that for complex conditions like heart disease and cancer, genes are not destiny.

Starr Gibson would agree. She’s happy she tried Promethease, but mostly because it was amusing to get a small peek at her genome. “Use this just for fun,” she says. “If you plan to use this for medical reasons, get a doctor.”