Illustration by Jillian Tamaki
This past September, Silicon Valley startup 23andMe took another step in its self-described mission of “democratizing personal genetics” when it slashed the price of its gene-scanning service from $999 to $399. Increasing its user base is part of the company’s goal of becoming a robust platform for web-based genetic research. And by starting to collect, compile, and compare phenotypic and environmental data from these users, 23andMe is catalyzing DNA’s move from labs to Facebook tabs.
23andMe.com has the feel of a social-networking website, but its users are sharing genetic information rather than photographs. For $399, you don’t get a full genome scan, but rather a look at 580,000 different single nucleotide polymorphisms (SNPs): the genetic variants most often tied to distinctive personal traits. Users can elect to share this information with family and friends via their site-based profiles. 23andMe’s competitors, deCODEme and Navigenics, have different price structures and focus on different selections of SNPs, but the basic premise of their services is the same.
Though these services have been heralded by some as a revolution in personal genetics, their utility is limited. While they all test for SNPs related to medical conditions, like Crohn’s Disease and breast cancer, the companies caution against using their reports in making healthcare decisions without first seeking professional advice. Other SNPs are trivial; 23andMe, for example, will show you the genetic basis for your earwax type.
The company, however, is aiming to be more than an expensive Q-Tip. “One person’s genome isn’t all that meaningful,” says Linda Avey, who cofounded 23andMe in December 2007. “But many people’s genomes catapult us ahead in terms of research.” Customers can keep their genes hidden from other users, but by submitting a saliva sample, they consent to have their anonymized SNPs compared in studies on a wide variety of traits and conditions.
With the costs of gene scans dropping rapidly, 23andMe isn’t the first to attempt such a broad collection of association studies. George Church’s Personal Genome Project is aiming to collect the whole genomes — not just SNPs — of 100,000 people. But looking at a pool of genes isn’t particularly interesting unless you know how those genes are expressed. In addition to a large sample size, such studies need phenotypes along with genotypes in order to be clinically useful. Church’s project will employ a network of physicians to collect DNA samples as well as other personal and biological data.
The challenge for 23andMe is how to replicate this process when its contact with subjects is limited to a vial of spit and an internet connection. To that end, 23andMe unveiled its research sector: 23andWe. Customers are now surveyed on physical traits, like whether they’re right-handed, generally optimistic, or can stomach the taste of broccoli. With the ideals of Web 2.0 as a guiding principle, 23andMe has described its user-generated matching of genotype and phenotype as “Research 2.0.”
When people are sick, they aren’t as concerned about privacy as those who are not.
While self-reported data collection presents its own set of problems, 23andMe’s approach has already attracted research partners, such as the Michael J. Fox Foundation for Parkinson’s Research (MJFF) and the Parkinson’s Institute. “Parkinson’s is a pretty complex mix of genes and environment,” says Dr. Brian Fiske, MJFF’s associate director of research programs. Though several related genes have been identified, which polymorphisms are tied to which aspects of the condition is largely a mystery. So beyond adding their patients to 23andMe’s pool, the research groups are helping to design 23andWe surveys that will measure a number of Parkinson’s symptoms. And with web-based motion tests that measure how quickly and steadily users can click a mouse, customers may eventually provide direct experimental data on muscle control.
Fiske says that 23 andMe presents “a unique opportunity” to reenvison future research. The biggest question that remains is “whether we can do surveys online that draw out r igorous data,” he says. “The jury is still out as far as the research community is concerned. But suddenly all the data is electronic and you have access to far more people, which is exciting.”
Tying personal information to the genetic profiles in 23andMe’s digital database might sound unsettling, but Avey argues that “when people are sick they aren’t as concerned about privacy as those who are not.” That logic is the basis for other bio-networking websites that have sprung up in recent years, such as PatientsLikeMe. Its users connect with others who have the same medical condition, then share their test results, pharamaceutical regimens, the progression of their symptoms, and other information. “We call ourselves the first phenome company,” says co-founder James Heywood.
Where 23andMe and PatientsLikeMe will meet in the not-too-distant future, he says, is marrying the former’s genetics with the latter’s longitudinal data. Adding more detailed medical information could open the door to clinical research on par with traditional models. With users of Google Health and Microsoft HealthVault already uploading this kind of information to the web, it may eventually be just a matter of integration.
But Heywood also believes that mashing up genomes and phenomes is only the beginning. Digital profiles could be easily be compared with other electronic resources — GPS data, or online weather feeds, for example — which could add another dimension of analysis for many conditions for which environmental factors are relevant.
For some, however, there still isn’t enough justification for signing away one’s DNA. Arthur Caplan, director of the University of Pennsylvania’s Center for Bioethics, derides these direct-to-consumer services as “Spitomics,” a cookie-cutter approach that he claims eschews personalized advice and the ability to ask detailed questions. And with business models that rely on paying customers, these services can’t control the size and scope of their data sets, leading to gaps and biases. “If you’re Asian-American or Indian-American, they might not have anybody in the database like you,” says Caplan.
23andMe declines to say how many clients have opted in, but it has its eyes on millions. “Genetics today is a very low-resolution picture,” says PatientsLikeMe’s Heywood, “but the only way to get started is to get started.” Flouting concerns over genetic privacy, individuals are taking those first steps. For example, users of the wiki SNPedia have begun sharing the raw data from their 23andMe, Navigenics, and deCODEme profiles using a homebrewed program called Promethease.
Whether such internet-driven associational studies can produce real findings, or can do so before stronger regulation on informed consent catches up with them, remains to be seen. But if 23andMe’s social-networking forebears are any indication of how quickly such technology can take hold, we may be recalibrating our conception of genetic research and privacy sooner than we think.
Originally published February 27, 2009