The microbiome has been big news lately. Ninety percent of the DNA in a human body belongs to the microbes in the gut, mouth, and nose, and on the skin and genitals. Estimates of the weight of all those microbes differ, with some researchers saying it could be up to five pounds—but whatever they weigh, it’s now widely agreed that they are effectively an organ of the human body.
For example, intestinal disorders such as irritable-bowel syndrome are now thought to be partly due to imbalances in gut bacteria, leading to innovative treatments such as transplanting stools from a healthy volunteer to a sick person to repopulate the intestines. Patients with chronic sinusitis have been found to have less bacterial diversity in their noses than do healthy subjects, suggesting that they might be helped with probiotics put directly into their noses. It’s even thought that there is a connection between gut bacteria and metabolic disorders such as diabetes.
That’s why scientists are eager to study the human microbiome. A first-line task is to characterize its composition. The federally financed Human Microbiome Project, which got under way in 2008, had collected samples from 238 subjects in the United States as of 2012.
Now garage-style start-ups such as American Gut and uBiome have jumped in as well, using new and relatively cheap tabletop devices that can rapidly identify bacterial species by sequencing parts of their DNA. They present their work as “citizen science,” offering the possibility that nonscientists can help map the microbiome by sending in samples from their own bodies.
Scientific endeavors ranging from astronomy to zoology have in recent years taken advantage of the wisdom of the crowd, such as by recruiting citizen volunteers to identify fish specimens or to sift through satellite data looking for signatures of exoplanets. But when it comes to sequencing the microbiome, some scientists ask, how much will crowdsourced research really contribute to our knowledge?
Data Bias at Issue
In February 2013, uBiome, a company based in San Francisco, finished raising $351,000 through indiegogo, a crowdfunding Web site. The American Gut project has raised $163,500 in similar fashion. Anyone can donate $90, along with a sample, and allow scientists to use the data in peer-reviewed publications. In exchange, donors get to see their own personal data.
At uBiome, says Zachary Apte, the company’s chief scientist, contributors will get a list of the microbes in their sample and their relative abundance, along with phylogenetic trees relating each bacterial species to the others. He says uBiome will also relate the list to each user’s self-reported information on diet, health status, and antibiotic use.
Skeptics, however, say that crowdsourced data from self-selected participants may not lead to good science. They have raised questions about how much the “citizen scientists” will learn that is personally useful. Melissa Bates, a researcher at the University of Wisconsin School of Medicine and Public Health, wrote on the Boundary Layer, a blog she co-authors, that uBiome’s approach “biases your data by limiting participation to people who can afford to participate.”
Janet D. Stemwedel, a philosopher of science at San Jose State University, agrees the data sets could be misleading. “There might also be a skew toward people with particular illnesses or medical conditions who have been led to believe that knowing something about their microbiome is going to be useful to them in their treatment,” she says. “I don’t know what that’s going to do to the samples. But I can’t imagine that it’s going to result in a sample that’s really representative,” even of people in North America.
But uBiome’s Mr. Apte argues that biases are always present, regardless of sampling methodology, and can be taken into account with demographic and statistical analyses. The company has the numbers on its side, he says, with 3,000 users, from more than 30 countries. (American Gut’s crowdfunding Web site said last week that it had 1,060 contributors.)
Justin L. Sonnenburg, a microbiologist at Stanford University, notes that traditional clinical studies, too, are vulnerable to sampling problems like biases. “The potential pitfalls of this sort of study are not unique to this citizen-science approach,” he says.
In an effort to increase sampling diversity, both uBiome and American Gut have offered contributors a chance to buy two kits, the second of which would be donated for use in developing countries. But only 47 of uBiome’s contributors and seven of American Gut’s had selected that option as of last week.
How Useful Is It?
While uBiome bills its work as basic research rather than diagnostic testing, Ms. Bates, of Wisconsin, asks if it is overselling its medical value. “Granted, the investigators do state that uBiome is not a diagnostic test,” she wrote on her blog, “but they couple it with statements that the results should be discussed with the participant’s doctor and that this information may be useful data in treating difficult disease.” She thinks that physicians would hardly know how to use such information.
In fact, microbiomes differ widely not only from person to person but from region to region, which would complicate any effort at medical use. For example, a study of children in an African village found that a large percentage of their gut bacteria were of the genus Prevotella, which was absent in European children of the same age. The study’s authors hypothesized that the Prevotella evolved with the population to maximize energy intake from the local diet, which is rich in fiber, while also warding off local causes of inflammation and colon disease. Interesting—but it’s not yet clear what a healthful microbiome is, let alone how to change it to help people with illnesses such as diabetes.
But Stanford’s Mr. Sonnenburg is optimistic that the start-ups will provide good material for researchers, arguing that the companies have reputable staffs and are using the latest equipment.
“It’s very likely that the data that is generated will be of very good quality, that it will be easy to analyze and to make sense of,” he says.
Microbiome data are so likely to become medically useful, he adds, that individuals would be smart to collect them now. If people submitting their samples become sick later, the data could provide a personal baseline for getting back to health with as-yet undeveloped treatments.
“Getting to know your gut microbiota is probably a really important thing for people to do,” Mr. Sonnenburg says. In a few years, he thinks, microbiome sequencing will become a routine part of the yearly physical.