Over the past decade, we have heard about the “fat gene,” the “diabetes gene,” the “alcoholism gene,” the “intelligence gene,” even the “God gene.” In the end, none of these so-called discoveries proved correct.

Genetics is more complex than scientists imagined. We have discovered highly predictive genes for about 2,000 rare ailments, such as Huntington’s disease, but not for most common diseases such as cancer and diabetes. This hasn’t prevented companies such as 23andMe Inc. from selling direct-to-consumer genetic testing, with claims that it offers beneficial health information.

Debates about the information these companies are providing boiled over last week after the Food and Drug Administration, which is responsible for ensuring that medications, medical devices and tests are safe and effective, sent a letter to 23andMe, demanding that the company stop selling its $99 genetic test kit. On Monday, 23andMe stopped marketing its test kit.

On its website, 23andMe still boasts that its kit “reports on 240+ health conditions and traits.” A small proportion of this information ― as it relates to breast cancer mutations, for example ― may be medically useful. For the most part, however, studies have shown that direct-to-consumer test results are inconsistent. For example, Craig Venter, a pioneer in genetics, found that for five patients, the results obtained by 23andMe and a competitor were in disagreement at least half the time on seven of 13 illnesses analyzed.

A doctor who only diagnosed patients correctly 50 percent of the time wouldn’t practice medicine for very long. We don’t want doctors to give patients ineffective or harmful drugs or false and inaccurate information. The same standard should apply to other health-care providers and organizations.

The risk with companies such as 23andMe is that by telling people they have very high odds of getting a particular disease, they may unnecessarily suffer anguish or invasive procedures. Or a false negative result could lead a complacent patient to forgo screening or useful preventive action.

The FDA says it has tried to work with 23andMe, holding 14 face-to-face meetings and sending hundreds of e-mails, to press the company for evidence of claims about the tests’ validity.

Certainly, individuals have a right to their own genetic information. But whether direct-to-consumer genetic test kits should be allowed despite their flaws misses a larger point. The problem with these test kits is scientific. Only part of a person’s DNA is tested, and scientists are still unsure how to interpret most of the information.

Human DNA consists of a sequence of four molecules (so-called letters) that form a 3 billion-letter sequence ― 99.9 percent of which are shared by all humans. Individual DNA sequences, genomes, differ by about 3 million letters.

Several years ago, many scientists believed they would find highly predictive genetic mutations for many if not most diseases. But for the vast majority of patients, the causes of common diseases, such as heart disease, diabetes and most inherited forms of cancer, remain unknown; most of these afflictions can probably be ascribed to a combination of multiple genes and various environmental factors.

The test from 23andMe, which looks at only about 1 of every 3,000 letters, misses many genes that may be involved in a disease and thus faces even more limitations. The FDA is arguing that the company is offering less complete information than it advertises. At the same time, the price of whole genome testing is falling. Although it costs about $3,000 today, it may soon be available for only a few hundred dollars more that 23andMe’s partial and not fully reliable test. Full genome testing doesn’t solve the major scientific obstacles associated with genetic testing and disease prevention but at least it has the potential to consider more genes that could be involved in diagnosis.

In some ways, 23andMe’s test kit detracts attention from the company’s larger presumed objectives. Much of 23andMe’s value is in the hundreds of thousands of genetic samples it has collected from its customers. Such a collection can aid medical research, but, more important, the data will probably be very useful in securing patents. This year, the U.S. Supreme Court ruled that human genes can’t be patented, but the court appeared to permit other kinds of patents within the field of genetics ― such as synthetic DNA.

Executives at 23andMe may have been gambling that the company could continue to offer its inexpensive product and accumulate samples before the FDA acted more forcefully to stop sales of its test (some companies battle with government agencies for years). By then, the company would have created a valuable product: a huge bio-bank.

Given the stakes involved, the FDA should take stronger action to protect consumers from the risks associated with direct-to-consumer testing ― especially overpromises about partial testing. In coming years, scientists will learn more about the causes of diseases. At the same time, patients, doctors and policy makers need to understand the complexities involved. When it comes to understanding the disease, $99 doesn’t get you much.

By Robert Klitzman

Robert Klitzman is a professor of clinical psychiatry and director of the master’s of bioethics and online certificate programs at Columbia University. He is the author of “Am I My Genes?: Confronting Fate and Family Secrets in the Age of Genetic Testing.” ― Ed.

(Bloomberg)