Column 5 min read

Genome Culture: How to Decide Which DNA Tests to Buy

Evaluating direct-to-consumer kits for tucking under your family tree.

By Laura Hercher featured image jxfzsy / Getty Images

‘Tis the season for aggressive marketing campaigns, and my inbox is littered with ads from HomeDNA, a company I’d never heard of until it started sending me two emails a day in November, with subject lines like “Save $100 on the Perfect Gift” and “Yule Like This Deal.” Deep analysis has determined that America is ready for genetic tests under the tree and, apparently, America agrees. According to reports, sold approximately 1.5 million DNA test kits over the Thanksgiving weekend —  a 200 percent increase, year-over-year. And on Black Friday, a “health test” from rival direct-to-consumer (DTC) testing company 23andMe was one of Amazon’s top five in total sales, right up there with the programmable cooker.

Are these tests for real? Sadly, there’s no simple answer to that question. A good starting point for evaluating consumer genetic tests is to look for CLIA certification. CLIA certification assesses laboratory competence. When they say you have a certain variant, it should make you more confident that they got that right. But CLIA is agnostic on what having that variant means. The DTC company may tell you that a particular variant means you are more likely to get cancer, or less likely to get diabetes, or prone to producing wet earwax. CLIA certification doesn’t say anything about whether or not they are right.

So you still have work to do to figure out what tests are credible. The HomeDNA lineup includes three different sorts of tests, so let’s use those as examples.

Ancestry Testing

Yes, this is a thing that exists. I am not endorsing this company or this product, but there are tests that can indicate, with varying degrees of exactness for different populations, in what part of the world your ancestors lived. Can they tell you what village you came from?  That’s a silly concept. Go back more than a few generations and your ancestors are legion. Heaven help you if they all lived in the same tiny village in Wales, or whatever. You would be as inbred as a lab rat and have the fitness of an English bulldog, which can no longer copulate without assistance. But I digress. Yes, ancestry tests can tell you something about where your ancestors lived and — if you use one of the major services and allow them to use your DNA to search for matches — it can find people to whom you are distantly (or less than distantly) related. But keep in mind, especially for gift-giving, that shaking the family apple tree can also turn up some pears. One person’s long lost second cousin could be someone else’s surprise half sibling. Caveat emptor.

Skin Care Analysis

If you believe a genetic test is better than a mirror at telling you what kind of skin you have, then boy do I have a bridge to sell you. But you won’t just be getting a sales pitch from me, because HomeDNA has friends with products to sell to you — it’s right in their abysmal privacy policy, if anyone bothers to check: “We may partner with other companies and individuals with respect to particular products or services.” Some examples of the characteristics ‘revealed’ by their genetic skin analysis include skin pigmentation, sun protection, and skin sensitivity.  Now seriously, I know we live in a post-factual world with alternate realities but please, if you want to know how your skin responds to sunlight, just go outside. If you think it is amusing to see what a testing algorithm believes about your tastes and your traits based on your genes, go right ahead, but remember that you are more than the sum of your data, so your lived experience is probably going to be a better guide than your test results.

Healthy Weight Analysis

This test claims it will tell you how you are likely to respond to a given diet or type of exercise. This health and wellness stuff is probably the hardest to assess. Be skeptical. For the most part, genetics is not that good yet at predicting outcome in dynamic systems, where many factors are in play. It’s like the algorithm that told HomeDNA I would probably buy one of their tests. Their data scientists may be right about people like me in general, but the actual individual me? I’m buying Instapots.

Digging down to see which of these tests are worthwhile requires looking at the science that backs up their claims, and that is a challenge. HomeDNA makes it easy by citing absolutely no science to back up their claims. Even their FAQs are a master class in the art of using words while saying nothing. If you click on “how dependable are the results,” you get this in response: “We have laboratory protocols that ensure very high accuracy, and so you can be sure your test has been processed correctly.” Oh, okay. How can I be sure the food at this restaurant will be delicious? Don’t worry ma’am, we have recipes.

Many tests will cite published research as evidence, but to establish credibility you actually need to know the journals (there are some that will publish anything, for a fee). Make sure the studies are run by scientists not connected with the company.   Both of these steps can be challenging.  Then look at the sample test reports. It’s not enough to say that there is an association between a gene variant and a health outcome; the report must explain how strong that association is. Some breast cancer susceptibility gene variants can raise a woman’s lifetime risk from 8 percent to as high as 85 percent.  That’s significant. Another variant might only raise her risk from 8 percent to 9.5 percent.  There might be great science behind the small increase, but it’s simply not useful. Finally, look at the type of recommendations they make. For diet-and-exercise type tests, most of it is probably going to be generic. If you need to be told to eat your vegetables and get some cardio, I can do that for you at a fraction of the price. You don’t even have to spit.

And while I am giving out advice, I have some for you, HomeDNA. In the future, you might want to tweak your marketing algorithm to avoid targeting genetics professionals. I’m just saying…