If you could know every letter in your DNA, would you be better off?
Our recent obsession with kits like 23andMe鈥檚 certainly seems to indicate that many of us believe there is important knowledge (and fun) to be gained from probing our genome. But there is a significant gap between recognizing the alphabet of a foreign language and understanding one of its books.
Genetics sure used to be simple. You may remember having learned about the 鈥�1 gene, 1 disease鈥� model in school, which states that a single mutation in a single gene leads to a single disease. That was the low-hanging fruit of early genetics research, and it holds true for many diseases that tend to be less common. Take cystic fibrosis for example. One mutation in one gene results in the impairment of a channel that mediates what goes into and out of a cell. This leads to the production of a thick and sticky mucus, which brings about cystic fibrosis.
If all conditions were as straightforward, rushing to buy a direct-to-consumer DNA kit (like the one 23andMe sells) might make some sense from a health perspective. We would receive a binary list: you have this, you have that, you don鈥檛 have this, you don鈥檛 have that. But the more we learn about our genetic code and how it influences our health, the more complicated things get.
Take a common condition like asthma. It鈥檚 not caused by a single mutation in a single gene. It looks as if over 100 genes are involved depending on the population, and the environment interacts with these genetic predispositions.
Then there鈥檚 the word 鈥渕utation鈥� itself which is being displaced. In popular culture, a mutation is an aberration, something deeply pathological. When I was in grad school, whether or not something was called a mutation by geneticists depended on how rare it was. Back then, we had a myopic view of what was normal. Early efforts had led to the establishment of a reference sequence, based off of the sequencing of the entire genome of a few people. This reference sequence was, we thought, what a normal human genome looked like. So every bit of DNA tested in laboratories around the world was compared to this reference. If you found a deviation from the reference sequence, you looked around to know who else had seen it. If it had been seen in 1% of people or more, it was called a polymorphism (from the Greek for 鈥渕any forms鈥�). But if it was rarer than that, it was termed a mutation.
As the cost of sequencing entire genomes went down, scientists realized that there is more diversity in human DNA than previously thought. Some mutations can now actually be seen in more than 1% of people, and some of these mutations can cause disease in one population but not in another. That is why the words 鈥渕utations鈥� and 鈥減olymorphisms鈥� are being phased out in the genetics community. Instead, we say 鈥渧ariants鈥�.
Some variants are benign. Some cause disease and are termed 鈥減athogenic鈥�. Some affect function without being dangerous (e.g. variants in the MC1R gene lead to dark hair, blond hair, and red hair, none of which is a disease). And for some variants, we don鈥檛 yet know what they do. They are variants of unknown significance, the UFOs of the genetics world.
And even when a variant is associated with a disease, it may only confer a risk for it. How big of a risk? Sometimes, it鈥檚 a 100% chance of developing the condition. Sometimes, it鈥檚 1%. And sometimes, it depends on other variants in other genes.
This necessary caution in the interpretation of many genetic findings is why I鈥檓 skeptical of the mantra that 鈥渕ore knowledge is good鈥� adopted by fans of direct-to-consumer genetic tests. This thirst for information can have particularly egregious consequences when combined with fear. The Atlantic that many people are ordering a 23andMe test for a highly dubious reason: to figure out if their child should be vaccinated or not.
They order the kit, get results back, download the entire raw data, and use a third-party app to look at a gene called MTHFR, which codes for an important enzyme which contributes to the building blocks of proteins. Some naturopaths and anti-vaccination misinformers believe that mutations in this gene predispose you to bad vaccine reactions including (of course) autism. This nonsense is based on which looked at side effects like mild fever and rashes following immunization against smallpox (a vaccine which is no longer administered, as the disease itself was declared eradicated in 1980). Its senior author has recently said his study is 鈥渘ot even a valid study by today鈥檚 methodology.鈥� The company for MTHFR, citing that 鈥渢he existing scientific data doesn鈥檛 support the vast majority of claims that common MTHFR variants impact human health.鈥�
Variants have been spotted in MTHFR, but what many of them do is still not clear. In many cases, the activity of its encoded enzyme is unaffected. Some variants were initially tied to a number of diseases, but many of these associations could not be reproduced in larger samples. And some could even be beneficial.
Knowing the letters of a foreign alphabet is one thing. We鈥檝e gotten really good at cataloguing the As, Ts, Cs, and Gs of our genome. But how much of a problem a change in these letters creates? We do have some good knowledge, but it鈥檚 very easy to cut corners and portray nuance as certainty, especially in the service of an anti-science agenda.
Direct-to-consumer genetic testing has inadvertently left the door wide open for mutation-based pseudoscience. It wasn鈥檛 enough to worry about the woo around us: we now have to worry about the woo hiding in our genes.
For those who want to delve deeper into the problems with ancestry testing, click here
The hype over direct-to-consumer genetic testing extends beyond health and into ancestry, where the central question of 鈥渨hose reference sequence are we comparing your data to?鈥� rears its head again.
Earlier this year, CBC鈥檚 consumer-protection television show Marketplace shocked some people with : 鈥渙ne set of identical twins, two different ancestry profiles.鈥� One sister was told by her 23andMe report that 2.6% of her DNA came from France and Germany, while the other sister had none of that in her report from the same company. And the results between different companies were even less alike.
In order to pin down someone鈥檚 genetic ancestry, the service provider will compare parts of your genome to the corresponding parts in other people鈥檚 genomes, people who have self-reported as being Italian, or Russian, or Malay. These people may be previous customers or they may have participated in research projects in the past and their anonymized data is now part of public libraries. This makes up the 鈥渞eference panel鈥�, and each company has its own reference panel. Smaller panels lead to a greater risk of errors. Some panels have excellent European representation but a tiny number of samples from East Asian groups, for example. When your sequence is acquired by the company, it is compared to its reference panel, and statistics are used that can lead to variability in the final breakdown. And when the reference panel is updated, your results will likewise be recalculated and updated.
Hence, even with ancestry DNA testing, the actual precision of the final results has been blown out of proportion. A geneticist who commented on the CBC story referred to these direct-to-consumer tests as a 鈥渞ecreational scientific activity.鈥� Caveat emptor.
Take-home message:
-聽The word 鈥渕utation鈥� is being replaced by 鈥渧ariant鈥�, and a variant can be benign, can cause disease, affect function, or its significance may be unknown for now.
- Some people believe that variants in the gene MTHFR put you at risk from a bad reaction to vaccines, but this claim is a wild distortion of a bad scientific paper looking at fever and rashes in people who got immunized against smallpox.
Want to comment on this article? View it on our Facebook page!
