How DNA Testing Services Work

You’ve no doubt seen the advertisements for DNA testing services that offer to test your DNA to find out your entire genetic history. Ancestry.com and 23andMe are the most famous of the DNA testing companies. And people love these services: over 26 million DNA tests have been performed across all the major companies. The tests are fairly affordable for most people too: the retail cost of these tests are about $50 to $140 per testing kit. And less if you use an Ancestry DNA coupon on your order. But how does the science behind these services work? And are the DNA results accurate?

According to Ancestry.com, their tests use something called “microarray-based autosomal┬áDNA testing“, which analyzes your genome at over 700,000 locations. Now, your human genome has over 3 billion base pairs. So the way these testing services work is that they analyze the areas of the genome where the most variations and differences are located. The result is that it only focuses on small genetic variations (called single nucleotide polymorphisms). These little differences will determine a person’s eye color, height, and more.

So when you do a genetic test, these services analyze your genome and compare it against their database of people with known genetic heritages (basically, past customers). Which is where there can be problems: as their database gets larger and has more information in it, you may do a test one year with particular results, but end up with different results if you do the same test later.

Initially, these services first built their database based on self-reporting of individuals in particular regions (i.e. Greece or Turkey). And took them on faith. But as time has gone on, these companies’ reliance on initial testing has lessened.

But there are biases based on the demographics of people who take these tests. The genetic history of Europe is well known and very well mapped out. If you’re of European ancestry and you take one of these tests, you can be relatively certain that your results will be accurate.

Also, of the 700,000 locations that are analyzed, the test can produce small errors that can add up to a different result if you take the same test twice. While the analysis is 99.9% accurate, when you’re examining 700,000 locations, that 99.9% accuracy will result in around 100 errors. These little errors can be enough to make identical twins get disparate results.

The other thing to consider is that siblings may get different results (even though they have the same ancestors) because different people will inherit a different amount of their genetic ancestry. So if the siblings are Irish and the family tree extends into Spain a couple hundred years ago, one sibling may inherit more of that Spanish genetic heritage than the other brother.

After you’re DNA has been analyzed, these companies use an algorithm to assign you a percentage of different ancestries. When you take one of these tests, you’ll receive a report that says: 23% chinese, 14% arabic, 12% German, etc. Their algorithm takes all the unusual single nucleotide polymorphisms (SNP’s) we mentioned above and comes up with a percentage from each genetic region. That’s where some mistakes can come in. For example, England’s genetic makeup is similar to Ireland, so you may end up with a test that says you’re 80% Irish when you are actually English. It won’t go way off-board and say that you’re 80% Japanese, but this algorithm’s estimations of percentages can sometimes be quite inaccurate.

But as more people use these services and their databases & algorithms improve, these DNA tests will start to become much more accurate and informative.