Your AncestryDNA® test results are the product of a multi-step process. First your DNA is measured, or read, in the lab. This lab processing generates raw DNA data. The raw data is then analyzed to generate your AncestryDNA results. There is no single measure of AncestryDNA test accuracy. Instead, the accuracy of each step can be measured independently.
Accuracy of the Reading of the DNA
Reading your DNA is a first step in generating your AncestryDNA results. Accuracy is very high when it comes to reading each of the hundreds of thousands of positions (or markers) in your DNA. With current technology, AncestryDNA has, on average, an accuracy rate of over 99 percent for each marker tested.
Accuracy of Regions in Your Ethnicity Estimate
When you take an AncestryDNA test, your test results will include an ethnicity estimate. Part of this is an estimate—reported as a percentage—of where your ancestors lived hundreds of years ago, as far back as around 1,000 years. An example would be 8% Italy, which reflects the amount of your DNA that has been inherited from Italian ancestors.
AncestryDNA determines this part of your ethnicity estimate in two steps. The first step is to collect the DNA of people whose family has a long history in a particular part of the world. This group is called the reference panel, and right now 43 different regions in the world are represented.
The second part of the process is to compare your DNA, bit by bit, to the DNA of the people from the 43 different regions in the reference panel, to see which groups’ DNA your DNA most resembles. For example, if ten percent of your DNA looks most similar to the DNA of people from France, AncestryDNA will assign ten percent of your ethnicity estimate to France.
AncestryDNA uses a number of different methods to determine the accuracy of this part of your ethnicity estimate. One method looks at how well AncestryDNA predicts the ethnicity estimate of people with a known ethnicity. For example, it looks at how well it works on people from the reference panel (who should theoretically come back with 100 percent of a certain ethnicity).
In terms of particular regions in an ethnicity estimate, the accuracy will—as is true for methods used by other testing companies —depend upon the region/population and the granularity of the prediction. If AncestryDNA restricts itself to the continental level (Europe, Asia, Africa, etc.), it is extremely accurate. As AncestryDNA drills down further into more specific regions, the accuracy will tend to go down. So it is a balancing act between accuracy and granularity.
See Table 4.1 of the Ethnicity Estimate 2018 White Paper . for more information on how well the AncestryDNA algorithm does for each region of the reference panel.
For your results, AncestryDNA calculates a range of most likely percentages for each ethnicity. If, for example, you see 41% France in your AncestryDNA results, the actual range of most likely results might go from 38-45%. France. In other words, between 38%-45% of your DNA is likely from French ancestor(s) with the most likely percentage being 41%. You can find this range by clicking on a specific ethnicity within your AncestryDNA report.
Accuracy of Communities in Your Ethnicity Estimate
In addition to the regions with percentages, AncestryDNA can also find communities your ancestors belonged to using our patented Genetic Communities™ technology. This method can tell you where your ancestors may have lived in more recent times, and how some people in these communities might have moved around the globe over the same time period. The patented technology uses AncestryDNA’s DNA database, the largest consumer DNA database in the world, and its vast treasure trove of millions of family trees.
Customers can identify how confident AncestryDNA is in assigning them to a community by following these steps:
1. Go to the the community.
2. Click on the little “i” for information.
3. Note there will be one of three categories: Very Likely, Likely, or Possible.
A community labeled as “possible” means there is a chance a customer may not be part of this community as they were on the edge of being included in the first place.
Accuracy of DNA Matching
DNA matching is an opt-in feature of AncestryDNA test results, in which we identify people you may be related to and how you may be related to them (for example they may be a 2nd to 3rd cousin).
The accuracy of AncestryDNA is extremely high for seeing if two people are related at the 3rd or 4th cousin and closer level. This high level of accuracy is based on the method by which relatives are found—sharing long segments of DNA. Usually this is only possible if two people have had a recent common ancestor.
AncestryDNA increases the accuracy even further by trying to rule out the cases where unrelated people happen to share a stretch of identical DNA. One way to do this is by limiting the size of the shared DNA. AncestryDNA also uses a special algorithm called TIMBER to filter out any larger pieces of identical DNA that may be shared for a reason other than a recent common ancestor. With this methodology, finding people you are genetically related to is extremely accurate.
Where things get a little harder to predict is when the level of relationship is assigned. AncestryDNA can assign a parent/child or a sibling relationship with a very high degree of accuracy. Other relationships are more difficult to assign for biological reasons.
Some relationships are difficult to assign because many happen to share the same amount of DNA. For example, relatives with grandparent/grandchild, double first cousin, half-sibling or avuncular (aunt/uncle) relationships all share around 25 percent of their DNA with each other. This obviously makes them hard to tell apart. So AncestryDNA reports back a “Close Family” relationship instead of one of these specific relationships. It is up to the customer to use other methods to determine which of these relationships is correct.
Other relationships are difficult to assign because the amount of DNA they share can overlap. For example, because of how DNA is passed down, sometimes you and your second cousin might share as much DNA as someone else and their first or third cousin. And there is more overlap the more distant the relationship: it is not easy to tell a 3rd cousin from a 5th cousin, for example.
Continued Commitment to Accuracy
The kind of science done at AncestryDNA is cutting-edge. This means that as the science advances, the AncestryDNA results will better reflect the history of where your ancestors lived and who you are related to. In addition, as more customers from more parts of the world are included, both the ethnicity estimate and matching results will also get better and better.