Which DNA test should you use to find your distant relatives?
There are pros and cons to each DNA test, but they can all be used to help you find your distant relatives.
The number of matches you get with distant relatives will vary from company to company. Each company has its own proprietary database, and you will find some cousins in one database who have not tested elsewhere.
Although the size of the database determines the number of distant relatives you get to some extent, each company also uses different algorithms and sets different thresholds for what it considers to be a match.
How do DNA tests find your distant relatives?
The amount of DNA shared between you and your distant relatives is measured in units known as centiMorgans (or cM). Current tests are run on microarrays or chips that cover about 700,000 markers scattered across the genome.
However, this represents a fraction of the total variation in our genomes, and there are some regions that have very poor coverage on the chips. This introduces the possibility of errors when inferring matches.
Some of the companies use a technique known as phasing to sort the DNA letters – all those As, Cs, Ts and Gs – onto the maternal and paternal sides.
If phasing is not done then the algorithms will zigzag between letters on the paternal and maternal chromosomes to produce false matches with distant relatives.
The companies that use phasing are able to set a lower threshold for finding distant relatives, because they feel more confident that the matches are real.
How accurate are DNA tests in finding your distant relatives?
The companies try to get the balance right, so that they are reporting a reasonable number of matches without too many false ones.
While matches with close relatives sharing large amounts of DNA can be called with confidence, many matches with more distant relatives are likely to be false.
I’ve been able to test both of my parents and I’ve found that 26–36 per cent of my matches at the different companies don’t match either of my parents.
It is mostly the distant relatives that are affected, particularly when matches share 15 cM or less.
Some firms assign confidence levels to matches. For instance, at AncestryDNA matches sharing 6–16 cM are rated as “moderate” with only a 15–50 per cent likelihood of sharing a single recent common ancestor, while MyHeritage has “moderate” or “low” confidence in matches under about 30 cM, stating, “This match may be a distant relative, but there is also a small chance that it is not related to you.”