I reviewed the results of Ancestry’s V1 chip in comparison with their V2 chip relative to matches recently in the article titled Ancestry V1 vs V2 Test Comparison.
I had previously tested on the V1 chip, and recently tested on the V2 chip to see how many of the same matches were present on both match lists. The results were better than expected. Out of my 333 V1 Shared Ancestor Hint matches, all but 7 were on the V2 match list. Given that Ancestry replaced almost half of the SNPs on their chip, that’s an amazingly high retained match number – about 97.5%.
Another genetic genealogist asked about how much of the DNA is the same, or in common for the individual matches. In other words, did the amount of shared DNA with individual matches change between the two chip versions?
While Ancestry does not provide us with a chromosome browser, they do provide us with the amount of DNA in common with a match after their Timber algorithm removes segments that Ancestry feels are “too matchy.” You can read more about how this is done, here.
In the screen shot above, you can see that the amount of shared DNA is displayed when you click on the “i” button beside the confidence level of the predicted relationship. In this case, I’ve looked at my V1 kit match to my V2 kit match. Clearly, I don’t have 26 chromosomes, so some of my chromosome segments have been severed, either by faulty reads or by Timber removing segments.
Because of Timber, the amount of shared DNA shown by Ancestry is not the actual amount of matching DNA when compared to matching DNA at any other vendor or Gedmatch. However, the amounts of shared DNA are consistently calculated between the V1 and V2 chips, so comparing Ancestry V1 to Ancestry V2 is certainly reasonable. What we don’t know is whether this is the same DNA that is matching between V1 and V2, or if the matching DNA is actually on different segments, partial segments or different combinations of segments. Without a chromosome browser or specific segment information, we have no way of knowing or discovering that information.
In the chart below, I’ve compared my 100 top shared ancestor hint (green leaf) matches (other than my own V1 to V2 kit comparison), meaning those with tree leaf hints indicating:
- That our DNA matches and
- That we share at least one common ancestor in our trees
Please note, for purposes of clarity, a shared ancestor hint (green leaf) does NOT mean or confirm that the DNA we share is from that common ancestor. The shared DNA could be from a secondary or different common line or the genealogy could be incorrect in one or both trees. The fact that we share DNA, and that we have an identified common ancestor in our trees are independent pieces of information that both serve as important hints. Both need to be verified. Without a chromosome browser and triangulation, we cannot confirm that the shared DNA is from that particular ancestor.
Amount of Shared DNA Between V1 and V2 Chips
For each of my 100 top V1/V2 shared ancestor hint matches, I recorded the amount of shared DNA as displayed by Ancestry and the number of shared segments. In addition, I also recorded the Ancestry predicted relationships and actual relationships as shown in my tree and my matches tree, as shown in the example below for Match 1.
My top 100 matches are shown in the table below, with their V1 and V2 results along with predicted and actual relationships.
- Bold=increases and decreases in the amount of shared DNA
- Red=increase or decrease of 2cM or greater
- Yellow=increase or decreases in the number of shared segments
Increases and Decreases
Of the various matches, 9 increased between V1 and V2, indicating that these individuals match on some of new newly included SNPs.
On the other hand, 52 decreased between V1 and V2 indicating that some of the SNPs where they previously matched have been removed on the new (current) chip.
Increases and decreases are bolded, including those in red which signify an increase or decrease of 2cM or greater. Nine matches had an increase or decrease of 2cM or more. Of those, 2 increased and 7 decreased.
The maximum increase was 5.3 cM.
The maximum decrease was 6 cM.
In most cases, the number of shared segments remained the same. Of the 4 that changed, 3 decreased and one increased, indicated by cells highlighted in yellow. In one case, the cMs dropped, but the segments increased, causing me to wonder if a segment was split in the V2 version. In another instance, the shared cMs remained the same, but the segments moved from 2 to 1. I’m not sure how to explain that one, except for the possibility that some of the removed SNPs caused the measured area to be counted as one instead of two, or perhaps the matching segments aren’t the same.
Actual vs Predicted Relationships
Eight people, or 8% had private trees meaning they can see the identity of our common ancestor, because my tree is public, but I cannot see the identity of that ancestor. That also means that I can’t determine the actual relationship for this comparison.
The 5 noted with ? means the ancestor is not the same ancestor or the match’s tree information is incorrect. In this case, that means 5% of the tree matches, or common ancestors as indicated in the trees are known to be inaccurate for one reason or another. There are likely additional inaccurate “common ancestors” given the amount of “tree grafting” that occurs.
In two cases the relationship was further out in time than predicted, although the predicted ranges are fairly broad and do significantly overlap. For example, one range is 4-6th cousins, and the next range is 5-8th cousins.
In 16 cases the relationship was closer than predicted.
I do have an endogamous Acadian line as noted.
In all cases, the amount of shared DNA was within the range of other people whose predictions were accurate, so this prediction variance is clearly a factor of the variability of inheritance of DNA.
The net-net of this exercise is that when comparing the shared DNA between the same match on the V1 and V2 chip, far more people lost matching DNA than gained – 52% vs 9%. In this comparison, all 100 of the people remained as matches, which isn’t surprising since these are my 100 closest shared ancestor hint matches, meaning those with the highest amounts of shared DNA. However, with matches that have “less to lose,” meaning more distant matches having fewer matching centiMorgans of DNA to begin with, matches are more likely to be lost.
In this comparison, the people who appeared as matches on the V1 chip remain as matches on the V2 chip, but just over half showed less matching DNA utilizing the V2 chip.