Shared matches with my son

[Susan48]

My son recently took an ancestry DNA test and is happy for me to manage the results. My husband has also tested and our son's results have been helpfully identified as Father's side or Mother's side. In looking at his shared matches with me I've been surprised to find that in some cases my son shares more DNA with the shared match than I do. It's not a lot more, but in a few cases it's pushed the match up into the 4th-6th cousin range. I would have expected him to share the same amount of DNA with matches as I do, or less - not more, but perhaps I'm missing something.

On a more bizarre note, my son's ethnicity estimate includes a surprising 1% North African element, which neither my husband nor I have.

 

[peter]

In looking at his shared matches with me I've been surprised to find that in some cases my son shares more DNA with the shared match than I do. It's not a lot more, but in a few cases it's pushed the match up into the 4th-6th cousin range. I would have expected him to share the same amount of DNA with matches as I do, or less - not more, but perhaps I'm missing something.

For a start you can and should ignore Ancestry's suggested relationships. Sometimes they're right and sometimes they're wrong, so ignoring them is best.

It isn't possible for your son to inherit DNA from his maternal grandparents unless he got it from you, but DNA doesn't have markers showing where one segment ends and the next starts; it also doesn't have markers saying which bits were inherited from one parent and which from another. Ancestry use a phasing algorithm to attempt to separate the sections that came from each parent, but whilst highly accurate, it's not 100% accurate. Finally, the test itself is not 100% accurate, though it is well over 99% accurate..

In short, discrepancies like these are something else you can and should ignore. Indeed there is little point looking at a child's matches if both parents have tested; if only one parent has tested then the child's matches are of interest in so far as they are not shared with the parent who has tested, but if the second parent is alive and prepared to test, they should test instead of the child.

 

[Susan48]

Indeed there is little point looking at a child's matches if both parents have tested; if only one parent has tested then the child's matches are of interest in so far as they are not shared with the parent who has tested, but if the second parent is alive and prepared to test, they should test instead of the child.

I know you have always encouraged us to get cousins to test as far as breaking down brick walls is concerned, and I accept that my son's results may not help me take the family trees further back, but I seem to remember that you also said that everyone should test, if possible, because otherwise our DNA will be lost to posterity (or words to that effect).

 

[peter]

I know you have always encouraged us to get cousins to test as far as breaking down brick walls is concerned, and I accept that my son's results may not help me take the family trees further back, but I seem to remember that you also said that everyone should test, if possible, because otherwise our DNA will be lost to posterity (or words to that effect).

A child's DNA is inherited from their parents, but they only inherit half of it - that's why it's important for the earliest surviving generations to test. Where both parents have tested, or are able to test, the DNA of their children cannot provide any genealogical clues that are not present in the parents' DNA.

(It might be important for a child to test for medical reasons - but in that case tests designed for genealogical purposes would probably be unsuitable.)

If only our ancestors had left perfectly preserved DNA samples, life would be so much simpler for genealogists!

 

[Helen7]

In looking at his shared matches with me I've been surprised to find that in some cases my son shares more DNA with the shared match than I do.

As reported in another thread it can happen that a son/daughter appears to share more DNA with a person than their parent does. This is because of downgrading of matches due to the filtering system (TIMBER algorithm) that Ancestry use, and the downgrading can be more for the parent than the child. As I reported in the other thread (post #3), I have 8 cases where my son has a match of >20 cM with someone whose match is <20 cM with me.

As the cutoff for appearing in Ancestry's 'shared match' list is 20 cM, if the discrepancy results in matches either side of 20 cM, it would result in someone showing up in a shared match list with your son but not with you (or your husband). This might prove useful when looking at shared matches - you may see some which you otherwise would have missed just looking at shared matches with yourself or your husband.

 

[peter]

As reported in another thread it can happen that a son/daughter appears to share more DNA with a person than their parent does. This is because of downgrading of matches due to the filtering system (TIMBER algorithm) that Ancestry use, and the downgrading can be more for the parent than the child. As I reported in the other thread (post #3), I have 8 cases where my son has a match of >20 cM with someone whose match is <20 cM with me.

No, TIMBER alone can't account for the difference between parent and child because the same algorithm will be applied to both. In any case Ancestry also provide the unweighted statistics, so you can eliminate the impact of TIMBER.

If one of Ancestry's algorithms is responsible for the discrepancy it is likely to be the phasing algorithm, because the other half of the DNA will come from a different source (ie the other parent in the case of the child).

 

[Helen7]

No, TIMBER alone can't account for the difference between parent and child because the same algorithm will be applied to both. In any case Ancestry also provide the unweighted statistics, so you can eliminate the impact of TIMBER.

If one of Ancestry's algorithms is responsible for the discrepancy it is likely to be the phasing algorithm, because the other half of the DNA will come from a different source (ie the other parent in the case of the child).

OK, I can see that the TIMBER algorithm alone might not account for the difference, and you can see the unweighted cM so you know how much the match has been downgraded (in one of mine it has gone from 29 cM to 9 cM and in another from 32 cM to 10 cM), but I can't really see how the phasing algorithm accounts for the discrepancy either when you can clearly see which parent the DNA comes from. The key thing though is not what causes it, but that it happens and has the effect on shared matches that I mentioned. One of my son's matches has 3 shared matches whereas I don't have any shared matches with that person (other than my son of course).

 

[peter]

I can't really see how the phasing algorithm accounts for the discrepancy either when you can clearly see which parent the DNA comes from

I'm not going to attempt to explain the technical details - it's well beyond the scope of what we, as genealogists, need to know.

But something to bear in mind is that whilst Ancestry know, in theory*, which parent your daughter got each of her chromosomes from, they don't know which parent you got yours from.

* my understanding is that they don't currently use this information, and process each test individually, but it's possible that they made changes when SideView was introduced

 

[JMCWSW]

"Ours not to reason why"

I've got my son, myself and my Mum all tested... there is no logical pattern, a lot of the time, as to why we don't DNA match to the same person, or when we do, the degree is topsy-turvey. In addition my Mum's first cousin has tested. He leads me to a further generation or so back on one of our lines, more often than not, but occasionally I've got the bigger degree of matching than he or Mum.

Right now I've got a bigger doozy I want to draw on DNA experts' opinion on... so there's a new thread on that. Similar problem, but different.

 

[peter]

"Ours not to reason why"

There isn't a simple straightforward answer, there are numerous possible explanations for these anomalies. But if yon really want to delve into the detail, you can do what I have done, and read all the technical articles about autosomal DNA, including Ancestry's White Papers and the research papers linked to from the ISOGG wiki.

The only one sentence explanation i can give is that the DNA tests used by genealogists are very cheap - it cost billions to sequence the first complete human genome.