DNA match anomaly

[ozdelver]

Hi,
I have noticed that I have several instances where a match on Ancestry DNA matches both my Mum and myself, but the amounts we match are reversed -that is, I share more DNA with the match than my mother does with the same match.
I have assumed that this means the match must be on both my paternal and maternal ancestral lines, and perhaps the lines linked up a generation or two further back on my mother's line than my father's.
Is this a correct assumption, or am I missing something here with regards to this anomaly?
Any advice will be most welcome!

[phil57]

To what extent do the amounts differ? What are the unweighted results for your and your Mum's match? (When viewing your match's profile page, click on the blue hyperlinked DNA result and it will open a box that shows the "headline" shared DNA and number of segments, the unweighted shared DNA and the longest segment).

Ancestry use a computational algorithm called Timber, which strips out matching lengths that they consider are more likely to be false, e.g. found commonly in population groups and unreliable to use for descendant matching, etc. It isn't perfect and can make errors. But the unweighted results are the shared amount before Timber is applied. If the unweighted amounts show a higher match length for your Mum, but it has become lower when weighting has been applied, that is probably the answer.

Are there notable differences in the shared matches between you and the match or your Mum and the match? Are any of those matches assigned differently by Sideview (Parent 1, Parent 2 or Maternal/Paternal if you have assigned them)?

The only time I have found something similar was a mother and her daughter who I matched on GEDmatch. I had a higher match to the daughter than her mother. When I asked, it turned out that the daughter had created a superkit by merging her own test with her mothers, which is not what their superkit function is intended for at all. But that won't apply to Ancestry matches.

[medpat]

I have had this with with a man and his father. They are 3rd cousin and 3rd cousin once removed.

The son has more DNA shared with me than his father. One section of son's DNA matches with mine but his father's doesn't on that chromosome. We discussed this and we can't find the link to his mother.

Since we tried to find the link we have had 4 people match with the son and myself on the DNA section where I meet the son and the 4 people do not match with the father. Unfortunately his mother has died so no DNA from her and I haven't been able to find where my family link is to the 4 people.

[Ruskie]

I haven’t tested with Ancestry (yet) but with My Heritage DNA my father, myself and my daughter have many examples of the child having higher cm matches than the parent.

With some of our matches I share higher cms with our common matches than my father does. My daughter shares higher cms with matches than I do, and so on.

In no instances is this related to matching on both the maternal and paternal lines.

I thought it was just down to the randomness of how we inherit DNA.

[phil57]

Even with randomness though, it is impossible for a child to inherit DNA from a parent that the parent doesn't have themselves.

Another possibility might be imputation. I'm by no means fully conversant with the process, but all testing companies use it to some degree, as no genealogical autosomal tests cover an individuals entire genome. They take a specific number of samples from certain locations and attempt to complete the gaps using knowledge of what is most likely likely to be adjacent to the known locations.

Ancestry changed from testing with their V1 chip to their V2 chip in 2016. Depending when the parent and child tests were taken, it could also be that they were read by different chips. The V2 chip only reads about half of the locations that were read by the previous chip, so there is more imputation involved in more recent test results.

Roberta Estes carried out a comparison between the two chips on her own DNA matches, and found that the V2 test only showed a 2.5% reduction in the number of matches though, so not as many as might be expected. Nevertheless, it could mean I suppose that some segment lengths could be missed, and overall match lengths reduced accordingly on the V2 chip, even though the overall match between two individuals is still mostly confirmed. 

[Ruskie]

Even with randomness though, it is impossible for a child to inherit DNA from a parent that the parent doesn't have themselves.

I understand that. In my examples parent and child both share cMs with the match but child has more than the parent. Looking for examples I found this one:

JS is a mutual match to my father, myself and my daughter.
My father shares 10.9 cM with JS
I share 55.2 cM with JS  (a lot more than my father)
My daughter shares 10.1cM with JS (almost as much as her grandfather)

Another example:
My father and MS share 32.6 cM
I share 59.9 cM with MS

This one is probably not unusual:
I share 60.2 cM with KC but my daughter shares nothing with KC.

I have numerous similar examples to this:
Me and VY share 48.5 cM
My father’s cousin’s daughter and VY share 15.2 cM (so I know which side of the family VY belongs to)
My father does not appear as having any shared CMs with VY.

I have found some other quite odd things with My Heritage matches, so I’m not sure how much I trust their estimates.

I will be testing with Ancestry very soon, so hopefully that might help sort some anomalies out. 

[brigidmac]

Hi Ruskie
I have a few matches like that too and as 2 of my maternal cousin's (from each of my mother's sisters ) and my nephew have now it's interesting to see when my generation have higher matches to someone than my mother .

I think under 60cm the amounts do not mean that you have necessarily inherited from the other parent
This is not scientifically based but from the cases where I've seen this .

Although it's always a possibility .

It can be that you  might be related to both of the matches parents but have inherited varying amounts compared to your father.

Occasionally there are some really unlikely co incidences . I found one of my distant Scottish matches from my father's side also matched my mother .
I'm not sure how the Canadian lumberjack family connect to my father but have found my mother's distant relative leaving Northampton going to Canada and marrying into the McMartin family whose descendants match my paternal aunt .....who did have a great uncle who never married but went gold  prospecting in Canada .
Looks like there must have been an illegitimate birth out there !

[Ruskie]

Yes with some matches I suppose there could be DNA inherited from both their maternal and paternal lines but my father should have inherited that too as he is a generation closer.

I’m treating it as just the luck of the DNA draw - you could drive yourself mad trying to analyse each match and each cM shared.

The matches in the examples given are all on my fathers line.

I don’t want to clog up the thread with any more unexplained examples, but just wanted Ozdelver to know I have similar instances to him/her in my results (as well as other weird things). 

[ozdelver]

Wow, thankyou all for your replies. I will have to work through them all and see if any of your suggestions apply to my strange case.

I have initially picked up this anomaly with these four cases, who are actually in pairs

With regard to phil57's comments, here are the details:

RS and CS are related. Looks like UK Tree.
My matches with RS and CS are 23(unweighted 24) and 20 (uw25)
My Mum's matches are 16 (uw17) and 16 (uw19).

The second pair are TS and JT. They are also likely related (?siblings). Tree is USA.
My matches are: TS 22 (uw22) and JT 20 (uw22)
My Mum's are: TS 9 (uw9) and JT no match.

So in all cases, even the unweighted amounts are greater in my matches than my Mum's.

In Mum's matches, the RS\TS pair are Maternal, and TS is Paternal
In my matches, all four are unassigned.

I have no evidence to suggest that these two sets of pairs are related to each other, which means that there are two separate examples where I have more DNA than my Mum from the same matches.

With regard to imputation, my memory is a bit dodgy but I am assuming that my test was after 2016, and Mum's was only a year or so ago. So that seems to discount the chip theory.

There may be more examples, but I have only just discovered these ones so can't say if it occurs frequently. I intend to look deeper into their trees, and see if my parents' ancestors do cross over somewhere in the past. 

I should add that I also have a couple of matches (not the above examples) which are matching to both my paternal and maternal lines. I am assuming that this might have occurred here in Australia in the 1800's when our population was relatively small. The various gold-rushes drew thousands of people in from the UK and the USA. But that's a puzzle for another day.....

Thanks again for all the responses, they are appreciated!