PDA

View Full Version : Tested kids - weird results



Grossvater
06-14-2017, 03:25 PM
With the recent offers of discounted DNA tests at 23andMe, I ordered two tests for each of my grown children. When I received the results, there were no earthshaking revelations. Since my wife is over a quarter Native American genetically, it was no surprise that my children were 13% and 14% Indian respectively. The rest of their ancestry is European with some minute African thrown in to spice up the mix.

What I found unsatisfactory was the feature that breaks down how much ancestry each of my children received from my wife and I, as we both have tested with 23andMe. My test (done several years ago) says I am 100% European, even though I know of a Native ancestor that died in the 1780s. When I run my raw data through the GEDMATCH calculators, minute amounts of Native American DNA shows up. 23andMe says my daughter got .5% of her Native ancestry from me. It also says that my son got over 2% of his Native ancestry from me. How can that be when by any stretch of the imagination I do not carry even one percent Native American.

Something's fishy with their testing methods, methinks? Has anyone else had similar results?

Wing Genealogist
06-14-2017, 04:21 PM
While I may be wrong, it is possible what 23andMe is saying is that of your son's 14% NA ancestry over 2% of THAT came from you (while the 97+% of it came from his mother). 5% of 14 is 0.7% so it appears your son inherited most (if not all) of your NA ancestry.

geebee
06-14-2017, 09:30 PM
With the recent offers of discounted DNA tests at 23andMe, I ordered two tests for each of my grown children. When I received the results, there were no earthshaking revelations. Since my wife is over a quarter Native American genetically, it was no surprise that my children were 13% and 14% Indian respectively. The rest of their ancestry is European with some minute African thrown in to spice up the mix.

What I found unsatisfactory was the feature that breaks down how much ancestry each of my children received from my wife and I, as we both have tested with 23andMe. My test (done several years ago) says I am 100% European, even though I know of a Native ancestor that died in the 1780s. When I run my raw data through the GEDMATCH calculators, minute amounts of Native American DNA shows up. 23andMe says my daughter got .5% of her Native ancestry from me. It also says that my son got over 2% of his Native ancestry from me. How can that be when by any stretch of the imagination I do not carry even one percent Native American.

Something's fishy with their testing methods, methinks? Has anyone else had similar results?

The short answer to your final question is, yes.

My father also shows as 100% European in Ancestry Composition. At the same time, my five siblings and I all show up with about 2% -- from just under that to just over it -- and my daughter has about 1.5%. This is a small percentage, but it's consistent across different companies and is also supported with documentary evidence.

For all six of the siblings, most of this ancestry is correctly attributed to our mother -- who was never tested. But for three of us, there's a Native American segment on chromosome 15 which is attributed to our father; and for one of us, there's an even smaller segment on chromosome 18 that is attributed to our father.

But here's the thing. In the case of the segment on chromosome 15, it was actually inherited by four of us. For the fourth sibling, the segment is correctly attributed to our mother. But it's exactly the same segment. Not only that, but all four of us happen to share a segment with a known maternal 3rd cousin in this same region, and in his case the segment is also identified as Native American. He's not on our mother's side, either; he is a fellow descendant of the same, known Native American ancestor.

But if this weren't enough, I passed this segment on to my daughter. It's correctly attributed to me. However, something I didn't pass on to my daughter is a recombined copy of chromosome 15. That is, on the copy of chromosome 15 that my daughter inherited from me, she has no DNA from her grandfather. She has plenty of other DNA from her grandfather -- including all of chromosome 9 -- it's just that her paternal copy of chromosome is purely from her grandmother.

So, it's pretty clear that my Native American segment on chromosome 15 -- which I passed on to my daughter -- can't be from my father.

The Native American segment on chromosome 18 was also inherited by four siblings -- though not exactly the same four. Amusingly enough, though, the only sibling for whom this is attributed to our father is also the only sibling who had the correct attribution of the segment on chromosome 15.

Her inherited segment on 18 is actually a little shorter, though I'm not sure if this is a factor. But here, too, I passed the segment on to my daughter. And guess what? She didn't get any DNA from my dad on her copy of chromosome 18, either. So this is definitely a maternal segment, and it has to be a maternal segment for all four of the siblings.

geebee
06-14-2017, 09:33 PM
While I may be wrong, it is possible what 23andMe is saying is that of your son's 14% NA ancestry over 2% of THAT came from you (while the 97+% of it came from his mother). 5% of 14 is 0.7% so it appears your son inherited most (if not all) of your NA ancestry.

That isn't how it works. 23andMe shows the percentage inherited from each parent as a percentage of your entire inheritance, not as a percentage from just that parent.

So, yes, they really are saying that 12% came from one parent and 2% came from the other parent. It's obvious that they're simply wrong, though I'm not sure what the cause is. I've thought it could stem from some sort of phasing error, yet it doesn't cause the "sliver" of misattributed ancestry to actually show up in the parent's Ancestry Composition.

wombatofthenorth
06-16-2017, 10:34 PM
That isn't how it works. 23andMe shows the percentage inherited from each parent as a percentage of your entire inheritance, not as a percentage from just that parent.

So, yes, they really are saying that 12% came from one parent and 2% came from the other parent. It's obvious that they're simply wrong, though I'm not sure what the cause is. I've thought it could stem from some sort of phasing error, yet it doesn't cause the "sliver" of misattributed ancestry to actually show up in the parent's Ancestry Composition.

Parents don't get phased as well from their children as children get from their parents at times and since you are mostly not Native American it probably had a hard time detecting it and washed it away and just called your bit European but for your son it picked it up.

Maybe.

hah

geebee
06-17-2017, 06:37 AM
Parents don't get phased as well from their children as children get from their parents at times and since you are mostly not Native American it probably had a hard time detecting it and washed it away and just called your bit European but for your son it picked it up.

Maybe.

hah

Again, no.

23andMe attributed the same segment to both my father and my mother. For three of us, it said that a Native American segment on chromosome 15 came from our father. This segment has the same start and end points as the segment of the fourth sibling, for whom the segment was identified as being from our mother.

It also happens that I match the sibling whose segment is maternal on both chromosomes in this region. So, even if she happened to inherit a maternal Native American segment that just coincidentally was located in exactly the same place as my maternal segment, we each should have both segments. We don't.

Also, I mentioned that I passed this very same segment on to my daughter. She shares absolutely none of her paternal chromosome 15 -- where the segment is located -- with my father. Essentially, I gave her an exact copy of chromosome 15 that I inherited from my mother. So no, it isn't that my "more accurate" phasing is picking up something my father has, but which is "washed out".

You're right that phasing against a parent -- which is what I have -- is more accurate than phasing against a child -- which is what my father has. This is especially true when that phasing is against both parents, which is what my daughter has but I do not.

A possible example of the sort of "washing out" that you're talking about is that my daughter has a small segment on chromosome 4 that is identified as "Middle Eastern and North African", and this segment is attributed to me. I don't have any such segments, nor does my daughter's mother. The segment is small enough that it could simply be noise, but I'm willing to allow that my daughter's phasing two-parent phasing might mean her result is more accurate than my one-parent phasing.

I say this not only because of the difference in phasing, but because my maternal grandmother was half Minorcan Spanish. That being the case, it's entirely plausible for me to have a small amount of North African ancestry along with my Spanish ancestry.

However, this is not the same thing that's happening with the Native American segment on chromosome 15 for me or for two of my siblings -- or on chromosome 18 for one of my siblings. These are purely errors on the part of 23andMe. It is possible that they're due to a bit of erroneous child phasing of my father's results. Keep in mind that even though 23andMe has results for all six of my father's children, they only use one for the child phasing. If they used all six of us, they could probably get even better results than with two-parent phasing.

EDIT: I've added a copy of pictures to illustrate what I'm talking about.

You should be able to see what I mean when I say my father and daughter don't match at all on chromosomes 15 or 18. That means that the Native American segment I passed on to her on each of these chromosomes can not possibly have come from him, but had to have come from my mother.

I think it's fairly obvious that the segment on chromosome 15 is the same segment for all four siblings. It therefore makes absolutely no sense to say that it's paternal for three of us but maternal for only one of us.

You can also see that in addition to not matching her grandfather, Kathryn does match Curt, Kim, and CJ in this region. That also means that I match these three here, and that each of the three matches the others.

It's a similar story for the segment on chromosome 18. In this case, 23andMe got it right for three out of four of us -- correctly identifying a maternal origin for the segment. But for my sister Kim, who inherited only a small portion of the segment, they've said it came from our father. Once again, doesn't match my father but does match Kim -- although only partially. That's why Kim is missing most of the segment. She didn't inherit the same maternal DNA on chromosome 15 that I did, or that Kathryn did.

1695816959

(Just so no one is confused, I took these snips from the same genome comparison, so Curt and Luci are included for both chromosomes, even though Luci doesn't have the segment on chromosome 15, and Curt doesn't have the segment on chromosome 18.)

wombatofthenorth
06-18-2017, 02:18 AM
I've heard that they have a bug that sometimes puts paternal on each top row instead of each bottom row.

geebee
06-18-2017, 02:41 AM
I've heard that they have a bug that sometimes puts paternal on each top row instead of each bottom row.

That's always possible, but I'm not basing my conclusions solely on which row the segment is in. Ancestry Composition says that 0.3% of my Native American ancestry is from my father, and 1.8% is from my mother. By this they don't mean that percentage of the total, they mean that this is how my 2.0% is split. And, yes, I know that 1.8% plus 0.3% would be 2.1%, but this is an effect of rounding.

At the same time, all of my Native American segments but one appear in my top row -- which does happen to be my maternal row. A single segment appears in the bottom row, on chromosome 15. Logically, this seems likely to be connected with that 0.3%.

My brother Curt and my sister CJ, who also are told that they have 0.3% NA from our father -- with the remainder being from our mother -- also have all of their NA segments but the one on chromosome 15 appear in the top row. Again, the segment on chromosome 15 appears in the bottom row.

Then there's my sister Kim. She's told that less than 0.1% (out of 1.8%) is from our father. For her, the segment on chromosome 15 is present, but it's in the top row. But she has a segment on chromosome 18 that is in the bottom row. Although three other siblings inherited a Native American segment at the same location of chromosome 18, Kim's segment is much shorter than the corresponding segment of the other three.

If you look at how the sibling genomes compare, it turns out that in a significant portion of her maternal chromosome 18, Kim inherited opposite maternal DNA from her siblings. This portion includes most of where the NA segment on chromosome 18 is located, which is why Kim's segment is so much smaller. Nevertheless, for the other three of us, this segment is in the top row, and only for Kim is it on the bottom.

So it's clear that this error can't be explained by the segments merely being displayed in the wrong row, unless the software can randomly display segments in the wrong row even when there's a tested parent, and somehow "forget" which parent the segment is associated with. I doubt that the software that determines the allocation is visually based, and goes by which row the segments are in. Presumably, it's the determination of which parent passed on the segment that determines the row it's displayed in -- so if it gets flipped, you'd think it would be consistently flipped. (And would not affect the percentage attributed to each parent.)