DNA Inherited from Grandparents and Great Grandparents

+8 votes

In this interesting post (January 14), Roberta Estes explains that few people actually get 25% of their DNA from each grandparent, and can actually inherit twice as much DNA from certain of their great grandparents as from others.  She also briefly discusses other reasons that we may have very different numbers of matches from the various branches of our family trees.

Update:  Please see link in JN Murphy post below.  Thanks, JN.
in Genealogy Help by Julie Kelts G2G6 Pilot (167k points)
edited by Julie Kelts

1 Answer

+5 votes
Really, the way to think of it is like if you flipped a coin 100 times.

People might think you should "get 50 heads", but in reality you should get something CLOSE to 50. The probability of getting EXACTLY 50 heads is pretty small (I'd guess less than 5%), but the probability of getting between 40 and 60 heads is pretty good (I'd guess over 90%). The probability of getting 50 heads is higher than for any other number, but nonetheless the chance of ANY specific exact number occurring is low..

So in that way, of course it will pretty much never be EXACTLY 25% from each grandparent. But it's interesting to see what the variability is, and it's a good thing to know.
by Frank Stanley G2G6 Mach 5 (55.3k points)
Now did anyone ever say DNA is an exact science and  the interpretation is what goes wrong????  :)

I've obviously a bit of tidying to do Frank.
With tests at FTDNA/23andme/Ancesty and uploaded to MyHeritage and GEDmatch there are still quite a few people who cannot be compared to others at the same spots for clues despite my having strategically added the closest rellie I could get from each side to 23andme, a few others at Ancestry for clues, and transferred heaps into MyHeritage and GEDmatch.

My first cut filtering by each category in turn totalled 113%, :)
compared with the 97% that seems to be total coverage possible and 93% I have overall to date!!!

A bit of quick tidying came up with a total of 97% supposedly accounted for by category, which included a definitely Paternal (or Maternal) but which grandparent unknown but is still too high and must have overlaps

Oddly enough the breakdown of the unbelievable 97% ended up with

Dad's father 17.5%
Dad's mother 12.6%
Paternal, unknown branch 6.2%
which is way too low at 36.3
There are a heap of spots I have NO paternal side matches over 12cMs, about 16%, which would bring that up to 52%  and needs more attention to see what's there under 12cMs
I don't have Jim Bartlett's dedication to mapping down to 7cMs, preferring to balance research into the brickwalls with the mapping / assigning.

Mum's father 20.5%
Mum's mother 23
Maternal, unknown branch 17.4
Total Maternal 60.9%, so again must be overlaps in there across companies to be reassigned.

All great fun nevertheless and has just led to yet another brickwall cracked just in the last couple of days.

I confess I use the mapping mostly as an aid to exclude people from further inquiries into whatever research I'm currently doing, eg  who all has to belong to the particular ancestral line of focus and can be researched further for more clues.

Frank, you can map your chromosomes back to your grandparents based on your DNA and that of two siblings.  I believe you could do it with one sibling, especially if you also had some cousins to compare to, although it would take longer.

I used the Bettinger method.  Below is an example.  Basically, by comparing your match to two siblings simultaneously, you identify the crossover points each of you has.  Once you have some other matches, you can identify which segments come from which grandparents.

A couple years ago I wrote up detailed instructions on how I'd made my maps, but could never interest anyone in trying it.  I will say it is a lot of work!  In my opinion, well worth it.

I have not mastered the art of getting my G2G exhibits to come out the size I want them.  Here is a link to an expanded version:


The top part of the chart is taken from the comparisons of me and my siblings on GEDmatch.  The bottom part is the reconstructed chromosomes by segment.
Well, this stuff is obviously a bit tricky. Lorna's has about a 60/40 mix on maternal vs paternal, so there's some trouble-shooting to do there. Julie's very nice graphic looks a little suspicious to me at one spot.

So let me go through Julie's. It seems like the way to go is generally right-to-left.

I see about 2/3 of it, on the right, as being all three siblings matching, except for two obvious segments where Julie is the "odd person out". I guess it kind of makes sense that both of those segments are on the same side (meaning her mom's side) since if the crossovers were on different sides, she wouldn't be returning to matching both siblings. Which grandparents these segments come from would all have to determined by matches with cousins.

So far, so good! Then we more to the left, and see that Lynn is now the "odd person out". Presumably, we know he difference is on her dad's side, due (again) to matching vs cousins. All is still good, as far as I can figure!

But the next piece (2nd from the left) is where I wonder if it doesn't add up. Does it really make sense for Lynn to have crossovers on both her dad's and her mom's sides, at the exact same place?

Instead, I would look at "Julie vs Lynn", and expect that Lynn remains "blue" and "orange", all the way back to the beginning. The "vs George" results are still explained by his having a different segment (right where you have it), but I would think it would need to be on his MOM's side, not his dad's.

So, in summary, I would think that Lynn has two segments on her dad's side (blue then pink), and is solid orange on her mom's side, while George is solid pink on his dad's side, with a short green segment of his mom's side interrupting what is otherwise orange.

Really, it also seems like it must be kind of unusual for all three of you to mostly have the same (pink, orange) combination, with some minor exceptions. The chances of having 3 siblings all match at a given point is 1/16, as best I can figure (64 possible combinations, 4 of which have all three matching).

Frank, thanks for your interest.  I see that you grasped the basics fairly well, but as you note, it's tricky.  Developing these maps is an iterative process, and the segments I've identified are consistent with all of our other matches.

I picked chromosome 18 because it was one of the simplest. (In contrast, look at chromosome 2:

https://www.wikitree.com/photo.php/b/b0/Chromosome_Mapping_Examples-3.png )

Lynn's double crossover is not really at exactly the same place; it just looks like it on the small chart.  Double crossovers are among the most perplexing problems encountered during this type of mapping, but they do happen.

You're suggesting that Lynn's double crossover is actually no crossover, but look at this match to our mother's cousin:

Yes, it probably is unusual for all three of us to be fully identical on much of the chromosome, but you can see from the green bars that we are.

I'll be happy to send you my detailed map if you'd like to audit it!

One day I'll set aside time to do visual phasing on my three sibling 2nd cousins I've had tested to see if that helps in any way with our mysterious 2*great grandparents parents.
Up until a week or so ago I didn't have a recent enough version of Excell to use the more automated tools.
My initial painstaking by hand mapping somehow gets lost when other things keep getting in the way. :)
I have to get more up to speed on that in the next 2 months as  I'm booked to attend a 3 day DNA workshop with Jonny Perl, Blaine Bettinger and Angie Bush. Which I'm very much looking forward to.
Julie, that's a bit chopped off, but if you say you're aware that something unusual is going on, and have proof, that's good enough for me!

Really, it's a good illustration of how that can happen occasionally (apparently, which is good to know), and how just looking at the 3 siblings - alone - could lead you astray occasionally.

As to the thing about all three siblings matching for the most part, that actually makes a lot of sense, the more I think about it. With the 1/16 odds, you're bound to have something like that SOMEWHERE in your chromosomes, and a chromosome where it dominates is one that's going to be less cluttered (in terms of how these diagrams look), and therefore one you'd be more likely to pick to show us! In short, you probably practically picked that one on purpose, just because it was like that, whether you realize it or not.
The smaller ones tend to be less cluttered, as you might expect. (And for anyone who doesn't know, the chromosomes are numbered by size, with no. 1 being the biggest.  Here is one illustration:

https://www.ncbi.nlm.nih.gov/books/NBK22266/ )

Frank, what that little snippet shows is that my sister had to have a crossover at 7.3, because that is where she starts matching our mother's cousin, even though my brother and I match him beginning at .1.

(In the absence of siblings to use for visual phasing, one's crossover points can be found through such differential matching.)

Related questions

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright