Can someone help me read DNA results from GEDMatch for inbreeding? I really don't understand Genetics at all.

+5 votes
316 views
Hi all! I have my hubby's DNA results and I have read through when I first got them with some online material trying to understand what they were trying to tell us about his heritage. We have had difficulty finding his grandfather. Can someone help me read his DNA results which may give us a clue?

Thanks so much!
WikiTree profile: Carter-32207
in Policy and Style by Connie Carter G2G4 (4.5k points)

3 Answers

+10 votes
 
Best answer

Hi, Connie. If I'm reading the hypothesis correctly, the "are your parents related" tool at GEDmatch won't be of any help. That looks only at runs of homozygosity--rather lengthy stretches of the chromosomes where all the alleles match--and would only indicate if both the father and the mother themselves had a recent biological relationship.

If the supposition is that the father's father is actually the biological parent--and the mother is unrelated--that's an entirely different story. And I'm making the assumptive leap here that the birth certificate father and presumed biological mother have both passed away and can't be tested.

A grandfather being the biological father should be fairly easy to detect if there are other closely-related test-takers and their paper-trail relationships are identified. But the paper trail would be necessary except for one instance. What I don't see mentioned is whether or not your husband has any siblings that could be tested. If so, that could give you an answer. They would each have received 50% of their DNA from the same mother, but the DNA distribution between your husband and his birth certificate siblings would be indicative. The amount of shared DNA would be lower than anticipated for full siblings, but higher than for half siblings. My rough expectation is that it would fall somewhere between 1700 and 2100cM.

Similarly, but more difficult to corroborate because the distribution of DNA becomes somewhat more random the more generations that are introduced, it could give you a strong clue if the grandfather had siblings who had children who could test. If the father is in fact the biological parent, the grandfather's siblings' children would be 1st cousins 1x removed (whose theoretical expected sharing would be about 425cM), and their children--who should then be of the same generation as your husband--should be 2nd cousins (or about 212cM). If the grandfather holds the biological keys, those relationships would move up the ladder one rung. Someone who, on paper, would be a 2nd cousin would test at around double the expected amount of shared DNA; same for a 1C1R, who would then test at around 850cM, the same as a full 1st cousin. But, again, there would need to be a paper trail to compare. The DNA results in and of themselves won't tell you.

A small tangent, but one I want to address. I see a fairly frequent misconception that pedigree collapse--biological relatives having children together or collapse due to instances like double 1st cousins--always has a lasting and telling impact on DNA sharing.

In fact, the detectable effects of the more common forms of pedigree collapse (discounting immediate family incest) dissipate fairly quickly with each successive generation if there is no further collapse. Otherwise, we'd never have had enough genetic diversity to survive as a species.

There are simple equations to help look at the theoretical DNA sharing in these instances, called the Coefficient of Relationship and the Coefficient of Inbreeding. But unless the lines are really convoluted, you can estimate the effect by simple addition, as Jessica showed, based on the hypothetical relationships involved. While not comprehensive, you can look at a chart I put together to show some expected sharing amounts. Note that half-relationships are just that: divide the full relationship's expected sharing in half, e.g., a 2nd cousin is expected to share about 3.125%, or 212cM, so a half 2nd cousin would be half that, about 106cM, the same as a 2C1R.

A quick pedigree collapse example, with a diagram I made last year (you can click it for a full-sized version).

Two 1st cousins, A1 and A2, have a child, B. B marries unrelated C and they have two children, E and F. Rather than four unrelated grandparents, E and F have two that are 1st cousins.

E marries unrelated D, and F marries unrelated G. Each couple has a child, H and I respectively. H and I are 1st cousins through their grandparents B and C, but they also share a genetic line to A1 and A2; they share two great-grandparents making them also 2nd cousins. 1st cousins can be expected to share about 12.5% of their DNA, or about 850cM, and 2nd cousins, as noted above, about 3.125%, or 212cM. Add those two genetic relationships together and you get 15.63%, or the equivalent of around 1,060cM.

At this point, the pedigree collapse should be fairly evident, or at least highly suspected. However, just one more generation removes that noticeable disparity into the realm of simply reshuffling variance.

The children of H and I--K and L--are 2nd cousins through B and C, but now they're one generation removed from the pedigree collapse, so A1 and A2 are their 2g-grandparents. To their 3.125%/212cM expected sharing as 2nd cousins, we'd now add their additional 3rd cousin relationship, which ups the total by only 0.78%/53cM to 3.9%/265cM.

If we were to take it yet one more generation with no additional pedigree collapse to the children of K and L, they would now be 3rd cousins via 2g-grandparents B and C, but also 4th cousins through A1 and A2. So we would add the 3C 0.78%/53cM to the 4C 0.195%/13cM for a total of 0.975% or about 66cM. At this point, there's really no discernible difference at all between N and O: there would be no indication in the DNA that there had been pedigree collapse generations earlier.

So the closer--both in immediacy of family members involved as well as chronology--are instances of pedigree collapse, the greater the impact on DNA sharing. But if there is no repetition of collapse along that same lineage, the difference in detectable sharing decreases rapidly.

In fact, and not coincidentally, the impact in total sharing between cousins decreases four-fold with each subsequent generation...exactly the decrease in theoretical expected sharing from one full level of cousinship to the next. For example, divide the expected sharing between 1st cousins, 12.5%, by 4 and you get 3.125%, the expected sharing between 2nd cousins. In the pedigree collapse illustration, the difference between the regular and the collapse-weighted sharing is 25%.

Edited to remove a "Hi, Connie" line. One was enough. wink

by Edison Williams G2G6 Pilot (259k points)
edited by Edison Williams
Thank you so much. I'm going to combine a few of these answers and try to figure this out. At least rule it out and keep on going. Thank you so much everyone for your help!

Thanks for the "best answer" star, Connie. And apologies for saying "hi" twice. Copy-and-paste caught me...but I'm going to hop over and correct that.  wink

Best of luck with the research. We've got some bright folks here. If you find matches that seem unusual or something else turns up that looks confusing, let us know!

Fascinating! And A very concise breakdown. I truly appreciate this explanation. Thank you so much.
+8 votes
The simplest thing to do is to run his kit through the 'Are Your Parents Related?' tool on GEDmatch.
by Jessica Key G2G6 Pilot (180k points)
I had done that. its his paternal grandfather that is the brickwall. the secret went to the grave with his grandmother. Thank you somuch for the suggestion though.
Well, did it indicate there was inbreeding? That is what you seem to be asking about.

With patience and persistence, it should be possible to determine the identity of the grandfather. You just have to sort his DNA matches via the Leeds method and keep working on it.
In case you didn't understand what Jessica said, the question of whether your husband's parents are related has nothing to do with whether you have identified them on your tree.  It is purely a matter of DNA.
I'm not sure...I can't rule it out as I have no leads to a grandfather anywhere but there could definitely be a grandfather out there somewhere. I'm trying to rule out one possibility at a time. So, I guess, yes that's one question...how could I tell from the DNA results if there was inbreeding by someone in the family with his grandmother? If that's ruled out, then how can the DNA be used to help locate a paternal grandfather somewhere? I've read many articles  on the DNA and how certain markers are indicative of certain traits and health conditions. when it comes to relationships, I'm just lost...So I appreciate any help, any leads on how I can use this information to best help find hubby's Grandad. Thanks again!
I did understand what Jessica was saying. Thank you so much though. I truly do appreciate any help that I can get.

Okay Connie, let's work through the incest scenario and explain how DNA can reveal it.

Let's say there's a young woman named Martha. Martha is impregnated by her uncle, Tom. She gives birth to a son, Jake.

Ordinarily, a mother and son will share about 50% (half) of their DNA. But Martha is not just Jake's mother, she's also his cousin via Tom. So they share approximately 62% of their DNA.

Usually DNA tests give away incest when two people know their "paper" connection but the DNA test is revealing they share too much DNA. This is not that common a revelation but it has happened a couple of times on the genetic genealogy forum I frequent. Please do realize that if it turns out your husband was conceived via incest, this is something that could devastate him. I never advocate keeping the truth from a person, but I do think people need to be prepared for worst case scenarios and counseling may be necessary. 

Thank you so much Jessica. My husband suspects it and has for years. His father used to tell him that HIS father (hubby's grandfather) was his father (Hubby's dad). when younger my hubby thought that meant its because he was raised by his paternal grandparents, not that he actually meant that biologically. After the DNA results came back, we looked up some results and thought it might be possible. In his mind hes convinced, me I'm not so sure and before I make it gospel truth in the tree, I have to know. So that's why I wanted to consult people more experienced in DNA results who knew what they were talking about and then I could move on from there to try to find his grandad. You have been very helpful. Thanks so much!
+3 votes
I've ancestors who were inbred.  My great great great grandfather married his 1st cousin.  It's not brother/sister inbreeding,but it's close.  There's a paper trail to follow.

My great grandmother,Sarah  Elizabeth Pitts,and her husband, my great grandfather,Eli Thomas Profitt, were more distantly related to each other.
by David Hughey G2G Astronaut (1.4m points)

surprise   

I really am dumb when it comes to DNA, Snps, ROH etc....but Im getting a better grasp I think....that's not saying much LOL! This is what my husband knows his ancestry:

His father, George, does not know who his father is, or so the family states. George lived with his mother Mary and her parents, George's grandparents, Ralph and Sarah Carter, (my hubby's g grandparents) until he was an adult. My husband suspects that his g grandfather Ralph impregnated his daughter Mary and she had baby George, my Hubby's dad. So that would make Hubby's g grandfather his grandfather as well, if this is true.

I ran the DNA thru Gedmatch "Are Your Parents Related" link again and there is no apparent inbreeding, according to that test.

Now it's possible, that it was another, more distant relative that impregnated her OR she had relations with a sailor passing thru! It really could be anything. My hubby really thinks its his great grandfather because his father always said that.

I will try David Pike's link and see if anything comes up with that. I just hope I can understand it. I'm trying the Leed's method, again, not sure if I can figure it out but I'm reading all I can on it.

Thank you sooo much for all this information.  It is greatly appreciated!

Content removed

i have some results here if anyone can help me read them:

https://www.math.mun.ca/~dapike/FF23utils/roh.php

I'm pretty sure now there is inbreeding somewhere.

Thanks so much!

I just wanted to thank you again! I think I have my answer. Do youknow of any other sites that I can run the results thru that can tell me more about the ROHs? Daves site was excellent at identifying the runs in the data.

Hiya again, Connie. I agree with your choice to remove the information in that earlier post. Since I'd replied to the topic, I received it via email. I'll make some general comments here that don't reveal any of the data.

First things first. Step one is to make certain we're looking at the returned data we want to look at. It isn't immediately obvious.

What you posted looked like it was taken from the very bottom of the reported results. Logically, that's where you'd think you should look, right?

But if you scroll up a bit from there, you'll see the heading, "The following percentages are for individual homozygous SNPs, not just those in ROHs." So that bottom section is looking at all locations on the chromosomes that show the same nucleic acid: we only have four of 'em--A, C, G, and T--so a bunch of the locations, locations that don't mean anything about pedigree collapse because they aren't in sequential runs, will be doubled-up, i.e., AA, CC, GG, TT.

These are reporting as homozygous SNPs, but we're only interested in them if they're strung together, a lot of them occurring contiguously, for example: AA, GG, GG, AA, CC, TT, etc.

Just above that heading about individual SNPs you should see three lines that look like:

Total ROH Mb: [number]
Total genome Mb: [number]
The detected ROHs account for [number]% of the reported genome

What we're interested in is that percentage, as well as the per-chromosome ROH sizes listed above it.

(Ed's Nerd Note [meaning this can be completely ignored]: That percentage is a measure of individual autozygosity--usually written as Froh--which is the proportion of the autosomal genome in runs of homozygosity above a specified length threshold. ROH are contiguous pieces of a chromosome that exist because parents passed down identical segments. Autozygosity is the degree to which both copies of a chromosome an individual carries have identical genetic sequences.)

Okay. Still with me? As I mentioned before, several studies have indicated that we all show some level of autozygosity--makes sense because all our DNA came from somewhere, and the whole "can't split-up genes that actively code for important stuff" thing. They found that the length of a run of homozygosity was important in determining whether or not that segment indicated recent pedigree collapse, or was simply inherited from longer ago...a population-level indicator. And the population-level genetic origins of the test-taker make a difference in evaluating the results.

The (somewhat) bad news: this means there really is no one hard-and-fast rule in using ROH to estimate pedigree collapse if the consanguinity is more than a few degrees of relationship removed; it's more assured if it's recent. If Aaron reads this, he might want to give us an idea of what evaluation thresholds GEDmatch uses.

African, East Asian, and European populations tend to show the shortest length of ROH, and smallest total genomic ROH (Kirin, et. al., "Genomic Runs of Homozygosity Record Population History and Consanguinity." PLoS One, Nov 2010; https://doi.org/10.1371/journal.pone.0013996).


Click the chart for a full-sized version.

(Ed's Nerd Note [meaning this can be completely ignored]: There is always discussion on G2G about Native American DNA and our common, inexpensive microarray tests. There has been--and will probably continue to be--little interest among the NA People regarding DNA; it isn't something that defines their ethnicity, and a single interaction with colonial European uniparental DNA--yDNA or mtDNA--will remove NA indicators for those anthropological-timeframe markers. However, the study linked above, though dated by current standards, found that NA populations as sampled from Central and South America showed the longest stretches of homozygosity for all ROH length categories, as shown in the chart. Of the NA samples evaluated, 67% showed at least one ROH longer than 10 megabase pairs. I don't really research NA genetics, and don't know if there have been other studies that looked specifically at this population-level ROH/autozygosity and whether it might be a reasonable indicator of NA descent within, say, five to 10 generations. For those involved, that could be worth some research.)

<coughcough> So... David Pike's utility will report short runs of homozygosity that, depending upon the circumstances and population-level genetics, may very well not be informative about pedigree collapse. You might choose to disregard very small ROH segments; your call. And because the utility includes the X-chromosome, we have to do some of our own math.
frown

We males are odd, and that throws some folks for a curve when looking at data like these. Males have only one Y chromosome, only one X chromosome, and then two tiny little areas, called the pseudoautosomal regions (PAR), that allow the Y and X chromosomes to link up. But because almost all of the formats the testing companies use for our microarray raw DNA results contain two columns of alleles (the A, C, G, and T) for each locus on a chromosome, they simply double-up the values of the X chromosome for us guys so that both columns are filled (i.e., every allele will simply be displayed twice, for instance: AA, GG, AA, CC). That's why Pike's utility is showing your husband's X chromosome as 99.988% homozygous: the only bits that can't be homozygous are in the PAR...which can have medical importance, but does absolutely nuthin' for genealogy. So we ignore everything but chromosomes 1 through 22.

I'll (gasp) reveal my actual results so you can follow along:

  • Chr  2 has a ROH of length 464 ( 0.20 Mb)
  • Chr  2 has a ROH of length 217 ( 0.03 Mb)
  • Chr  2 has a ROH of length 252 ( 0.09 Mb)
  • Chr  3 has a ROH of length 313 ( 0.09 Mb)
  • Chr  6 has a ROH of length 961 ( 0.48 Mb)
  • Chr  7 has a ROH of length 628 ( 0.12 Mb)
  • Chr  9 has a ROH of length 524 ( 2.66 Mb)
  • Chr 10 has a ROH of length 209 ( 0.91 Mb)
  • Chr 11 has a ROH of length 334 ( 0.12 Mb)
  • Chr 11 has a ROH of length 221 ( 0.89 Mb)
  • Chr 13 has a ROH of length 251 ( 0.02 Mb)
  • Chr 14 has a ROH of length 338 ( 0.13 Mb)
  • Chr 15 has a ROH of length 232 ( 0.15 Mb)
  • Chr 16 has a ROH of length 200 ( 1.39 Mb)
  • Chr 17 has a ROH of length 243 ( 0.10 Mb)
  • Chr 19 has a ROH of length 449 ( 5.81 Mb)
  • Chr  X has a ROH of length 8878 (87.44 Mb)
  • Chr  X has a ROH of length 7096 (62.90 Mb)
  • Total ROH Mb:  163.52
  • Total genome Mb:  2884.10
  • The detected ROHs account for 5.670% of the reported genome.

The utility is taking the total of all identified ROH segments (163.52 megabase pairs) and comparing that to the total genome estimate (2884.1 Mb) and showing that as a percentage. But I'm a male. There's no such thing as ROH on the X chromosome for me. So that has to go away.

My X is showing 150.34 Mb in total ROH. So we take 163.52 and subtract 150.34 for an actual total of 13.18 Mb...and an adjusted percentage of 0.47%. Big difference!

Barring any significant population-level differences, what can you expect with recent pedigree collapse? We can look to a 2011 article in the medical journal The Lancet for some info. Referring back to what I mentioned uptopic, the Coefficient of Inbreeding (CoF), a mating of 1st-degree relatives (CoF of 1/4; father/daughter, brother/sister), would be expected to show a total ROH of about 716 Mb....or about 25% of the genome.

A pairing of 2nd-degree relatives (e.g., double 1st cousins, an uncle and a niece; a CoF of 1/8), would be expected to show a total ROH of about 358 Mb...or, ta dah!, about 12.5% of the genome.

Similarly, a child of 3rd-degree relatives, like full 1st cousins, would have a CoF of 1/16 and be expected to show a total ROH of about 179 Mb...or about 6.25% of the genome. On David Pike's page for his ROH utility, he uses as an example one set of results where "ROHs were found on five different chromosomes of a person whose parents were first cousins." The total of those ROHs are 171.32 Mb. Pretty darned close to 179.

Connie, only looking at the numbers you initially posted I can't offer an informed opinion. I think you need to either check the results from David's utility if you saved them, or run the analysis again. But if I were forced to make a guess, I'd say you'll find that your husband has no significant autozygosity and that there is no close-relation inbreeding in his past between his patrilineal and matrilineal lines.

Back to my confusion over the actual scenarios involved, that doesn't tell you anything about a paternal grandfather impregnating his son's otherwise unrelated wife, or a maternal grandmother who was impregnated by someone related to her but unrelated to the son's patrilineal line.

Autozygosity also won't show in your husband if his father were, say, the product of two 1st cousins but his mother was biologically unrelated to the father's line. So Hubby's paternal grandfather might be the result of relational consanguinity, and the father's results would show indicative ROH...but the son's, Hubby's, ROH results wouldn't reveal that because he inherited one set of chromosomes from his father, and one from his mother. ROH happens only if segments of the father's and mother's genetic sequences match. If the mother is not biologically related to the father, the resultant ROH will look like a standard, outbred population.

"I just wanted to thank you again!"

You're very welcome; don't know much help I've been, though. BTW, I didn't see your post because I was typing away...mostly one-handed; copy-and-paste done with fork held in mouth. You joined me for lunch at my desk; we had a grilled-chicken Caesar salad, heavy on the chicken, light on the dressing.
smiley

There are some other ROH utilities I'm aware of, but nothing designed like David's or GEDmatch's to be simple to use with our typical microarray data. Most of the others are standalone tools needing installation and configuration, and require the DNA data be input in very specific formats. Some are: an R programming language package developed by Filippo Biscarini and others at the University of Guelph in Canada; one from Samtools, which is one of the more popular Open Source initiatives to work with data from whole genome sequencing; and a third called PLINK, from Harvard, another whole genome association analysis toolset.

Tell Hubby "hi" for me. I feel like I know him...

Thank you! I'm beginning to understand some! LOL! So, just to be clear, If a father impregnated his daughter, then DNA declaring that would not show up  in the Daughter's baby's DNA or that of the baby's offspring?

Just making sure that I'm understanding...then I won't bother you kind, knowledgeable people anymore and I'll have my answer to look for a British/Scandanavian looking grandfather! laugh

 

LOL! I bet you do know him by now inside and out! and how was lunch? Good I hope...Just an FYI...I ran numbers again only this time with "ROHs of length at least 300 tested Snps" (i read an article that said that would be a good idea). I got the same answers! I'll email them to you. 

I have heard of PLINK. I havent heard of the other one. Plink looks complicated. Or just over my head. LOL!frown

If a father impregnated his daughter, it would absolutely show up as autozygosity in the baby's DNA, in runs of homozygosity. The expected result in the child's DNA would be somewhere around a total ROH of about 716 Mb....or about 25% of the genome. Because in that case the biological mother (the daughter) and the biological father (her father) are very definitely related, as in 1st degree of relationship.

If the baby of that union (let's assume the baby is female and call her Alice) subsequently grows up and marries a completely unrelated man and has a child (let's call this child Mary), then Mary's DNA won't display any unusual ROH because all of the inbred DNA is on only one of each pair of her diploid chromosomes. There won't be matching sets of genetic segments on the chromosomes contributed by the father.

But that only deals with runs of homozygosity in the results of a single autosomal test, that of Mary, the daughter of Alice and an unrelated man. In other words, looking at Mary's DNA results in vacuo, with nothing else to compare them to, can't tell you about inbreeding because there wouldn't be significant ROH. Looking at Alice's results, again in vacuo, would tell about the inbreeding because her parents are 1st-degree relatives.

However, Mary would still be carrying the DNA of that parent/child union. To illustrate the pedigree collapse, Mary would have the normal four grandparents on her unrelated father's side, but biologically only two on Alice's side, the incestuous father's two parents.

This would show up in Mary's DNA, but probably only when it is compared to someone else's. For example, if the incestuous father had a sibling who then had a child--call the child Mortimer--Mary would be both the niece and the 1st cousin of Mortimer. The theoretical baseline averages say that Mary should share about 25% (1700cM) of Mortimer's DNA as his niece, plus another 12.5% (850cM) as his 1st cousin, for a total of 37.5%. or around 2550cM...which is indistinguishable from a full sibling.

Given the generational distance, it's perhaps more likely that Mary would find a child of Mortimer's (call him Murgatroyd) who had taken a DNA test. To Murgatroyd, Mary would be both 1st cousin and 2nd cousin, so 12.5% (850cM) plus 3.25% (212cM) for a total of somewhere around 15.75%, or  about 1062cM...which wouldn't fit neatly with any baseline relationship: it would peg right between aunt/uncle and 1st cousin.

So Mary's ROH results wouldn't reveal the pedigree collapse. But as soon as you start comparing results--and paper trails--with other test-takers from that incestuous father's side of the family, the issue would show up fairly starkly. (See what I did there. Starkly. Game of Thrones. Lannisters. Inbreeding. <cough> Never mind. I'm infamous for not only the World's Worst Metaphors, but also terribly bad puns and attempts at humor...)

I get it! and I get the humor! haha!! I will check out all those matches that I havent really looked at yet! you all are geniuses!

Thank you so so much I really appreciate the explanation and all the time you took with me Edison. I'm a little thick but I finally got it!

I'll check it out and figure it all out!

Related questions

+2 votes
1 answer
149 views asked Jun 24, 2020 in Genealogy Help by Michelle West G2G Crew (520 points)
+4 votes
2 answers
+3 votes
2 answers
+3 votes
2 answers
1.2k views asked Mar 27, 2018 in Genealogy Help by anonymous
+16 votes
0 answers
+9 votes
1 answer
220 views asked Jul 20, 2018 in The Tree House by Eddie King G2G6 Pilot (551k points)
+6 votes
1 answer

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

disclaimer - terms - copyright

...