GEDmatch: how to tell how closely related?

+4 votes
1.6k views
How can I tell how closely related to someone I am via a GEDmatch? We don't have any blue, but we have a lot of green, including some pretty big chunks of green. My friend only has her family tree back a couple generations. So far, we do have a surname in common: Green. I am stuck at 1858 with my Greens, and she is stuck at 1899 with her Verdes. I have no reason to think that my Greens were Spaniards, but I do have Spaniard in my genetic admixture, and her grandparents/great-grandparents immigrated from Spain.

I ran her GEDcom against many on my tree, she has been related to all of them, one branch in particular has a lot of green. (No GEDcoms are available for my Green side.)

Is there some way to determine how closely we are related?

Our GEDcoms are:

G131070
G006887

Thanks for any help in figuring this out!
in Genealogy Help by Susannah Rolfes G2G6 (8.5k points)
edited by Ellen Smith

1 Answer

+6 votes
I compared the two of you on Gedmatch and I see no matches between you and this other person. As you note, there are some green patches on the chromosome map (indicating short segments where the two of you share DNA on both copies of the chromosome), but these are very short segments and I see no reason to think they are meaningful. This person is not a DNA match for you.

If you are looking for hitherto-unknown cousins among your DNA matches, it would be better to concentrate on the good matches that appear near the top of your Gedmatch one-to-many list.
by Ellen Smith G2G Astronaut (1.5m points)

Yep; what Ellen said. I did a one-to-one compare of those two kits and even lowered the thresholds, just to check, to 400 SNPs/5cM...generally a pointless thing to do. Those kits share no detectable DNA.

Unless prepared to do some detailed analysis--both of the DNA and of the paper-trails--I generally recommend folks set the one-to-many thresholds at GEDmatch to 600 SNPs and 25cM. That puts you roughly at the level of 3rd cousins 1x removed. The next step farther out, 4C or 3C2R, would put you at about 13cM and can't be taken at face value; it will require other tested cousins and triangulation work to establish that level of match as valid evidence to an ancestral most-recent couple.

A couple of quick sources to look at if you have an autosomal DNA match and want to get an idea in what ranges the relationship may fall. First is this nifty utility by Jonny Perl that uses the data from Blaine Bettinger's Shared cM Project. Just type in the total amount of shared DNA in centiMorgans, and it will highlight the corresponding relationships in which the match might occur per the Shared cM Project data. Second is a chart showing the theoretical, mathematical DNA sharing averages. In a perfect world in which there is no pedigree collapse and DNA is halved evenly at each generation, these would be the shared amounts.

The more generations removed, the more randomized the sharing. The ranges are narrow between parents/children and full siblings. From there, the estimates become progressively rougher. The Shared cM tool is a good one for pragmatic numbers, but a few things should be recognized. The Shared cM Project data is crowd sourced, meaning individuals provide the information based on "known" relationships and the amount of DNA shown as being shared; errors are possible, and it's almost certain that zero matches are under-reported. Second, if a more distant match (say, 4th cousin or greater) shows a shared amount way higher than the mathematical average, it's indicative of enough pedigree collapse or endogamy that it may be extremely difficult to align the results with any accuracy to a given MRCA. Third, but similar, neither chart takes into account directly the possibility of double-cousins. In recent generations that may have a good paper-trail, and it would be important to understand the impact of, for example, two brothers marrying two sisters. If you know where and how the pedigree collapse occurred, you can adjust for it accordingly.

Thanks, I have been on GEDmatch for 2 months and really didn't understand the site, someone sent me a message that we were related and the amount of green between us looked really similar to what came up when I looked at my friend and I, so I thought there was only blue if it was a really close tie, especially because my husband and I don't share any green, and I saw a lot of green but no blue when I ran matches to people in my tree.

The person who sent me their kit said that my 5th great-grandmother was their 3rd great-grandmother. We share no blue. Would that be right? Or did they make a mistake?

Kits are:
G131070
T021621

Are you saying this person and I are also not related?

Based on this info, I thought my friend and I are related, since my husband and I share no green at all, and the distant cousin who claimed we were related (T021621) and I had only a little more green than my friend and I did.

Am I wrong? Was the person whose kit is T021621 wrong?

Thanks!
Regarding T021621, I just changed it to 400/5 from 500/7, and I see the match now, between me and this person who says her 3rd great-grandma is my 5th great-grandma. So, you are saying maybe this is a red herring? I clearly don't knkw enough about this subject!
If this other person claims to have found a genealogical relationship, that's worth investigating. But (if my mental arithmetic is correct) this person would be a 4th cousin twice removed. At that distance, a good DNA match is not common -- you have to be lucky to get one.
So, look if there is a paper trail, but don't bother looking if there is just a lot of green?

My friend and I share 7.6 cM on chromosome 2, with 327 SNPs.

More when I reduced it further. Does this mean anything? Is it too small to calculate? This could be hundreds of years ago? Thousands?

We also have the same mt haplogroup, we have 4.1 cM,     209 SNPs, on the X chromosome.

Any idea how distantly we would be related?
Unfortunately, that is a match that could arise by coincidence between 2 unrelated people. You may be distantly related, but you can't determine that from DNA.

My first recommendation is to consider the GEDmatch graphics display when you do a one-to-one comparison as interesting, but not informative of detail. Keep in mind that you have almost 3 billion base pairs of nucleotides in those 22 chromosomes displayed, that only about 700,000 (or some 0.03%) have been tested, and that GEDmatch is trying to squish that into a visual bar graph that fits on your screen. The bars can't even be presented in proportion horizontally with each other because their relative size differences are too great. For example, look below each bar and you'll see the image size reduction for each chromosome: at the default display of 1,000 pixels wide, chromosome 1 is at a 1/44 scale, and chromosome 22, the smallest of the lot, is at 1/8 scale...so Chr 22 is shown 5.5 times larger than it really is in relation to Chr 1.

If you want to see the graphics, watch for the blue sections. Small splotches of green are common and almost always mean a segment is what's referred to as IBS: Identical By State. Some people instead call this Identical By Chance--which is more descriptive and explanatory, but isn't a standard, recognized term. Segments of substantial length that show in green are atypical. GEDmatch is showing these as Fully Identical Regions. I've linked it so you can read more because it takes a bit to explain. Valid FIR segments are typically seen only in siblings, double-cousins, and in trees from endogamous populations or that have significant pedigree collapse. And, simply because of the way current testing works, segments that look like they're FIR may not be. So in the vast majority of instances, it's only GEDmatch's blue-colored segments that matter, the ones that show as Half-Identical Regions (HIR) shared by both kits.

The matter of segment length. This subject is like standing up in the middle of an automobile show and shouting, "Which manufacturer makes the best cars?" You'll get a cacophony of answers which range in supporting evidence from none to some. The truth is, there have as yet been no scientific, peer-reviewed studies published on the subject of segment size as produced by current DNA testing, the methods of determining same (reference genome maps, linear equations used, and the significant differences in the rate of recombination between females and males), and the validity or lack thereof in genealogy.

I'm on the side of small segments being highly suspect. One place to start exploring the issue is one of the most well-known genetic genealogists, Blaine Bettinger. Fact of the matter is that somehow 7 centiMorgans became some sort of baseline, but no one has ever been able to explain to me exactly how that was pulled out of the hat. It's what GEDmatch defaults to, and what WikiTree policy say is satisfactory for relationship triangulations. Some not-scientifically-tested but empirical data suggests that 6cM segments are false 74% of the time; 7cM segments are false 58% of the time; and 8cM segments are false 38% of the time. Not "difficult to validate"; false. Phasing helps improve how low you can go and hope a segment size is valid, but we never get to see both base-pair-level detail and phasing information (which, if done, is phasing by genotype, not actual trio phasing)...anywhere: not at GEDmatch or any of the testing providers.

What Jim Bartlett, noted genetic genealogist--who some point to as the source for saying 7cM segments are sufficient--actually wrote is this:

"However, my triangulation process involves a lot of work... I highly recommend starting with 15cM as a threshold (or even higher, if you don’t have the time or inclination). Setting a personal threshold is a good way control the amount of work you are willing to put in."

Jim's point is also that working to establish valid evidence using small segments is very complex and difficult...if it can even be done. Some think that "triangulation" with autosomal DNA simply means finding two cousins with whom you, all three of you, share a small segment and you're done. Conversely, Jim, as an example, has said that it's typical for him to work with triangulation groups that have upwards of 20 people in them before he's willing to accept a 7cM segment as valid: "If 1 or 2 of those triangulated shared segments turns out to be IBS, it's not harmful in the grand scheme."

In choosing what to examine, IMHO there are two starting points. Either, "Cousin Jane Doe and I show as both descended from our shared N great-grandparents and I want DNA substation for that," or "I've found a seeming DNA match and I want to see if it is valid and fits in the family tree."

For the latter, I recommend looking at the two charts I linked earlier: one for the mathematical average DNA sharing, and one using the data from the Shared cM Project. The farther removed the relationship, the more difficult and time consuming to validate. And, in fact, to even consider more distant cousin DNA validation, you have to have built a dataset of closer cousins as a benchmark to understand how the family DNA fits together. A free utility to help with that effort is DNA Painter; others, like Genome Mate Pro, are also available.

If you're simply looking at reported matches and want to start exploring how they might fit into your genealogy, I strongly recommend doing it in steps. Start with the matches that show the largest amount of in-common DNA. I'd set my sights on GEDmatch by limiting the display threshold to 75cM; at that level the SNP density really won't matter. That will encompass some 3rd cousins and closer. Some you'll know right away where they fit in the family tree, and others won't respond to emails; but keep trying. Work this group first and start mapping the chromosomes so that you'll better understand if and where a more distant match fits into the family.

When you get down to 3C2R and 4th cousins (shared 3g-grandparents) you're getting into rarefied air where it takes significant detail and effort to validate the match. At that point you'll be roughly at around 15cM to 30cM total DNA shared. Looking under 10cM is not for the meek, and adequately validating the atDNA evidence for 5th cousins is not easy or quick. Only 14.9% of your 5th cousins are going to share any detectable DNA with you at all, so it takes luck as well as work. That puts the odds at 17:3...or for every 20 5th cousins of yours who take a DNA test, you can hope to match only three of them.

We all can use some luck!  :-)

Related questions

+5 votes
3 answers
+1 vote
3 answers
+11 votes
7 answers
+11 votes
2 answers

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

disclaimer - terms - copyright

...