PDA

View Full Version : FTDNA Matches



Joel
07-28-2014, 02:11 PM
Does anyone know how the FTDNA mtDNA matches work? I have only HVR1 & 2 results right now and show 8 or 9 matches. However, I don't think those can be perfect matches as I have not seen any results that match mine so far. I couldn't find anything on how they do their matches. I'm guessing their 'matches' include at least one difference in the results.

Joel

geebee
07-28-2014, 03:16 PM
Well, they *could* be perfect matches as far as only HVR1 and HVR2 are concerned.

All I can really tell you is that if I switch from "Check matches for: HVR1, HVR2, Coding Region" to just "HVR1, HVR2", my number of matches goes from just 2 to 67. Unfortunately, no genetic distance is given. So I don't even actually know whether the matches are perfect even for the included regions.

Only when I add in the coding region does a column labeled "Genetic Distance" appear. My number of matches drops to just two, but they are both in my precise haplogroup: H1bg. One of the matches shows a genetic distance of 3, but the other is an exact match (GD of 0).

And sure enough, the "Most Distant Ancestor" for both me and the exact match is the very same woman. In addition, because we've both also taken the Family Finder test, I can see that we share segments on chromosomes 8, 10, 12, and the X.

EDIT: The 3 GD match has also taken the Family Finder test, but does not appear among my FF matches.

2nd EDIT: I should probably add that the "Most Distant Ancestor" is a 6th great grandmother to me; not sure what distance she is from my match, but probably something similar. It seems likely, given how much FF matching we also have, that she may not also be our most recent (mtDNA) ancestor.

vettor
07-28-2014, 06:32 PM
Well, they *could* be perfect matches as far as only HVR1 and HVR2 are concerned.

All I can really tell you is that if I switch from "Check matches for: HVR1, HVR2, Coding Region" to just "HVR1, HVR2", my number of matches goes from just 2 to 67. Unfortunately, no genetic distance is given. So I don't even actually know whether the matches are perfect even for the included regions.

Only when I add in the coding region does a column labeled "Genetic Distance" appear. My number of matches drops to just two, but they are both in my precise haplogroup: H1bg. One of the matches shows a genetic distance of 3, but the other is an exact match (GD of 0).

And sure enough, the "Most Distant Ancestor" for both me and the exact match is the very same woman. In addition, because we've both also taken the Family Finder test, I can see that we share segments on chromosomes 8, 10, 12, and the X.

EDIT: The 3 GD match has also taken the Family Finder test, but does not appear among my FF matches.

2nd EDIT: I should probably add that the "Most Distant Ancestor" is a 6th great grandmother to me; not sure what distance she is from my match, but probably something similar. It seems likely, given how much FF matching we also have, that she may not also be our most recent (mtDNA) ancestor.

how accurate are Genetic Distance matches?............I have a 0 GD with a person on Ftdna and yet when I enter ysearch, that same person is a GD of 15

dp
07-28-2014, 06:49 PM
Have you checked to make sure that you've both tested on the same level.
Hypothetically, they could have had say a Y-25 test from FTDNA, and subsequently tested with a different company for say a Y-111 test. Then they updated their ysearch account to the 111 STR level, and haven't ported their new results to their FTDNA kit. That could explain the disparity in GD values.
David Powell
dp :-)

vettor
07-28-2014, 07:00 PM
Have you checked to make sure that you've both tested on the same level.
Hypothetically, they could have had say a Y-25 test from FTDNA, and subsequently tested with a different company for say a Y-111 test. Then they updated their ysearch account to the 111 STR level, and haven't ported their new results to their FTDNA kit. That could explain the disparity in GD values.
David Powell
dp :-)

yes tested on same level ( 67)

dp
07-28-2014, 07:14 PM
don't know...
dp
I would print out the ysearch return for that kit and manually compare the STR values to your FTDNA kit. Look for any differences between the two kits for each STR value. If your not a match at the 12 STR level (a 12-0) I doubt you'll make a genealogical connection. Maybe it belongs to someone with just the same name as your match.
dp :-)

geebee
07-29-2014, 04:01 AM
Keep in mind that the OP had to do with mtDNA matches, not Y DNA matches. I'm not sure how genetic distance in one case is comparable to genetic distance in the other.

Not only that, but you're comparing STR matching to SNP matching -- different animals. I think that where the Y is concerned, STR matching is more useful for genealogical purposes than SNPs, though I'd imagine that should be changing with more extensive testing.

For mtDNA, I think the only testing worth looking at is FGS. Even exact matches in HVR1 and HVR2 might have *many* mismatches in the coding region. I now am up to four FGS matches, all in my haplogroup of H1bg. But only the one with a GD of 0 is definitely a recent cousin.

I don't yet know with the new addition. She and I have a GD of 1. Unfortunately, she has not also taken the Family Finder test, so we can't see if we match there; and she does not have a family tree uploaded.

I did learn, though, from FTDNA's email that they report FGS matches of up to 4 GD. I'm not sure what their criteria are for HVR1 and HVR2, but I would guess that they're only reporting exact matches. But again, since you have no idea how much mismatching there may be in the coding region, this is pretty close to meaningless.

For HVR1 and HVR2, I now have 69 matches. Most of these seem only to have tested HVR2, and their haplogroup is just reported as "H". However, I see that one of these matches has reported his Most Distant (mtDNA) ancestor as Angelique Ladner -- who was the daughter of my own MDA.

The list includes only two of my FGS matches, as far as I can tell, which tells me that at least one of the three non-matching SNPs must be in HVR1 or HVR2.

geebee
07-29-2014, 04:30 AM
how accurate are Genetic Distance matches?............I have a 0 GD with a person on Ftdna and yet when I enter ysearch, that same person is a GD of 15

The only time I ever had a match like that -- actually it was just 46/46 -- the person proved to be a 4th cousin twice removed, and on our surname line.

But I know that for Y matching, FTDNA first looks to see if you have matches at the highest level of testing you've had. If you don't, it moves to the next highest level, and so on, until it finds matches. I have no matches at 111, 67, or 37; so they have to go to 25. Here I have six matches, but all are at a GD of 2.

Strangely, 4 out of the 6 matches tested at 67 markers. So the obvious question is, why aren't they 67-marker matches with a GD of 2? Or even 37-marker matches? The only thing I can think of is that they must have an even larger GD at these levels -- too large to be considered a "match" at all.

So if FTDNA is showing you as a perfect match at less than 67 markers -- even though you've both tested at that level -- then this is probably your answer. But if they're actually showing you as 67-marker matches with a GD of 0, while ySearch shows a GD of 15 for the same level, something is very wrong.

I'd expect two people with a 67/67 match to be very close relatives; possibly even brothers, but at least cousins of no more than a very few degrees.

Joel
07-29-2014, 08:05 AM
Well, they *could* be perfect matches as far as only HVR1 and HVR2 are concerned.

All I can really tell you is that if I switch from "Check matches for: HVR1, HVR2, Coding Region" to just "HVR1, HVR2", my number of matches goes from just 2 to 67. Unfortunately, no genetic distance is given. So I don't even actually know whether the matches are perfect even for the included regions.


The reason I don't think I have a perfect match is by checking https://www.familytreedna.com/public/h%20mtdna%20haplogroup/default.aspx?section=mtresults

I know these are not all results, but it shows quite a few. In the group that I am in, I see no matches. I am 275990.

H 5 '36 - C456T
149583 Bridget Delia Hough/Hare/Howe b. 1849 & d. 1890 US H5'36 16129A 263G, 315.1C, 456T
274168 CATHERINE PEARSON b. 1850 d. 1934 H5'36 CRS 263G, 309.1C, 315.1C, 373G, 456T
275990 Nancy Roebuck (1795-1875) H5'36 CRS 263G, 315.1C, 456T

H5 is normally 456T with 16304C. H5'36 is 456T without 16304C.

Well now things are changing. When I check Matches now under FTDNA it says "new". Now all my matches are under H5'36 and I've gone down in matches from 9 to 5. However the match page doesn't show kit numbers and only has one with distant ancestor, so that makes it difficult to compare results. I see. I assume my coding region is now in as I have a list of genetic distance (as of 7/28). I have one GD - 0, 3 GD's of 1 and one of 3. When I take out the coding region, the matches go from 5 to 6 (where I used to have 9). Still a bit confusing - but interesting nonetheless.

Joel

geebee
07-29-2014, 09:27 AM
Joel

It's interesting that even without the coding region, your matches only go from 5 to 6. Mine go from 3 to 69. Some of the difference could simply be that your results are newer, and you may see your number of matches go up rapidly -- especially on HVR1, HVR2.

However, I also believe that neither 456T nor 16304C lies within the coding region, so does that mean your haplogroup is defined without having to reference the coding region? Without the coding region, I would only be in H, I think. For H1, you need G3010A -- which is in the coding region. And for H1bg, you also need G5054A and C7849T. These all lie within the coding region. (Which also means that just knowing a person's mtDNA haplogroup could allow certain health-related inferences to be drawn, in some cases.)

Anyway, this means that if FTDNA were to try to limit my matches to only persons in my haplogroup, this would be impossible except when the coding region is included. It looks as if, for you, those "near matches" that aren't in your haplogroup have now been excluded. For me, that would result in my getting zero matches except when all three regions are included.

EDIT: Or to put it differently, excluding the coding region actually "downgrades" my haplogroup, from H1bg to just H.

2nd EDIT: Now I'm confused, because I know that HVR1 is 16024-16569. and HVR2 is 00001-00576. The coding region is 00577-16023, which means that even to be in H (G2706A, T7028C) requires referencing the coding region. But once you're classified as belonging to H, the subclade H5 does not require any further reference to the coding region (nor does H5'36); whereas both H1 and H1bg do.

Joel
07-29-2014, 01:04 PM
Hi geebee,

My HVR1 was CRS, so I believe that should place me in H. C456T only occurs once on the H phylotree under H5'36. It also appears under L, D, J and R, but it should be obvious from the testing that I am already in H due to the presence or absence of other HVR1 and 2 markers.

Your group is H1bg. If you look at the phylotree, you see the control group markers in blue. It looks like from the tree that you have to be C16239T if I'm reading the tree correctly as H1bg branches from there. I assume that you are not C16242T or you would be H1bf1. That means, to me, that with just the control region, you should have been able to narrow your choices down to H1bf, H1bg, or H1bh. H17C also has C16239T but it has to also have A73G! which is in the control region.

geebee
07-29-2014, 04:43 PM
I think you're right that I could be placed in H1 without referencing the coding region directly. Even though both H and H1 are defined by SNPs within the coding region, I clearly belong to a subclade of H1. As you point out, I have C16239T, which is in HVR1; and the CRS version G73A, in HVR2.

The distinction between G73A and A73G! might be a source of confusion, but I think that simply reflects a difference between RSRS and rCRS. RSRS follows the order of ancestral version to derived version; but rCRS reports only differences from CRS. Since G73A is one of the defining SNPs for R0 -- which is "parent" to H -- CRS as a subclade of H also has G73A. I don't have a difference from CRS here, so nothing is reported.

However, H17c has a back mutation to the ancestral form -- which is why the SNP is reported as A73G! In this case, you have the derived value "A" reverting back to the ancestral value "G".

All of this said, it remains the case that out of 69 HVR1, HVR2 matches, 67 are listed simply as "H". The other two are "H1bg". Not one is identified as "H1", even though some of them probably are; you simply can't tell except by referencing the coding region. The only exception, again, would be cases in which a control region SNP allows you to identify a subclade of H1. Technically, FTDNA must be "cheating" to report the two HVR1, HVR2 matches that it calls "H1bg". Without the coding region, there would be know way to know if it's H1bf, H1bg, or H1bh. (Unless, as you note, there is also C16242T, the defining mutation for H1bf1.)

This is a bit different from your situation, in which your exact haplogroup can be determined without recourse to the coding region. (It does seem a bit puzzling, though, that H5 appears from the PhyoTree to be a subclade of H5'36, rather than the other way around.)

Joel
07-29-2014, 05:44 PM
(It does seem a bit puzzling, though, that H5 appears from the PhyoTree to be a subclade of H5'36, rather than the other way around.)

My assumption is that H5 is what a lot of people tested as being positive for over the past many years. Then later on, someone discovered a parent of H5 that was also a parent of H36, so they named it H5'36. As I look at the Phylotree under H, I see H1 to H100 all along one vertical line. However, some H numbers have been grouped in a secondary group under SNPs or in my case under H5'36. I look at it as a case of the tail wagging the dog.

Joel

geebee
07-29-2014, 05:57 PM
My assumption is that H5 is what a lot of people tested as being positive for over the past many years. Then later on, someone discovered a parent of H5 that was also a parent of H36, so they named it H5'36. As I look at the Phylotree under H, I see H1 to H100 all along one vertical line. However, some H numbers have been grouped in a secondary group under SNPs or in my case under H5'36. I look at it as a case of the tail wagging the dog.

Joel

Yes, this makes a great deal of sense. I suppose that among other reasons, too many papers have been written using "H5" and "H36" to simply shift the existing names. Things are confusing enough as is. :)

EDIT: I can also see that the whole phylotree includes some other examples of this, such as "L2'3'4'6". Presumably that means the "parent" of L2, L3, L4, and L6. So I wonder if that means the "nameless" branch defined by C16239T could be H1bf'bg'bh?

Joel
07-30-2014, 01:13 PM
I've finally answered my original question, though it is now somewhat out of date:

"On the mtDNA - Matches page, are only exact matches shown?

There are currently two sets of rules for mtDNA matching.

For those who have tested HVR1 (mtDNA) or HVR1 and HVR2 (mtDNAPlus), only exact matches are shown. SmartMatching is also used to prevent a match between different haplogroups.
For those who have tested the mtDNA Full Sequence (mtFullSequence), three differences are allowed. These differences include cases of heteroplasmy. Two high frequency insertion/deletion locations are completely excluded from difference counts. These are mutations at positions 309 and 315."

This surprised me a bit as 3 of my 'exact' matches at that time (before my full sequence came in) were H13a1a, H2a2 and H4a1a1.

On the other issue, I found the following in A ‘‘Copernican’’ Reassessment of the Human
Mitochondrial DNA Tree from its Root:

"(1) Monophyletic clades that are composed of two or more
previously named haplogroups are labeled by concatenating
their names and separating them by apostrophe
(e.g., L0a’b). This is not applied in the case of capital letter-
only labeled haplogroups (e.g., JT);"

dp
08-01-2014, 06:21 PM
geebee,
My bad, I caught the wrong end of the stick.
dp :-)


Have you checked to make sure that you've both tested on the same level.
Hypothetically, they could have had say a Y-25 test from FTDNA, and subsequently tested with a different company for say a Y-111 test. Then they updated their ysearch account to the 111 STR level, and haven't ported their new results to their FTDNA kit. That could explain the disparity in GD values.
David Powell
dp :-)

vettor
08-01-2014, 06:23 PM
The only time I ever had a match like that -- actually it was just 46/46 -- the person proved to be a 4th cousin twice removed, and on our surname line.

But I know that for Y matching, FTDNA first looks to see if you have matches at the highest level of testing you've had. If you don't, it moves to the next highest level, and so on, until it finds matches. I have no matches at 111, 67, or 37; so they have to go to 25. Here I have six matches, but all are at a GD of 2.

Strangely, 4 out of the 6 matches tested at 67 markers. So the obvious question is, why aren't they 67-marker matches with a GD of 2? Or even 37-marker matches? The only thing I can think of is that they must have an even larger GD at these levels -- too large to be considered a "match" at all.

So if FTDNA is showing you as a perfect match at less than 67 markers -- even though you've both tested at that level -- then this is probably your answer. But if they're actually showing you as 67-marker matches with a GD of 0, while ySearch shows a GD of 15 for the same level, something is very wrong.

I'd expect two people with a 67/67 match to be very close relatives; possibly even brothers, but at least cousins of no more than a very few degrees.

Does ftdna by your studies still use, for GD indications, both the Infinite method and the Step method?

dp
08-01-2014, 06:30 PM
Joel,
Besides the C-stretch variant that you mentioned, I believe that they omit an AC repeat that can occur in the hypervariable segment #2 at position 522.
David Powell,
dp :-)

GailT
08-01-2014, 11:50 PM
2nd EDIT: Now I'm confused, because I know that HVR1 is 16024-16569. and HVR2 is 00001-00576. The coding region is 00577-16023, which means that even to be in H (G2706A, T7028C) requires referencing the coding region. But once you're classified as belonging to H, the subclade H5 does not require any further reference to the coding region (nor does H5'36); whereas both H1 and H1bg do.

FTDNA does test a few coding region SNPs with the basic HVR test to confirm the haplogroup. Also, they will not predict the subclade of H based on the HVR test alone, so anyone who tested HVR will simply be assigned to H, even if they have informative HVR mutations that identify a more specific subclade. It's possible in some cases the prediction could be wrong based on HVR, so perhaps they are being overly cautious on interpreting HVR.

geebee
08-02-2014, 07:54 AM
FTDNA does test a few coding region SNPs with the basic HVR test to confirm the haplogroup. Also, they will not predict the subclade of H based on the HVR test alone, so anyone who tested HVR will simply be assigned to H, even if they have informative HVR mutations that identify a more specific subclade. It's possible in some cases the prediction could be wrong based on HVR, so perhaps they are being overly cautious on interpreting HVR.

I pretty much figured they would have to, to be able to assign anyone without the FMS to haplogroup H.

Something I find curious is that R.A. Canada of FTDNA wrote in January 2014 that


Higher testing levels (mtDNAPlus and mtFullSequence) include matching from lower testing levels. This means that someone who tests at the mtFullSequence testing level will have matching at the HVR1 level, the HVR1 & HVR2 level, and the HVR1, HVR2, & Coding Region level. https://www.familytreedna.com/learn/mtdna-testing/tell-closeness-relationship/
[emphasis mine]

The part I've highlighted doesn't quite bear out in my experience. At the FMS level, I have three matches. All are in the same haplogroup of H1bg, but the genetic distances are 0, 1, and 3 from me. However, when I change the filter to show only HVR1 and HVR2 matches, one of the matches disappears -- even though the total number of matches increases to 69. The FMS match with a GD of 3 drops out entirely, which is why I think that at least one of the mismatches must be in HVR1 or HVR2.

As I've mentioned, either in this thread or another one, the person with the GD of 0 actually does descend from the same "most distant" mtDNA ancestor. But I suspect he also must be even more closely related, since he has Family Finder results and we have matching segments on three different autosomes, and another segment on the X chromosome.

In addition, one of the 67 HVR1, HVR2 matches shows a daughter of my most distant mtDNA ancestor as his most distant mtDNA ancestor. I presume he simply doesn't know the identity of her mother.

It seems almost certain that if he were to have the FMS, his haplogroup would turn out to be H1bg, he would appear as another match to me, and his GD would most likely be 0.