PDA

View Full Version : Living DNA files to GedMatch Genesis - matches



A Norfolk L-M20
06-26-2017, 10:29 PM
Are yours working okay? Do your Living DNA matches line up on Gedmatch Genesis? I've had a few odd results, and feel the need to check with others for validity of the system.

Mike_G
06-27-2017, 12:06 AM
Are yours working okay? Do your Living DNA matches line up on Gedmatch Genesis? I've had a few odd results, and feel the need to check with others for validity of the system.

Something's screwy because you're on my match list. As is Amerijoe.

Sorry fellas. :P

Edit: Strange thing is you and I have a 7.5 CM match at one chromosome, yet I show zero for Joe.
Edit 2: Had it set to default. Joe and I have a 5 CM match.

C J Wyatt III
06-27-2017, 01:10 AM
Something's screwy because you're on my match list. As is Amerijoe.

Sorry fellas. :P

Edit: Strange thing is you and I have a 7.5 CM match at one chromosome, yet I show zero for Joe.
Edit 2: Had it set to default. Joe and I have a 5 CM match.

Consider the possibility that GEDmatch Genesis is correct and there is a common male ancestor about three hundred years ago. I say male, because females cannot reproduce fast enough to have that kind of impact.

If you try triangulating 'X one-to-many' matches using GEDmatch defaults, it is hard not to come to that conclusion after doing a few dozen. My mother has over 10,000 matches. I've seen other females with over 25,000 X-matches and even one with over 50,000. With a 7.0 cM minimum segment for matching, and only 196 cM's on one X-chromosome for a male, there is a hell of a lot triangulation going on.

These segments are not IBD, but a reconstructed piece of the ancestor's DNA by having multiple lines going back to him.

When comparing males, you only have a single X comparing with another single X, yielding no reconstructed matches. Male to female allows two X's for the females to make a reconstructed match and you do a little bit better on the matching front. However with a pair of females, each having two X chromosomes, the reconstruction possibilities explode. That is why females have so many X-matches.

Something to think about out of the mainstream.

Jack Wyatt

firemonkey
06-27-2017, 01:15 AM
Is anyone a match with me ? NG6639872.

Mike_G
06-27-2017, 01:44 AM
Is anyone a match with me ? NG6639872.

Yes we share a 5 CM segment.

That makes 4 Anthrogenica/Living DNA matches for me so far. Three Brits, one Yank. Sorry for the threadjack, Norfolk.

But for possible anomalies, my highest match on Genesis (he's a Living DNA match as well) is also on Ancestry, but he doesn't show up on my matches over there.

Solothurn
06-27-2017, 01:51 AM
Comparing Kit XA9576266 (*Solothurn) and NG6639872

Largest segment = 7.9 cM
Total of all matching segments > 1 cM = 160.0 cM (4.5 Pct)

Could be a glitch as you don't show on my one to many list :(

http://www.anthrogenica.com/images/AegeanE/misc/quote_icon.png Originally Posted by firemonkey http://www.anthrogenica.com/images/AegeanE/buttons/viewpost-right.png (http://www.anthrogenica.com/showthread.php?p=252313#post252313)
Is anyone a match with me ? NG6639872.

C J Wyatt III
06-27-2017, 01:51 AM
Yes we share a 5 CM segment.

That makes 4 Anthrogenica/Living DNA matches for me so far. Three Brits, one Yank. Sorry for the threadjack, Norfolk.

But for possible anomalies, my highest match on Genesis (he's a Living DNA match as well) is also on Ancestry, but he doesn't show up on my matches over there.

AncestryDNA sometimes cuts out what it doesn't like or can't explain. TIMBER!

Jack Wyatt

jortita
06-27-2017, 06:29 AM
My top match for Living DNA on Genesis Gedmatch is an East Asian person who when I run calculators appears as Han (60-75%)+Dai (25-40%). Reza is also among my top matches

firemonkey
06-27-2017, 02:22 PM
Comparing Kit XA9576266 (*Solothurn) and NG6639872

Largest segment = 7.9 cM
Total of all matching segments > 1 cM = 160.0 cM (4.5 Pct)

Could be a glitch as you don't show on my one to many list :(

http://www.anthrogenica.com/images/AegeanE/misc/quote_icon.png Originally Posted by firemonkey http://www.anthrogenica.com/images/AegeanE/buttons/viewpost-right.png (http://www.anthrogenica.com/showthread.php?p=252313#post252313)
Is anyone a match with me ? NG6639872.




You show up using the validity measurement tool.



Chr B37 Start Pos'n B37 End Pos'n Centimorgans (cM) SNPs
20 58,678,932 60,234,239 7.9 503


Largest segment = 7.9 cM
Total of all matching segments > 7 cM = 7.9 cM (0.2 Pct)
Estimated number of generations to MRCA = 7.4

596888 SNPs used for this comparison.

06-27-2017, 02:29 PM
Is anyone a match with me ? NG6639872.

No sorry :behindsofa:

I think I only Match
XA9576266 A *Solothurn Largest cm 5.4
my kit is Kit: QA3736362 (Stephen Grant-Davies) [Living DNA]

Pylsteen
06-27-2017, 02:29 PM
How do they look for matches? In my list the "closest" match I share 38 cM with (seemingly through shared Ashkenazi ancestry, indeed corresponding to this type of ancestry on four chromosomes it shows shared cM's with) but in the one to one and the validity, the amount of shared cM's is 0.

Solothurn
06-29-2017, 06:32 AM
When comparing on 'GEDmatch Validity Measurement Autosomal Comparison Entry Form' if the threshold is changed to 1 as FTDNA can do, I think the more ancient segments are shown.

Then again I tried this with 3 LDNA kits (including QA3736362 and NG6639872) and got well above 100cM so not sure what is going on :(

chelle
06-29-2017, 10:50 AM
Is anyone a match with me ? NG6639872.

If I lower it to 5 we match at 5.4 on ch. 16 and 5.3 on 19. My kit is HA8458205


I also match with Solothurn at 8.6 on 15.

chelle
06-29-2017, 11:25 AM
Something's screwy because you're on my match list. As is Amerijoe.

Sorry fellas. :P

Edit: Strange thing is you and I have a 7.5 CM match at one chromosome, yet I show zero for Joe.
Edit 2: Had it set to default. Joe and I have a 5 CM match.

I think all of you, along with some others are on my list, but when I do the comparison, I have to lower the threshold to see a match.

Solothurn
06-29-2017, 05:17 PM
You must be psychic as you are the other LDNA kit I compared :)

When lowered to 1cm we match a total of 190cM, do you think this is 'real' or glitch?


If I lower it to 5 we match at 5.4 on ch. 16 and 5.3 on 19. My kit is HA8458205


I also match with Solothurn at 8.6 on 15.

Amerijoe
06-29-2017, 06:07 PM
O.K. Let's see who's my closest cousin.

Firemonkey >6=6.6cM=169.1cM
Solothurn. >5=5.8cM=170.1cM
sgdavies. >7=7.6cM=175.4cM*
chelle. >6=7.0cM=158.2cM

* Winner! Prizes are awarded with proof of match. :)

chelle
06-29-2017, 07:19 PM
O.K. Let's see who's my closest cousin.

Firemonkey >6=6.6cM=169.1cM
Solothurn. >5=5.8cM=170.1cM
sgdavies. >7=7.6cM=175.4cM*
chelle. >6=7.0cM=158.2cM

* Winner! Prizes are awarded with proof of match. :)

It is an honor just to be in the running. :lol:

sktibo
06-29-2017, 07:23 PM
O.K. Let's see who's my closest cousin.

Firemonkey >6=6.6cM=169.1cM
Solothurn. >5=5.8cM=170.1cM
sgdavies. >7=7.6cM=175.4cM*
chelle. >6=7.0cM=158.2cM

* Winner! Prizes are awarded with proof of match. :)

If that Firemonkey match is genuine it might mean you actually have Aberdeenshire ancestry and that your new Aberdeenshire LDNA percentage might be real?

chelle
06-29-2017, 07:24 PM
You must be psychic as you are the other LDNA kit I compared :)

When lowered to 1cm we match a total of 190cM, do you think this is 'real' or glitch?

;) Good minds think alike. I'm not sure about accuracy. I haven't really played around on the genesis site too much yet.

Amerijoe
06-29-2017, 07:34 PM
It is an honor just to be in the running. :lol:

Cousin of mine, when it's morning here and evening at your end, if I contact you do I say good morning or good evening? Well, I'll play it safe, CHEERS! :)

Amerijoe
06-29-2017, 07:54 PM
If that Firemonkey match is genuine it might mean you actually have Aberdeenshire ancestry and that your new Aberdeenshire LDNA percentage might be real?

As I have previously mentioned, a couple of family surnames may have origins in this particular region. Unfortunately no paper trail. Also may be from my paternal side which is shrouded in mystery.

A note of interest, my new BigY match is from England, but whose surname's origin is very likely to be from this region considering a town carries the family name. The match should be made available for public view within a short time. The TMRCA should be closer to modern times hopefully.

chelle
06-29-2017, 08:12 PM
Cousin of mine, when it's morning here and evening at your end, if I contact you do I say good morning or good evening? Well, I'll play it safe, CHEERS! :)

hahaha After living overseas for so many years and timezones I have given up. I usually just say good morning, afternoon and/or evening. Luckily, since no one talks on phones that much anymore, it is rarely an issue.

firemonkey
06-29-2017, 08:51 PM
O.K. Let's see who's my closest cousin.

Firemonkey >6=6.6cM=169.1cM
Solothurn. >5=5.8cM=170.1cM
sgdavies. >7=7.6cM=175.4cM*
chelle. >6=7.0cM=158.2cM

* Winner! Prizes are awarded with proof of match. :)

What is your kit number and what settings did you use?

Amerijoe
06-29-2017, 09:18 PM
What is your kit number and what settings did you use?

I started out with the standard one to one settings and kept lowering minimum segment size downward one digit at a time until I get a confirmed segment match. The totals are at 1cM which all are in the 4%+ range for us all. Let me get over to g.g for my I'd and get back to you.

JonikW
06-29-2017, 09:52 PM
O.K. Let's see who's my closest cousin.

Firemonkey >6=6.6cM=169.1cM
Solothurn. >5=5.8cM=170.1cM
sgdavies. >7=7.6cM=175.4cM*
chelle. >6=7.0cM=158.2cM

* Winner! Prizes are awarded with proof of match. :)

Doesn't this level of relatedness seem remarkable given that people usually get 1,000 or so matches on 23andme with more than a million testers? Do any of you guys who match here also match on other tests? I haven't managed to upload my results yet.

evon
07-18-2017, 02:00 PM
I uploaded mine, but it gave me a bunch or errors when processing, basically an endless list of missing SNP's, should I reupload, or is this common for a LivingDNA file?

Pylsteen
07-18-2017, 02:26 PM
I uploaded mine, but it gave me a bunch or errors when processing, basically an endless list of missing SNP's, should I reupload, or is this common for a LivingDNA file?

On genesis or on regular gedmatch? The regular one doesn't work properly (yet?).

ollie444
07-18-2017, 02:33 PM
I uploaded mine, but it gave me a bunch or errors when processing, basically an endless list of missing SNP's, should I reupload, or is this common for a LivingDNA file?


On genesis or on regular gedmatch? The regular one doesn't work properly (yet?).

I didn't get any errors on either.

Pylsteen
07-18-2017, 02:36 PM
I didn't get any errors on either.


Ok, the normal gedmatch works, it is the batch processing that doesn't work properly there; or has your kit no two stars in front anymore?

evon
07-18-2017, 03:13 PM
On genesis or on regular gedmatch? The regular one doesn't work properly (yet?).

Genesis, I will try again... Worked now, I think I might have pressed the other option by mistake as I was in a hurry when I uploaded it.. Will be interesting to see the differences with my 23andme V3 file..

Kathlingram
07-18-2017, 03:34 PM
Genesis, I will try again... Worked now, I think I might have pressed the other option by mistake as I was in a hurry when I uploaded it.. Will be interesting to see the differences with my 23andme V3 file..

My cousin Bill Howard's blew up the first time and he did again.. now both are there.. My results say *KatieC or Kathleen Carrow Ingram ..
I only identify a FEW match..10 or so..
Living_DNA tells me that is the "Third party's" but I think it is both.. the ancient samples we may match at Living may or may not get kicked out with other algorithms

Amerijoe
07-18-2017, 03:56 PM
Both the VCF file at genesis and the text file at Gedmatch have approx. 52K snps, which is about 1/3 that of the other testing companies. I would not give too much affirmation to it's admixture results. The LDNA text works with the exception of one to many matches. It has two stars but has not been fully tokenized since uploaded over a week ago. Slight differences between the VCF and text files in admixing.

Kathlingram
07-18-2017, 04:37 PM
Both the VCF file at genesis and the text file at Gedmatch have approx. 52K snps, which is about 1/3 that of the other testing companies. I would not give too much affirmation to it's admixture results. The LDNA text works with the exception of one to many matches. It has two stars but has not been fully tokenized since uploaded over a week ago. Slight differences between the VCF and text files in admixing.

Amerijoe What does this mean:"It has two stars but has not been fully tokenized since uploaded over a week ago" where are those stars?
Thank you! Kathleen

Pylsteen
07-18-2017, 04:46 PM
Both the VCF file at genesis and the text file at Gedmatch have approx. 52K snps, which is about 1/3 that of the other testing companies. I would not give too much affirmation to it's admixture results. The LDNA text works with the exception of one to many matches. It has two stars but has not been fully tokenized since uploaded over a week ago. Slight differences between the VCF and text files in admixing.


If you look at the option "DNA file diagnostic utility" it says (at least with me) that it won't be batch processed due to too few SNPs.

evon
07-18-2017, 06:43 PM
This is interesting, my grandmother (who is a 23andme V3 kit) has a larger segment match with my LivingDNA kit than with my 23andme V3 kit (more in total too actually).

Largest Seg - Total cM
148.052 - 1748.22 LivingDNA
129.87 - 1740.94 23andme V3

I also see I have a lot more matches with my LivingDNA file, but I dont match the BURUSHO sample, as does my 23andme V3 kit.

Amerijoe
07-18-2017, 07:33 PM
If you look at the option "DNA file diagnostic utility" it says (at least with me) that it won't be batch processed due to too few SNPs.

Thanks, didn't think about diagnostic tool. Ran through and stated the following. Kit did not tokenize properly. Try deleting the kit and uploading it again. I'll try again. :)

Amerijoe
07-18-2017, 07:37 PM
Amerijoe What does this mean:"It has two stars but has not been fully tokenized since uploaded over a week ago" where are those stars?
Thank you! Kathleen

It means I should have ran a file diagnostic check as mentioned by Plysteen. :)

Kathlingram
07-18-2017, 07:58 PM
It means I should have ran a file diagnostic check as mentioned by Plysteen. :)

So I after 6 years did not know there WAS a diagnostic tool.. :\

Pylsteen
07-18-2017, 08:09 PM
Though I don't believe re-uploading will work, since the problem remains "too few SNPs" in the txt-file. Good luck though.

Reza
07-18-2017, 10:30 PM
My top match for Living DNA on Genesis Gedmatch is an East Asian person who when I run calculators appears as Han (60-75%)+Dai (25-40%). Reza is also among my top matches

So I only just discovered this today! You match with my father too.

Interestingly, noman comes up with a total of 13.2cM and a small segment with heksindhi.

There are various pathans, more on mother's side, and baloch / brahui that khana has kindly uploaded.

Is there any difference to the standard gedmatch matches when comparing with 23andme?

Pylsteen
07-19-2017, 03:56 PM
I have a lot of matches on a specific part of chromosome 22; that can't be real.

Now I came across this site (http://www.khazaria.com/genetics/excess-ibd-regions.html), that lists problematic areas (many potential false matches):

Chromosome 1 from 118 million to 153 million
Chromosome 2 from 85 million to 99 million
Chromosome 2 from 132 million to 141 million
Chromosome 2 from 192 million to 198 million
Chromosome 8 from 10 million to 13 million
Chromosome 9 from 38 million to 72 million
Chromosome 10 from 44 million to 53 million
Chromosome 15 from 20 million to 25 million
Chromosome 15 from 27 million to 30 million
Chromosome 16 from 19 million to 24 million
Chromosome 17 from 59 million to 65 million
Chromosome 17 from 77 million to 78 million
Chromosome 21 from 16 million to 19 million
Chromosome 22 from 16 million to 25 million

Amerijoe
07-19-2017, 04:29 PM
Though I don't believe re-uploading will work, since the problem remains "too few SNPs" in the txt-file. Good luck though.

Verified, reload produced 51.9K snps. Is there any timeframe from LDNA as to the resolution of this issue?

Kathlingram
07-19-2017, 04:51 PM
Verified, reload produced 51.9K snps. Is there any timeframe from LDNA as to the resolution of this issue?

The NEW Download has many many more SNPS.. Are you sure you are using the latest one? They rereleased it a week or so ago
Mine has 58,981 SNPS and that is compared to 23andme which has 182,701

They asked you to not use the original one at all.. They had said Mid July but mine was replaced on 7/11 and they messaged me this:"We have now updated your Autosomal Raw Data and this can now be downloaded in the Raw Data tab of your results portal if you would like to try again to download this."

ollie444
07-19-2017, 05:20 PM
The NEW Download has many many more SNPS.. Are you sure you are using the latest one? They rereleased it a week or so ago
Mine has 58,981 SNPS and that is compared to 23andme which has 182,701

They asked you to not use the original one at all.. They had said Mid July but mine was replaced on 7/11 and they messaged me this:"We have now updated your Autosomal Raw Data and this can now be downloaded in the Raw Data tab of your results portal if you would like to try again to download this."

Mine has 178102 'regular' SNPs on the file diagnostic utility and 58,232 on K13. My kit still has two red stars by it. I thought someone received an email response from Living DNA saying that the problem was at gedmatch's end, as they don't currently use lots of the SNPs Living DNA test for? Not sure about that though.

ollie444
07-19-2017, 05:23 PM
Just looking at the no-calls rate. Living DNA seems to be pretty good compared to my ancestryDNA kit, where the no-calls are over 2%!

Kathlingram
07-19-2017, 05:26 PM
Mine has 178102 'regular' SNPs on the file diagnostic utility and 58,232 on K13. My kit still has two red stars by it. I thought someone received an email response from Living DNA saying that the problem was at gedmatch's end, as they don't currently use lots of the SNPs Living DNA test for? Not sure about that though.

That is probably true.. I was only going with the SNPS required to do teh K13 which was 58, 232 and 23andme was 178,102 or at least that is what the Utility said when completed..

Kathlingram
07-19-2017, 05:29 PM
Mine has 178102 'regular' SNPs on the file diagnostic utility and 58,232 on K13. My kit still has two red stars by it. I thought someone received an email response from Living DNA saying that the problem was at gedmatch's end, as they don't currently use lots of the SNPs Living DNA test for? Not sure about that though.

I received an email but what they said was that Gedmatch being the 3rd party was responsible for false matches or small matches being reported as 4-5 generations.. My feeling is that it is a process that will be worked through and has to do with the geographical informative matches.. some companies remove them..

evon
07-19-2017, 09:34 PM
I see I match a few forum members on my LivingDNA kit... Does anyone know what the lowest level of SNP's Genesis uses for their matching algorithm?

evon
07-19-2017, 09:39 PM
My top match for Living DNA on Genesis Gedmatch is an East Asian person who when I run calculators appears as Han (60-75%)+Dai (25-40%). Reza is also among my top matches

I see we match :P I also match with Norfolk, Amerijoe, etc..

I have 178 LivingDNA matches in total, mostly Germans and Anglo sounding names on top, sprinkled with a few Latin American names and various other groups such as Jews further down.. I am actually surprised by the high number of non-Brits to have tested with LivingDNA.. I see khanabadoshi has been going nuts with the HGDP etc samples... :P I match him and several of the uploaded samples, my top being a Pathan individual..

Pylsteen
07-19-2017, 10:01 PM
Everything below 7cM is generally seen as uncertain (though could be real in some cases); everything above 7cM is likely real, if it does not occur on certain positions which people often seem to share, see also here (http://www.khazaria.com/genetics/excess-ibd-regions.html).

evon
07-19-2017, 10:10 PM
Everything below 7cM is generally seen as uncertain (though could be real in some cases); everything above 7cM is likely real, if it does not occur on certain positions which people often seem to share, see also here (http://www.khazaria.com/genetics/excess-ibd-regions.html).

I am familiar with that, I have recently had a round with excessive IDB's on the 10th chromosome. My concern is the SNP count, as some of these matches must have a very low SNP count, which makes them highly suspect, this includes those above 7cM... I am also curious if the total cM includes tiny matches..I guess I will have to e-mail John about it (The man behind Gedmatch), but he is a very busy man so I was hoping somebody here already knew the number... I will take a look at some of the Asian matches tomorrow as I suspect some of them will have a low SNP count..

I just had a quick look, the SNP count is very low for some of the matches, I match a Pathan sample at 8.2cM with only 253 SNP's, that is very low, so low in fact I dont trust it at all...

Pylsteen
07-19-2017, 10:21 PM
Ok. I'm not sure how many SNPs are needed for a significant potential match.

Kathlingram
07-19-2017, 10:22 PM
I am familiar with that, I have recently had a round with excessive IDB's on the 10th chromosome. My concern is the SNP count, as some of these matches must have a very low SNP count, which makes them highly suspect, this includes those above 7cM... I am also curious if the total cM includes tiny matches..I guess I will have to e-mail John about it (The man behind Gedmatch), but he is a very busy man so I was hoping somebody here already knew the number... I will take a look at some of the Asian matches tomorrow as I suspect some of them will have a low SNP count..

I just had a quick look, the SNP count is very low for some of the matches, I match a Pathan sample at 8.2cM with only 253 SNP's, that is very low, so low in fact I dont trust it at all...

Evon, When you message John ask him if the high number that indicates probablility relates to the low SNP count.. The matches that I have with Living DNA testers seem to have 95 and 96.. others are nice and low..( per what he says in instructions.) Thanks also ( reason for edit) if you do their version of One to One match ( the proof one), yes they do have small matches.. like 3 cM, 5 cM etc.

evon
07-20-2017, 09:39 AM
Ok. I'm not sure how many SNPs are needed for a significant potential match.

That is the big question. Most DNA companies have a threshold between 500-800 SNP's, while most of my top matches via Genesis have between 200-450 SNP's, which makes me very skeptical..

From Gedmatch genesis:

Segment threshold size will be adjusted dynamically above 200 SNPs
Minimum segment cM to be included in total = 5.0 cM


Evon, When you message John ask him if the high number that indicates probablility relates to the low SNP count.. The matches that I have with Living DNA testers seem to have 95 and 96.. others are nice and low..( per what he says in instructions.) Thanks also ( reason for edit) if you do their version of One to One match ( the proof one), yes they do have small matches.. like 3 cM, 5 cM etc.

It seems the lowest threshold for matching segments in the "one to many" are 200SNP's and 5cM, so I dont really have to send that e-mail after all. I also looked the confidence estimation, and I noticed that my matches with a probability at around 1.00 C overlaps with matches where the SNP count for at least one segment is also high. So low confidence = low SNP count (at least it seems to be the case with most of my matches).

Let me illustrate with a few of my matches:



Largest Seg - Total cM - Confidence (in bold) -Highest SNP count
1. 9.46 - 40.3126 - 0.98 - 1,867
2. 11.531 - 28.168 - 0.12 - 346
3. 6.0227 - 6.0227 - 0.48- 313


With match 3 there is actually an error, as we match on two segments at 6.2cM each (with different SNP count), but in the list he is listed as only matching me on one segment, thus the total should be 12cM and not 6cM. So maybe his low SNP count and high confidence is a mistake.

To illustrate the problem of low SNP count, here are some matches on my X Chromosome that I found very peculiar:
http://i67.tinypic.com/ek25hv.jpg

It is quite possible that they are real, as they appear to overlap but my guess is they are very old and should not be taken too seriously, mainly due to the low SNP count.

Kathlingram
07-20-2017, 12:57 PM
That is the big question. Most DNA companies have a threshold between 500-800 SNP's, while most of my top matches via Genesis have between 200-450 SNP's, which makes me very skeptical..

From Gedmatch genesis:




It seems the lowest threshold for matching segments in the "one to many" are 200SNP's and 5cM, so I dont really have to send that e-mail after all. I also looked the confidence estimation, and I noticed that my matches with a probability at around 1.00 C overlaps with matches where the SNP count for at least one segment is also high. So low confidence = low SNP count (at least it seems to be the case with most of my matches).

Let me illustrate with a few of my matches:



With match 3 there is actually an error, as we match on two segments at 6.2cM each (with different SNP count), but in the list he is listed as only matching me on one segment, thus the total should be 12cM and not 6cM. So maybe his low SNP count and high confidence is a mistake.

To illustrate the problem of low SNP count, here are some matches on my X Chromosome that I found very peculiar:
http://i67.tinypic.com/ek25hv.jpg

It is quite possible that they are real, as they appear to overlap but my guess is they are very old and should not be taken too seriously, mainly due to the low SNP count.

Evon
Good for you!! Thanks.. I was thinking also that there is something wrong and it was because they are ancient BUT am not as scientific as you are.. Boy we have been hanging around together a long time!! Good Job :D

Pylsteen
07-20-2017, 01:00 PM
If I compare my vcf to my txt file on genesis, it says it uses 596938 SNPs. This would imply that the file(s) are complete. I would like to see X-chrom data though.
Likely later this year gedmatch and genesis will merge into a new database, I heard.

Kathlingram
07-20-2017, 01:54 PM
If I compare my vcf to my txt file on genesis, it says it uses 596938 SNPs. This would imply that the file(s) are complete. I would like to see X-chrom data though.
Likely later this year gedmatch and genesis will merge into a new database, I heard.

Neither Evon and I can see that as our download is newer and Living DNA asked me not to use the other one.. I do remember many of you indicating they matched each other? That seems odd..My matches that can be confirmed were from the regular Gedmatch.. FTDNA and AncestryDNA.. I cannt seem to find the 23andme matches..
ALL of them phase these samples I think..

Pylsteen
07-20-2017, 02:09 PM
Neither Evon and I can see that as our download is newer and Living DNA asked me not to use the other one.. I do remember many of you indicating they matched each other? That seems odd..My matches that can be confirmed were from the regular Gedmatch.. FTDNA and AncestryDNA.. I cannt seem to find the 23andme matches..
ALL of them phase these samples I think..

They match almost perfectly, a few SNPs seem slightly different. When I open the vcf in excel, it shows about 602k SNPs. The txt-file seems to contain those, and additionally the X-chromosome, which doesn't occur in the vcf. I think the problem with "too few SNPs" therefore lies within gedmatch.
And yes, we should proceed with the txt-file.

Ann Turner
07-20-2017, 03:17 PM
Let me illustrate with a few of my matches:

The C column is not confidence -- it's a metric to quantify the SNP coverage for cross-platform comparisons. John Olson explained it to me:


C = (2*overlap)/(size1 + size2)

Varies between 0 and 1, depending on how similar the SNP templates of the 2 kits are.

It's very highly correlated with the Overlap column, although I haven't figured out the exceptions.

This table from the ISOGG Wiki has some stats for cross-platform comparisons.

https://isogg.org/wiki/Autosomal_SNP_comparison_chart

LivingDNA files don't get tokenized by the regular GEDmatch site, although they can be used for one-to-one. My hunch is that there's a hard threshold of 5000 SNPs per chromosome.

Ann Turner
07-20-2017, 03:25 PM
I did a comparison of Genesis match lists for my son's 23andMe v3 results and his LivingDNA results. I also uploaded phased data for my son based on 23andMe results.

LivingDNA has fewer SNPs available for cross-platform comparisons (180,565 for those two platforms, probably as high as it will get). LivingDNA gave more total matches but fewer matches that survived phasing. I think it is going to be very tricky to develop a robust algorithm when the number of SNPs drops that low.

17635

Kathlingram
07-20-2017, 04:45 PM
[QUOTE=Ann Turner;263123]The C column is not confidence -- it's a metric to quantify the SNP coverage for cross-platform comparisons. John Olson explained it to me:

Thanks for that Ann Turner I thought that was what I read..
apparently I read too fast.. So if myself and someone called Lizzy have this value:"0.96" as oppsed to this one which George and I have :"0.25".. what is the difference? Lizzy is better or worse?

Ann Turner
07-20-2017, 09:14 PM
[QUOTE=Ann Turner;263123]The C column is not confidence -- it's a metric to quantify the SNP coverage for cross-platform comparisons. John Olson explained it to me:

Thanks for that Ann Turner I thought that was what I read..
apparently I read too fast.. So if myself and someone called Lizzy have this value:"0.96" as oppsed to this one which George and I have :"0.25".. what is the difference? Lizzy is better or worse?
You did see the word confidence in the headers of the match list, but it's in quotes. It's more a measure of how similar two kits are in SNP coverage. If you compare a 23andMe v3 test with another version 3 test, the C value will be close to 1. If you compare a LivingDNA file with another LivingDNA file, you will also see a number close to 1. If you compare a v3 test with a LivingDNA test, you will see a number in the 0.25 range (based on my son's match list). It doesn't mean that the match is bad, just that it's based on fewer data points. The real question is the minimum number of data points we need.

Kathlingram
07-20-2017, 09:33 PM
I guess we don't know that answer.. well I really don't as I have a Social Science degree..Anthropology lends itself to migrations but not numbers and segments.. but I learned a lot through this..Thanks so much

psampson
07-22-2017, 09:40 PM
Are yours working okay? Do your Living DNA matches line up on Gedmatch Genesis? I've had a few odd results, and feel the need to check with others for validity of the system.

A Norfolk, you are on my list, 5 segments, largest segment 6.3cM total 33.3cM. SNP count between 344-990.
firemonkey is also matching, but only 2 segs each of 5.1cM
sgdavie matches on 3 segments Largest segment = 5.5 cM total of 15.8 cM
chelle - 5 segments, largest segment 7.1 cM Total of all matching segments > 5 cM = 30.8 cM

Pylsteen
07-22-2017, 09:54 PM
psampson, I think we all match each other on genesis :-) I share with you 8.3cM (1956 snps, quite solid I think?) on chrom 5.
Well, it can't be repeated often enough: double check with the Validity measurement tool, beware of anything below 7cM,
and watch out for false positives which often occur on certain chromosome parts (excess IBD sharing, see here (https://isogg.org/wiki/Identical_by_descent), point 4).

evon
07-22-2017, 10:12 PM
You did see the word confidence in the headers of the match list, but it's in quotes. It's more a measure of how similar two kits are in SNP coverage. If you compare a 23andMe v3 test with another version 3 test, the C value will be close to 1. If you compare a LivingDNA file with another LivingDNA file, you will also see a number close to 1. If you compare a v3 test with a LivingDNA test, you will see a number in the 0.25 range (based on my son's match list). It doesn't mean that the match is bad, just that it's based on fewer data points. The real question is the minimum number of data points we need.

This is quite interesting, look at the cM and SNP when I compare my grandmothers 23andme V3 kit to my 23andme V3 kit and my LivingDNA kit:

My grandmothers 23andme v3 vs my LivingDNA:

Largest segment = 148.0 cM
Total of all matching segments > 5 cM = 1824.7 cM (50.9 Pct)

176313 SNPs used for this comparison.

My grandmothers 23andme v3 vs my 23andme v3 kit:

Largest segment = 129.8 cM
Total of all matching segments > 5 cM = 1809.6 cM (50.5 Pct)

923693 SNPs used for this comparison.

firemonkey
07-22-2017, 11:26 PM
Quoting from gedmatch genesis.



Confidence' in a result is indicated by a value between zero and one in the 'C' column. This value is primarily based on the amount of overlap between the 2 kits being compared.

chelle
07-23-2017, 10:20 AM
I had someone reach out to me to say I matched their father on Genesis at 5.2 generations back When I compared my kit to the other kit on one to one there was absolutely no match that showed up. I compared his kits to each of my parents on the regular Gedmatch page and there was no match until I changed the threshold to 3 and then there was a slight match to both. (shrug) I wish I knew more about how to interpret these things. Every time I think I am getting the hang of this DNA thing, I get confused all over again.

evon
07-23-2017, 10:42 AM
Quoting from gedmatch genesis.

Yes this becomes obvious when one looks at the SNP numbers when comparing between two kits. I found the cM values to be quite interesting in this respect though, as I was expecting it to follow a similar pattern to the SNP, but it seems the lack of SNP's do not impact the cM count in the manner I was expecting... My guess is that Genesis simply fills in the blanks between segments in the name of uniformity, in a somewhat similar way to 23andme's smoothing algorithm.

My 23andme v3 and my LivingDNA kit matching on my grandmothers 23andme v3 6th chromosome (notice the cM and SNP count):
http://i64.tinypic.com/4r9jbl.jpg

firemonkey
07-23-2017, 10:58 AM
I am not sure how seriously to take a low snp match. Which is deemed more relevant a high cM low snp match or low cM high snp match? The fact that quite a few of us are listed as matches with each other seems a little suspicious to me. I suspect that the % of matches between members here is higher than could reasonably be expected.

Pylsteen
07-23-2017, 11:15 AM
I am not sure how seriously to take a low snp match. Which is deemed more relevant a high cM low snp match or low cM high snp match? The fact that quite a few of us are listed as matches with each other seems a little suspicious to me. I suspect that the % of matches between members here is higher than could reasonably be expected.


Since the LivingDNA kits use ca. 600k snps, and they all overlap, they would likely result in many more matches, real or not, I suppose.

Kathlingram
07-23-2017, 12:34 PM
Yes this becomes obvious when one looks at the SNP numbers when comparing between two kits. I found the cM values to be quite interesting in this respect though, as I was expecting it to follow a similar pattern to the SNP, but it seems the lack of SNP's do not impact the cM count in the manner I was expecting... My guess is that Genesis simply fills in the blanks between segments in the name of uniformity, in a somewhat similar way to 23andme's smoothing algorithm.

My 23andme v3 and my LivingDNA kit matching on my grandmothers 23andme v3 6th chromosome (notice the cM and SNP count):
http://i64.tinypic.com/4r9jbl.jpg

Evon I think I did match your grandmother.. wonder if that is still a fact.. M103328

Kathlingram
07-23-2017, 12:37 PM
I am not sure how seriously to take a low snp match. Which is deemed more relevant a high cM low snp match or low cM high snp match? The fact that quite a few of us are listed as matches with each other seems a little suspicious to me. I suspect that the % of matches between members here is higher than could reasonably be expected.

I think they have a lot of work to do before they begin cousin matches.. That may not come easily especially since they indicated this to me when I queried the Genesis match :"We have collaborated with GEDmatch who are a third party tool so that they can accept our Living DNA raw data, unfortunately we have not been directly involved in their analysis developments.

If you have any questions about uploading your raw data, what GEDmatch can do for you, or what your GEDmatch results mean, please contact GEDmatch support/administrators. You can contact them at [email protected]"

Kathlingram
07-24-2017, 03:40 PM
from Living_DNA :"As our Family Ancestry Matching product is still be developed we are yet to release information about how the matching will work.
The best way to follow the development and progress of the matching functionality and how it will work will be to check our blog and social media streams, as and when we are able to release information and updates about the service and functionality it will be released on these platforms."

firemonkey
07-24-2017, 04:06 PM
In https://www.livingdna.com/en-gb/blog/274/new-feature-downloading-your-raw-data it talks about helping to continue research on other platforms. I can only really think of Gedmatch genesis so far(although I did upload to geneplaza) and that seems buggy.

Amerijoe
07-24-2017, 05:37 PM
Though I don't believe re-uploading will work, since the problem remains "too few SNPs" in the txt-file. Good luck though.

Tried re-up several times on Apple and Wintel platforms with close to identical results. Here is what I get time after time at Gedmatch.

The file indicates error free with the exception of a notice.

17727

17726

Here are the number of snps used for the one to one comparison.

23andMe to Livingdna 96344 snps
FTDNA. to Livingdna 159248 snps
23andMe to FTDNA 298444 snps

At Gedmatch Genesis performed the Validity, one to one comparison.

23andMe to Livingdna 107774 snps
FTDNA. to Livingdna 163040 snps
23andMe to FTDNA. 298858 snps

evon
07-24-2017, 08:56 PM
Evon I think I did match your grandmother.. wonder if that is still a fact.. M103328

You do match her 23andme V3 file with your LivingDNA file, but the matching segments at 5-6cM have very few SNP's.


Largest segment = 6.1 cM
Total of all matching segments > 5 cM = 17.1 cM (0.5 Pct)

176084 SNPs used for this comparison.

Pylsteen
08-01-2017, 11:57 AM
Mmm. I found my first "Asian" match, a Philippine matches me, only 5,5 cM, but it is on one my "Asian-heavy" chromosomes, so I tend to think it is real (I am aware it can also be "ancient", anyway, it is Asian).