PDA

View Full Version : Why do Gedmatch results differ for an individual that has tested with dif



Almagest
10-26-2016, 07:57 PM
Why do Gedmatch results differ for an individual that has tested with different commercial DNA testing companies? Which company should I do autosomal DNA test with to get the most accurate Gedmatch results?

sktibo
10-26-2016, 08:19 PM
I think it's because of different population samples / references. I think the best thing about gedmatch is that you can compare numbers/stats with others of a certain population... And tons of people have posted their GEDmatch numbers here on anthrogenica.
I read somewhere that FTDNA is best for gedmatch.. but MyOrigins is less than fantastic. Please get more opinions on this before deciding.

MitchellSince1893
10-27-2016, 12:10 PM
From my own experience I don't see much of a difference. Here's what my father gets on one of the more popular gedmatch tools Eurogenes K13. No more than a .27% difference

23andme
# Population Percent
1 North_Atlantic 52.25
2 Baltic 21.91
3 West_Med 10.46
4 East_Med 5.39
5 West_Asian 5.2
6 South_Asian 1.76
7 Red_Sea 1.27
8 Amerindian 0.75
9 Siberian 0.39
10 Oceanian 0.36
11 Sub-Saharan 0.26

Ancestry.com
# Population Percent
1 North_Atlantic 52.06
2 Baltic 21.95
3 West_Med 10.68
4 West_Asian 5.25
5 East_Med 5.25
6 South_Asian 1.83
7 Red_Sea 1.18
8 Amerindian 0.75
9 Oceanian 0.36
10 Sub-Saharan 0.36
11 Siberian 0.33

FTDNA
# Population Percent
1 North_Atlantic 52.33
2 Baltic 21.68
3 West_Med 10.55
4 East_Med 5.39
5 West_Asian 5.33
6 South_Asian 1.75
7 Red_Sea 1.18
8 Amerindian 0.78
9 Siberian 0.39
10 Oceanian 0.36
11 Sub-Saharan 0.27

dp
10-28-2016, 07:59 PM
I believe @Kurd explains it as testing companies are trying (sometimes with imputation) to identify segments that belong to a particular ethnicity. Admixture looks for the presence of ancestry specific alleles throughout your genome. @Kurd also has mentioned that the estimate of the testing companies can be low compared to what you may harbor.
I have noticed that testing companies will try to find the nearest population to your main hotspot that can explain any deviation.

@Kurd customizes his GEDmatch admix (under GedrosiaDNA) routines to work for 23andme.com
@Vadim is the creator of those under MDLP.
Apologies to those I can't think of your handles.
dp :-)

Missouri1455
12-05-2016, 02:17 PM
Why do Gedmatch results differ for an individual that has tested with different commercial DNA testing companies? Which company should I do autosomal DNA test with to get the most accurate Gedmatch results?

It differs, because gedmatch goes way deeper into your ethnicity and breaks it down; for example in my case, in genographic 2.0 I scored 9% jewish diaspora and 9% North africa, but when i used ged match it really broke down for me what all of this meant; i basically came to a conclusion that i do have an ancestor who was either Iranian Jew, Iraqi Jew, or Samaritan; with also iranian shirazi along the bloodline not sure where they fall; and the north african(Maghreb) was mostly Libyan because Libyans were usually clustered as being one of my closest populations beside yemenite, egyptian, jordanian, and bedouin.

The most accurate test i found for my results was Deocad V3; i think it broke down my ethnicity specifically my only issue with was it included a term "Palaeo African" which is considered more ancient and clustered it together with more modern ethnicity, when in reality "Palaeo african" does not even fall under the same category as East African and West African; and this was shown in a graph by the group that made it.

A Norfolk L-M20
12-05-2016, 02:34 PM
Different number of SNPs tested, and different range of SNPs tested by the different chips used by different companies. Yes, I do sometimes get quite different results on GEDmatch calculators and other analysis sites, depending in whether I use my 23andme raw file, or my FTDNA raw file. The ftDNA file measures more SNPs, but is it better?

Missouri1455
12-05-2016, 02:40 PM
Different number of SNPs tested, and different range of SNPs tested by the different chips used by different companies. Yes, I do sometimes get quite different results on GEDmatch calculators and other analysis sites, depending in whether I use my 23andme raw file, or my FTDNA raw file. The ftDNA file measures more SNPs, but is it better?

how much does it cost to do FTDNA? and how much more snps is used for it on gedmatch?

kikkk
12-05-2016, 03:39 PM
how much does it cost to do FTDNA? and how much more snps is used for it on gedmatch?

FTDNA costs 59 $ now and 704636 snps are used on gedmatch from my FTDNA kit.

Missouri1455
12-05-2016, 03:57 PM
FTDNA costs 59 $ now and 704636 snps are used on gedmatch from my FTDNA kit.

Thats nice, How different was your result in comparison to geno 2.0 and 23andme? and how far does it go back?

A Norfolk L-M20
12-05-2016, 03:58 PM
You'd have to check the ftDNA website yourself in your country to check price, but their autosomal DNA family finder package is still reduced to $59 USD. But I am not qualified to say more SNPs are always better. I believe that 23andme tested more SNPs on their v3 chip then on their v4 chip? It must also depend on which are tested. However, for comparison:

http://isogg.org/wiki/Autosomal_DNA_testing_comparison_chart

The new chip used by Living DNA I understand tests even more SNPs. All of this though, should serve to warn consumers, how unreliable autosomal DNA testing for Ancestry still is. Test on a different chip, different company, different calculator, phase with parents, different sibling - all change your results, but your real ancestry does not change.

kikkk
12-05-2016, 04:01 PM
Thats nice, How different was your result in comparison to geno 2.0 and 23andme? and how far does it go back?

I have tested solely with FTDNA.

Missouri1455
12-05-2016, 04:09 PM
I have tested solely with FTDNA.

Interesting, how many years it goes back? 400-500 years?

Also do they have accurate haplogroup system?

kikkk
12-05-2016, 04:33 PM
Interesting, how many years it goes back? 400-500 years?

Also do they have accurate haplogroup system?

TBH I have no idea about how many years they go back.

Ignis90
12-05-2016, 04:40 PM
Interesting, how many years it goes back? 400-500 years?

Also do they have accurate haplogroup system?

They don't go back anywhere in time: it only depends on their sampling strategy and how drifted their reference populations are. Genetic testing companies are oversimplifying things for their clueless customers to the point of stating straight up lies/misinformation.

Missouri1455
12-05-2016, 05:50 PM
They don't go back anywhere in time: it only depends on their sampling strategy and how drifted their reference populations are. Genetic testing companies are oversimplifying things for their clueless customers to the point of stating straight up lies/misinformation.

You're right, even though im new to this, i have been getting the hang of it and im starting to realize lies/misinformation that are put out there and how easy it is to fool clueless customers who don't know much about genetic testing.

Ignis90
12-05-2016, 06:00 PM
You're right, even though im new to this, i have been getting the hang of it and im starting to realize lies/misinformation that are put out there and how easy it is to fool clueless customers who don't know much about genetic testing.

Not singling you out as a "clueless customer" though! :biggrin1:
But yes, I think simplifying things to an extent is necessary since over 99% of the customer population does not have any background or training in basic human population genetics. But the "500 years back in time" thing is too much and misleading.

What's great about a forum like this one is that everybody can learn from each other and eventually get a less inaccurate picture of one's ancestry.

A Norfolk L-M20
12-05-2016, 06:07 PM
The "how much time they go back" claims made by different DNA companies? I am very skeptical. They all appear to struggle to tell ancient population background admixture from recent (300 year) family admixture, regardless of their claims.

Here's some boring comparison data between my 23andMe and FT-DNA raw files:

My 23andMe AC in spec mode (after phasing with one parent) is:
100% European
96% NW European
2% South European
2% broadly European

37% British & Irish
22% French & German
1% Scandinavian
36% broadly NW European
2% broadly South European

FTDNA Family Finder - My Origins
36% British Isles
32% Southern European
26% Scandinavian
6% Eastern European

I thought that it would be interesting to compare how a few important GEDMATCH calculators, see my raw data from Family Tree DNA, in comparison to the raw data from 23andMe:

GEDMATCH

23andMe raw data V ftDNA raw data

Eugenes K13 Oracle

23andMe data
Admix Results (sorted):
# Population Percent
1 North_Atlantic 47.58
2 Baltic 22.36
3 West_Med 15.65
4 East_Med 8.03
5 West_Asian 3.05
6 Red_Sea 1.42
7 Amerindian 0.74
8 South_Asian 0.71
9 Oceanian 0.46
Single population Sharing:
# Population (source) Distance
1 South_Dutch 3.89
2 Southeast_English 4.35
3 West_German 5.22
4 Southwest_English 6.24
5 Orcadian 6.97
6 French 7.63
7 North_Dutch 7.76
8 Danish 7.95
9 North_German 8.17
10 Irish 8.22

Family Tree DNA data
Admix Results (sorted):
# Population Percent
1 North_Atlantic 47.89
2 Baltic 22.68
3 West_Med 15.45
4 East_Med 7.41
5 West_Asian 3.11
6 Red_Sea 1.38
7 South_Asian 0.84
8 Amerindian 0.72
9 Oceanian 0.52
Single Population Sharing:
# Population (source) Distance
1 Southeast_English 3.75
2 South_Dutch 4.03
3 West_German 5.42
4 Southwest_English 5.68
5 Orcadian 6.33
6 North_Dutch 7.15
7 Danish 7.36
8 Irish 7.59
9 West_Scottish 7.62
10 North_German 7.7

Euogenes EUtest V2 K15

23andMe data
Admix Results (sorted):
# Population Percent
1 North_Sea 33.42
2 Atlantic 27.98
3 West_Med 12.24
4 Baltic 10.42
5 Eastern_Euro 7.04
6 West_Asian 3.52
7 East_Med 3.14
8 Red_Sea 1.48
9 Amerindian 0.39
10 Oceanian 0.19
11 South_Asian 0.18
Single Population Sharing:
# Population (source) Distance
1 Southwest_English 2.7
2 South_Dutch 3.98
3 Southeast_English 4.33
4 Irish 6.23
5 West_German 6.25
6 North_Dutch 6.79
7 West_Scottish 6.84
8 French 6.85
9 North_German 6.89
10 Danish 7.26

Family Tree DNA data
Admix Results (sorted):
# Population Percent
1 North_Sea 33.81
2 Atlantic 28.23
3 West_Med 12.04
4 Baltic 10.59
5 Eastern_Euro 6.84
6 West_Asian 3.66
7 East_Med 2.47
8 Red_Sea 1.46
9 Amerindian 0.35
10 South_Asian 0.31
11 Oceanian 0.25
Single Population Sharing:
# Population (source) Distance
1 Southwest_English 2.29
2 Southeast_English 4.02
3 South_Dutch 4.48
4 Irish 5.78
5 West_Scottish 6.41
6 North_Dutch 6.43
7 West_German 6.63
8 North_German 6.73
9 Danish 7.01
10 Orcadian 7.19

Gedrosia Eurasia K6 Oracle

23andMe data
Admix Results (sorted):
# Population Percent
1 West_European_Hunter_Gartherer 39.18
2 Natufian 38.8
3 Ancestral_North_Eurasian 20.85
4 Ancestral_South_Eurasian 0.82
5 East_Asian 0.35

Family Tree DNA data
Admix Results (sorted):
# Population Percent
1 West_European_Hunter_Gartherer 39.57
2 Natufian 38.66
3 Ancestral_North_Eurasian 20.75
4 East_Asian 0.86
5 Ancestral_South_Eurasian 0.16

23andMe V4 raw file for myself on DNA.land:

100% West Eurasian.
77% North West European
19% South European (broken into 13% Balkan / 6.1% South/Central European
2.4% Finnish
1.3% Ambiguous

FT-DNA FF raw file for myself on DNA.land:
100% West Eurasian
75% North West European
25% Balkan

Conclusion

Not a lot of difference - however, there are some shifts in some places. The FT-DNA file might have tested more SNPs, but for some reason, it doesn't produce as much admixture on DNA.land as the 23andMe file. Not that any of it is likely to be true! The FT-DNA gets it right on Eurogenes K13 , with SE English as my closest Single Population (I am SE English), whereas my 23andMe file in K13 sees South Dutch as closer. Am I guilty though, of positive sampling - picking a test and a file, that just happen to match my family history and paper trail?

But not a lot of difference. The reduced price FT-DNA Family Finder Package does give a good value autosomal DNA raw file, and also a matching service. In my opinion though, it's My Origins ancestry analysis is very poor. Also, there is no haplogroup testing.

Mixed
12-05-2016, 06:24 PM
I've experienced the same shifts as well.

Oleg (Rus)
12-05-2016, 06:43 PM
Different companies test different sets of SNPs, so Gedmatch uses different SNPs in its calculations. It's the reason of different results.

Oleg (Rus)
12-05-2016, 07:58 PM
Interesting, how many years it goes back? 400-500 years?

Also do they have accurate haplogroup system?

I don't think that it directly goes back to 500, 1000 or 2000 years, for example. It's rather a complex vision, and each component may have its own age.