PDA

View Full Version : Gedmatch Admixture (heritage) Utilities



Björnsson
11-01-2019, 03:15 PM
Which are your favourites and why?

Mine is MDLP Project, Admixture Proportions (With link to Oracle) and K23b. I don't bother with pie chart and percentages, because distance plotting reveals my nearest ethnicities within 20 spaces and in this case, matches my family tree. Also, I wish Davidski would link all of Eurogenes to Oracle, especially K36, but he's busy trying to sell another product. Hats off to the apps that include Utah European and Utah White, not just the ones profiling Latin America. Makes one wonder where Canada, Australia, New Zealand and South Africa or Rhodesia are.

mildlycurly
11-03-2019, 12:37 AM
It's difficult to say, especially as they're quite outdated.

However, I tend to prefer the Eurogenes calcs.

Björnsson
11-05-2019, 05:09 PM
MDLP K23b Oracle

The only app matching my North Sea family tree, which includes Carolingian nations during the colonial period and those of Knut the Great further back in the Viking age. First seven places are more or less the same.

LDNA

Single Population Sharing:

# Population (source) Distance
1 English_Kent_GBR ( ) 2.85
2 Frisian ( ) 3.37
3 Belgian ( ) 3.49
4 Irish ( ) 3.58
5 English ( ) 3.71
6 English_Cornwall_GBR ( ) 4
7 North_European ( ) 4.75
8 CEU ( ) 5.29
9 Welsh ( ) 5.31
10 British ( ) 5.39
11 French ( ) 6.93
12 Dutch ( ) 7.12
13 Norwegian_East ( ) 7.38
14 South_German ( ) 7.78
15 Norwegian_West ( ) 8.07
16 Scottish_Argyll_Bute_GBR ( ) 8.71
17 Swede ( ) 8.75
18 German-Volga ( ) 8.96
19 North_German ( ) 9.02
20 Dane ( ) 9.24

ADNA

Single Population Sharing:

# Population (source) Distance
1 Frisian ( ) 2.6
2 Belgian ( ) 3.31
3 Irish ( ) 3.5
4 English ( ) 4.38
5 English_Kent_GBR ( ) 4.49
6 English_Cornwall_GBR ( ) 5.52
7 North_European ( ) 5.77
8 Dutch ( ) 6.37
9 CEU ( ) 6.74
10 South_German ( ) 6.79
11 Welsh ( ) 6.83
12 British ( ) 6.89
13 German-Volga ( ) 7.82
14 Norwegian_East ( ) 7.86
15 French ( ) 8.26
16 North_German ( ) 8.36
17 Swede ( ) 8.86
18 Dane ( ) 9.05
19 Norwegian_West ( ) 9.19
20 Scottish_Argyll_Bute_GBR ( ) 9.95

Finn
11-05-2019, 07:08 PM
MDLP23b is quite good especially in my fathers's case (but South German before Frisian?) in my mother's case it's more off road although when you take the middle between East-Norway and Belgium you are in here case quite in the middle of the road ;) But it's not 100% bull's eye....

Father
Using 1 population approximation:
1 Dutch_ @ 2.785799
2 North_German_ @ 4.288587
3 South_German_ @ 4.311460
4 Frisian_ @ 5.029346
5 Dane_ @ 5.336345
6 Belgian_ @ 5.389698
7 Norwegian_East_ @ 5.747798
8 Irish_ @ 6.072299
9 English_ @ 6.331868
10 Swede_ @ 6.521797

Using 2 populations approximation:
1 50% Frisian_ +50% North_German_ @ 2.060322

Using 3 populations approximation:
1 50% Frisian_ +25% Irish_ +25% Slovak_ @ 1.366006

Mother
Using 1 population approximation:
1 Belgian_ @ 4.037003
2 Norwegian_East_ @ 4.329999
3 Dutch_ @ 4.725513
4 North_European_ @ 4.785549
5 Frisian_ @ 4.812660
6 South_German_ @ 5.059114
7 English_ @ 5.313992
8 North_German_ @ 5.526011
9 Norwegian_West_ @ 5.578958
10 Dane_ @ 5.653434

Using 2 populations approximation:
1 50% Belgian_ +50% Norwegian_East_ @ 2.467062

Using 3 populations approximation:
1 50% Belarusian-East_ +25% Frisian_ +25% Spanish_Pais_Vasco_IBS_ @ 1.837897

Björnsson
11-05-2019, 07:38 PM
I've noticed that there is a difference in single populations between Oracle (more accurate) and Oracle+4 (a little off).

Björnsson
11-05-2019, 08:34 PM
It's difficult to say, especially as they're quite outdated.

However, I tend to prefer the Eurogenes calcs.

What about Eurogenes do you like more than the others? Aren't you bothered by the lack of Oracle and Oracle+4 in most of Eurogenes? Also, what is so much better in Davidski's new G25 set-up?

msmarjoribanks
11-05-2019, 11:13 PM
I tend to like Eurogenes, but agree it's outdated at this point.


MDLP23b is quite good especially in my fathers's case (but South German before Frisian?) in my mother's case it's more off road although when you take the middle between East-Norway and Belgium you are in here case quite in the middle of the road ;) But it's not 100% bull's eye....

I apparently average as Frisian 1.91 there.

Then:

2 Belgian ( ) 2.5
3 Irish ( ) 3.36
4 Dutch ( ) 3.92
5 English ( ) 4.33
6 South_German ( ) 5.19
7 North_German ( ) 5.9

What I don't really like about it is that I have such a low number to start that in order to get lower it takes my little bit of Native American (which I assume is probably noise) and gives me Frisian (98.7) + 1.3% of various indigenous Latin American populations for 1.36, which is just off.

Björnsson
11-06-2019, 12:12 AM
I tend to like Eurogenes, but agree it's outdated at this point.



I apparently average as Frisian 1.91 there.

Then:

2 Belgian ( ) 2.5
3 Irish ( ) 3.36
4 Dutch ( ) 3.92
5 English ( ) 4.33
6 South_German ( ) 5.19
7 North_German ( ) 5.9

What I don't really like about it is that I have such a low number to start that in order to get lower it takes my little bit of Native American (which I assume is probably noise) and gives me Frisian (98.7) + 1.3% of various indigenous Latin American populations for 1.36, which is just off.

I don't follow the mixed population percentages, just the top 20 singular ones and what you have is very similar, from the looks of it. If you like Eurogenes, is K36 your favourite, or is that another?

Jessie
11-06-2019, 03:05 AM
This is mine. I prefer G25 over Gedmatch these days but for people starting off they are quite interesting.

Some oddities in this is the difference between East Norwegian and West Norwegian. Shouldn't they be closer together? It is also odd for me as an Irish person to have Orcadian so low compared to some other populations. Doesn't really make sense.

Single Population Sharing:

# Population (source) Distance
1 Irish ( ) 1.62
2 English ( ) 1.85
3 North_European ( ) 2.94
4 Frisian ( ) 3.77
5 Belgian ( ) 4.02
6 Norwegian_East ( ) 4.09
7 English_Kent_GBR ( ) 4.72
8 Dutch ( ) 4.75
9 CEU ( ) 4.83
10 English_Cornwall_GBR ( ) 4.84
11 Welsh ( ) 5.6
12 British ( ) 5.62
13 Swede ( ) 5.75
14 Norwegian_West ( ) 6.07
15 Dane ( ) 6.15
16 North_German ( ) 6.76
17 Icelandic ( ) 6.77
18 Scottish_Argyll_Bute_GBR ( ) 7.13
19 South_German ( ) 7.67
20 Orcadian ( ) 8.63

Greekscholar
11-06-2019, 03:30 AM
I prefer MDLP because it normally has fewer basic components and a better set of references for their Oracle. I like Eurogenes K15 and K36 for creating PCAs/mapping though. Their raw estimates have too many moving parts for me to make much sense out of. Their Oracles on GEDmatch don't have enough reference samples, IMO. It is hard to make sense out of them without the mapping/plotting tools.

I do prefer G25 these days, but still visit GEDmatch daily to check and see if I have any new matches, etc.

msmarjoribanks
11-06-2019, 02:37 PM
I don't follow the mixed population percentages, just the top 20 singular ones and what you have is very similar, from the looks of it. If you like Eurogenes, is K36 your favourite, or is that another?

I don't like K36 because of the lack of a spreadsheet. I like K13 and K15, because of the components chosen and the spreadsheets. I do have to catch up and use G25, though.

My results for K13:

1 North_German 3.52
2 West_German 5.03
3 South_Dutch 5.27
4 Danish 5.3
5 North_Dutch 5.63
6 Southeast_English 5.97
7 Orcadian 6.57
8 Norwegian 7.38
9 Irish 7.4
10 Swedish 7.61

Top mixed mode: #
1 60.5% North_German + 39.5% West_German @ 1.66

(Reality is a mix of English, Welsh, maybe some Irish, German, Swedish.)

Results for K15

# Population (source) Distance
1 North_German 2.62
2 Danish 5
3 South_Dutch 5.05
4 North_Dutch 5.42
5 Southeast_English 5.47
6 Southwest_English 5.82
7 Irish 6.06
8 West_Scottish 6.66
9 West_German 6.95
10 Norwegian 7.96

Top mixed:
1 61.4% Orcadian + 38.6% Austrian @ 1.89

Björnsson
11-07-2019, 02:27 AM
K13 is consistent except for the flip between fourth and fifth groups. K15 is only good for the first two groups. I don't know why anyone thinks it's better to have their top proximity single population at 3.0 or further off, when K23b gives about 2.75. MDLP doesn't waste my time with Spanish and Hungarian, because I haven't had those ancestors since the 14th century through descent from the royal families of Europe, specifically the Plantagenets (Edward IV), although I do have a closer Stewart line (James IV) and that may be why Scottish and Scandinavian are just above the Eurogenes groups I just mentioned were further back. Furthermore, what's the point of Oracle+4, in chasing crazy admix configurations between two, three, or four populations that don't fit?

K13

LDNA

Single Population Sharing:

# Population (source) Distance
1 Southeast_English 3.09
2 South_Dutch 4.44
3 Southwest_English 5.15
4 Orcadian 5.3
5 West_German 5.66
6 North_Dutch 6.17
7 Danish 6.35
8 Irish 6.58
9 West_Scottish 6.64
10 North_German 6.83
11 Norwegian 8.51
12 French 9.19
13 Swedish 10.02
14 Austrian 12.13
15 East_German 12.95
16 North_Swedish 15.15
17 Spanish_Cataluna 16.24
18 Hungarian 16.67
19 Spanish_Castilla_Y_Leon 17.41
20 Spanish_Galicia 17.67

ADNA

Single Population Sharing:

# Population (source) Distance
1 Southeast_English 3.23
2 South_Dutch 4.07
3 Southwest_English 5.17
4 West_German 5.24
5 Orcadian 5.52
6 North_Dutch 6.36
7 Danish 6.6
8 Irish 6.71
9 West_Scottish 6.81
10 North_German 6.93
11 Norwegian 8.78
12 French 8.79
13 Swedish 10.29
14 Austrian 12.03
15 East_German 12.96
16 North_Swedish 15.43
17 Spanish_Cataluna 15.88
18 Hungarian 16.58
19 Spanish_Castilla_Y_Leon 17.05
20 Spanish_Galicia 17.36

K15

LDNA

Single Population Sharing:

# Population (source) Distance
1 Southwest_English 3.1
2 Southeast_English 3.99
3 South_Dutch 5.33
4 Danish 5.73
5 North_Dutch 5.82
6 West_Scottish 5.92
7 Irish 5.93
8 Orcadian 6.31
9 West_German 6.37
10 North_German 6.75
11 Norwegian 8.1
12 French 8.88
13 West_Norwegian 9.06
14 Swedish 9.69
15 North_Swedish 11.74
16 East_German 12.16
17 Southwest_Finnish 16.11
18 Austrian 16.21
19 Spanish_Galicia 16.34
20 Spanish_Cataluna 16.43

ADNA

Single Population Sharing:

# Population (source) Distance
1 Southwest_English 3.29
2 Southeast_English 4.34
3 North_Dutch 5.75
4 Orcadian 5.86
5 Danish 5.91
6 South_Dutch 5.99
7 West_German 6.05
8 West_Scottish 6.09
9 Irish 6.24
10 North_German 7.19
11 Norwegian 7.83
12 West_Norwegian 8.57
13 French 9.23
14 Swedish 9.54
15 North_Swedish 11.83
16 East_German 12.73
17 Spanish_Galicia 16.69
18 Southwest_Finnish 16.69
19 Austrian 16.96
20 Spanish_Cataluna 16.98

Björnsson
11-07-2019, 03:06 AM
This is mine. I prefer G25 over Gedmatch these days but for people starting off they are quite interesting.

Some oddities in this is the difference between East Norwegian and West Norwegian. Shouldn't they be closer together? It is also odd for me as an Irish person to have Orcadian so low compared to some other populations. Doesn't really make sense.

Single Population Sharing:

# Population (source) Distance
1 Irish ( ) 1.62
2 English ( ) 1.85
3 North_European ( ) 2.94
4 Frisian ( ) 3.77
5 Belgian ( ) 4.02
6 Norwegian_East ( ) 4.09
7 English_Kent_GBR ( ) 4.72
8 Dutch ( ) 4.75
9 CEU ( ) 4.83
10 English_Cornwall_GBR ( ) 4.84
11 Welsh ( ) 5.6
12 British ( ) 5.62
13 Swede ( ) 5.75
14 Norwegian_West ( ) 6.07
15 Dane ( ) 6.15
16 North_German ( ) 6.76
17 Icelandic ( ) 6.77
18 Scottish_Argyll_Bute_GBR ( ) 7.13
19 South_German ( ) 7.67
20 Orcadian ( ) 8.63I have no Orcadian, low Scottish and high Irish on K23b, which reflects my family tree. East and West Norwegians are real groups. East Norse are closer to other Scandinavians, while West Norse are closer to the British Isles, hence Orcadian (really should be Faroese) and Icelandic.

Björnsson
11-07-2019, 03:08 AM
I prefer MDLP because it normally has fewer basic components and a better set of references for their Oracle. I like Eurogenes K15 and K36 for creating PCAs/mapping though. Their raw estimates have too many moving parts for me to make much sense out of. Their Oracles on GEDmatch don't have enough reference samples, IMO. It is hard to make sense out of them without the mapping/plotting tools.

I do prefer G25 these days, but still visit GEDmatch daily to check and see if I have any new matches, etc.

If K36 had Oracle and Oracle+4, wouldn't that be better than G25? Isn't G25 only slightly better than K23b?

Björnsson
11-07-2019, 03:15 AM
I don't like K36 because of the lack of a spreadsheet. I like K13 and K15, because of the components chosen and the spreadsheets. I do have to catch up and use G25, though.

My results for K13:

1 North_German 3.52
2 West_German 5.03
3 South_Dutch 5.27
4 Danish 5.3
5 North_Dutch 5.63
6 Southeast_English 5.97
7 Orcadian 6.57
8 Norwegian 7.38
9 Irish 7.4
10 Swedish 7.61

Top mixed mode: #
1 60.5% North_German + 39.5% West_German @ 1.66

(Reality is a mix of English, Welsh, maybe some Irish, German, Swedish.)

Results for K15

# Population (source) Distance
1 North_German 2.62
2 Danish 5
3 South_Dutch 5.05
4 North_Dutch 5.42
5 Southeast_English 5.47
6 Southwest_English 5.82
7 Irish 6.06
8 West_Scottish 6.66
9 West_German 6.95
10 Norwegian 7.96

Top mixed:
1 61.4% Orcadian + 38.6% Austrian @ 1.89

What about Jtest and EUtest?

Whoever picked puntDNAL, I'd like to know why. They're almost as close as MDLP. Dodecad, I'm also curious why.

Björnsson
11-07-2019, 12:35 PM
Since I keep seeing K16 in the forum, I post it here and now. This app takes away my Low Countries and tucks them away into France, but otherwise rightly gives mostly British Isles and a stray case of Germany. It seems that the first five approximations are more or less the same.

LDNA

Single Population Sharing:

# Population (source) Distance
1 French (France) 3.12
2 English (Kent) 3.77
3 French (WestFrance) 3.83
4 Scottish (Grampian) 4.08
5 Irish (Connacht) 4.2
6 English (Cornwall) 4.37
7 English (England) 4.49
8 Irish (Cork_Kerry) 4.5
9 Scottish (Highlands) 4.52
10 Irish (Leinster) 4.53
11 Irish (Ulster) 4.75
12 Scottish (Fife) 4.79
13 German (Germany) 4.88
14 Scottish (Dumfries_Galloway) 4.89
15 Shetlandic (Shetland_Islands) 5.14
16 Irish (Munster) 5.2
17 Scottish (Borders) 5.23
18 French (NorthwestFrance) 5.65
19 French (EastFrance) 6.06
20 Welsh (Wales) 6.3

ADNA

Single Population Sharing:

# Population (source) Distance
1 French (WestFrance) 3.19
2 Scottish (Grampian) 3.4
3 English (Kent) 3.43
4 French (France) 3.5
5 Irish (Connacht) 3.58
6 English (England) 3.68
7 Scottish (Highlands) 3.83
8 Irish (Leinster) 3.83
9 Irish (Cork_Kerry) 3.83
10 English (Cornwall) 3.87
11 Scottish (Fife) 3.99
12 Irish (Ulster) 4.06
13 Scottish (Dumfries_Galloway) 4.12
14 Shetlandic (Shetland_Islands) 4.38
15 Irish (Munster) 4.43
16 Scottish (Borders) 4.53
17 German (Germany) 4.65
18 French (NorthwestFrance) 5.35
19 Welsh (Wales) 5.73
20 French (EastFrance) 5.8

Björnsson
11-08-2019, 06:37 AM
LDNA makes me Norwegian and English, German and Frisian, Austrian and Belgian, French and Croatian, Albanian and Hungarian, all of which fit in Utah, lol. Top Oracle results per app:

MDLP K11 Modern: British_Celtic @ 4.159802 (Oracle+4)
MDLP K16 Modern: French (France) 3.12 [ADNA: French (WestFrance)]
MDLP K23b: English_Kent_GBR ( ) 2.85 [ADNA: Frisian]
MDLP World-22: CEU (derived) 4.57 [ADNA: German-South (derived)]
MDLP World: Austrian 2.5

Eurogenes K13: Southeast_English 3.09
Eurogenes EUtest V2 K15: Southwest_English 3.1
Jtest: English 3.74
EUtest: English 3.65

Dodecad V3: CEU (HapMap) 2.89
Africa9: North_Italian 4.86
World9: Mixed_Germanic 2.79
Dodecad K7b: Mixed_Germanic 3.33
Dodecad K12b: Mixed_Germanic (Dodecad) 4.42

HarappaWorld: utahn-white (1000genomes) 4.43 [ADNA: french]

EthioHelix K10 + French: french 25.24
EthioHelix K10 + Japanese: egypt 6.33
EthioHelix K10 + Palestinian: morocco-n 6.81
EthioHelix K10 Africa Only: algeria 8.46

puntDNAL K10 Ancient: German_South 2.8 [ADNA: Croatian]
puntDNAL K12 Ancient: Alberstedt_LN_I0118 5.7
puntDNAL K12 Modern: German_South 2.22 [ADNA: Belgian]
puntDNAL K13 Global: German_South 3.72
puntDNAL K15: Utahn_White 2.94
puntDNAL K8 African only: Egyptian 4.31

Gedrosia K3: French @ 0.000000 (Oracle+4) [ADNA: Albanian]
Gedrosia K12: Norwegian 16.82
Ancient Eurasia K6: Hungarian 1.81

Björnsson
11-08-2019, 06:50 AM
For Gedrosia K3, my wife has Bulgarian from LDNA and Georgian from ADNA. Same Norwegian as me. Ancient Eurasia K6, LDNA gives her English and ADNA gives her Czech.

I'm not finding GedrosiaDNA all that reliable or consistent, while EthioHelix is just plain irrelevant. The rest seem fine.

As expected, most of my DNA is from Anglo-Saxons and Franks, in the time of Offa and Charlemagne. That is echoed in my colonial period infusions from the Continent, but I was surprised to see Croatian and it's likely the source of my Aegean. I remembered that it was one of the regions paying tribute to the Holy Roman Empire, even before Styria and Carniola were assimilated by the Bayerische Ostmark, becoming the Babenberger Herzogtum. That's the extent of my most exotic DNA, unless my Pashtun is Anglo-Indian.

Going by the same parameters of top Oracle results, my wife is just British Isles and Germany, which makes sense, except for negligible Iberian from Portugal or Madeira. I think that this canvassing and polling has finally completely matched each other's family trees, without really missing or adding anything.

Tanchik
07-19-2020, 12:00 PM
Dodecad World 9 the most