Mega QAPDM Models thread

Printable View

Show 40 post(s) from this thread on one page

07-13-2025, 08:29 PM
Opie

Mega QAPDM Models thread

Now that IllustrativeDNA introduced QAPDM service, I thought starting a thread where everyone shares QAPDM models would be resourceful.

As a start, I am sharing this model:

Sources

Turkey_Marmara_Barcin_N.AG (Anatolian Neolithic Farmer)

Russia_Sidelkino_HG.SG (Eastern European Hunter-Gatherer)

Luxembourg_Mesolithic.AG (Western European Hunter-Gatherer)

Georgia_Kotias_Mesolithic.SG (Caucasus Hunter-Gatherer)

Iran_GanjDareh_N.AG (Neolithic Iranian Farmer)

References

Mbuti.DG

Russia_AfontovaGora3_UP.AG

Italy_Epigravettian.SG

Russia_Kostenki14_UP.SG

Turkey_Central_Pinarbasi_Epipaleolithic.AG

Turkey_Central_Boncuklu_PPN.AG

Russia_Tyumen_HG.AG

Israel_Natufian.AG

Morocco_Iberomaurusian.AG

Georgia_Satsurblia_LateUP.SG

Iran_Wezmeh_N.SG

India_GreatAndaman_100BP.SG
07-13-2025, 09:00 PM
AndreiDNA

Remove kostenki and Andaman from your rights. They're irrelevant to the model and will lead to worse pval
07-13-2025, 09:56 PM
majevica

https://i.postimg.cc/JhTKrp1D/Screenshot-447.png

Works I would say
07-13-2025, 10:38 PM
Gannicus

1 Attachment(s)

Attachment 141692

Z-score would be 1.6ish and I looked up where there have been published studies where z-scores like that are acceptable. P-value is well above 0.05
07-14-2025, 12:25 AM
djipon

I seem to get red p-value and feasibility

https://i.postimg.cc/Gt6XjbGw/image.png
07-14-2025, 12:44 AM
Gannicus

1 Attachment(s)

Quote:

Originally Posted by Opie

Now that IllustrativeDNA introduced QAPDM service, I thought starting a thread where everyone shares QAPDM models would be resourceful.

As a start, I am sharing this model:

Sources

Turkey_Marmara_Barcin_N.AG (Anatolian Neolithic Farmer)

Russia_Sidelkino_HG.SG (Eastern European Hunter-Gatherer)

Luxembourg_Mesolithic.AG (Western European Hunter-Gatherer)

Georgia_Kotias_Mesolithic.SG (Caucasus Hunter-Gatherer)

Iran_GanjDareh_N.AG (Neolithic Iranian Farmer)

References

Mbuti.DG

Russia_AfontovaGora3_UP.AG

Italy_Epigravettian.SG

Russia_Kostenki14_UP.SG

Turkey_Central_Pinarbasi_Epipaleolithic.AG

Turkey_Central_Boncuklu_PPN.AG

Russia_Tyumen_HG.AG

Israel_Natufian.AG

Morocco_Iberomaurusian.AG

Georgia_Satsurblia_LateUP.SG

Iran_Wezmeh_N.SG

India_GreatAndaman_100BP.SG

I did this with your sources and references except India_GreatAndaman_100BP.SG and
Russia_Kostenki14_UP.SG

Attachment 141698
07-14-2025, 01:15 PM
Gannicus

2 Attachment(s)

Testing the near east spice I see in G25 and Eurogenes. I have had AndreiDNA run this 2-way combination. I’m testing it myself with I believe a slightly different set of right populations. I also still cannot use my merged file, so I’m using my DNA file from MyHeritage.com. One is allowed two free kit uploads. It’s visible in the list, however, I cannot select it. I used the Human Origins Dataset for this. Attachment 141699 Attachment 141700
07-14-2025, 03:51 PM
Feiichy

I've been trying some medieval models but they keep getting rejected. Any tips how to improve?
07-14-2025, 06:38 PM
majevica

Quote:

Originally Posted by Feiichy

I've been trying some medieval models but they keep getting rejected. Any tips how to improve?

Try Sunghir + Shtike
07-15-2025, 04:36 AM
Gannicus

7 Attachment(s)

Another model run with EGY_TIP_JK2134:

Sweden_IA_2 VK522:

Attachment 141727

Raw Output:

pat wt dof chisq p f4rank Sweden_IA_2.SG_VK522.SG Egypt_ThirdIntermediatePeriod.AG_JK2134.AG feasible best dofdiff chisqdiff p_nested
00 0 13 17.260705192920177 0.18765859345185965 1 0.8362559026573537 0.16374409734264636 TRUE NA NA NA NA
01 1 14 29.014137137307152 0.010404434189942862 0 1 NA TRUE TRUE 0 -43.42078257238162 1
10 1 14 72.43491970968877 6.968636070378395e-10 0 NA 1 TRUE TRUE NA NA NA

Denmark_IA VK532:

Attachment 141731

pat wt dof chisq p f4rank Denmark_IA.SG_VK532.SG Egypt_ThirdIntermediatePeriod.AG_JK2134.AG feasible best dofdiff chisqdiff p_nested
00 0 13 11.39641196621108 0.5776461449763756 1 0.8878444657143866 0.11215553428561346 TRUE NA NA NA NA
01 1 14 15.971954105271262 0.31509001573731876 0 1 NA TRUE TRUE 0 -56.16501842952157 1
10 1 14 72.13697253479283 7.896541490264353e-10 0 NA 1 TRUE TRUE NA NA NA

Scotland_MIA I2982:

Attachment 141732

pat wt dof chisq p f4rank Scotland_MIA.AG_I2982.AG Egypt_ThirdIntermediatePeriod.AG_JK2134.AG feasible best dofdiff chisqdiff p_nested
00 0 13 21.56134889984603 0.062542307243352 1 0.8882680436329078 0.11173195636709223 TRUE NA NA NA NA
01 1 14 25.82020873179249 0.027282545571266602 0 1 NA TRUE TRUE 0 -48.52234493063585 1
10 1 14 74.34255366242834 3.1231520993584695e-10 0 NA 1 TRUE TRUE NA NA NA

I had AndreiDNA model me with Old Kingdom Egyptian (NUE001) + Denmark IA and from there I copied the reference populations he used in that model run:

Attachment 141725

I had to substitute Russia_Karelia_HG to Russia_Samara_HG. For some reason I couldn't find it in IllustrativeDNA.
Also, I'm not sure If he did this, but I pooled together a number of reference populations into one. For example, I combined Taforalt 9-13 into the Iberomaurusian reference:

Attachment 141728

Full reference list:

Attachment 141729
Attachment 141730
07-15-2025, 05:00 AM
majevica

Quote:

Originally Posted by Gannicus

I had to substitute Russia_Karelia_HG to Russia_Samara_HG. For some reason I couldn't find it in IllustrativeDNA.

Karelia is Russia_YuzhniyOleniyOstrov_Mesolithic, the one I used in my model since Samara and Sidelkino already have CHG/Iran_N admix
07-15-2025, 05:49 AM
Gannicus

1 Attachment(s)

Quote:

Originally Posted by majevica

Karelia is Russia_YuzhniyOleniyOstrov_Mesolithic, the one I used in my model since Samara and Sidelkino already have CHG/Iran_N admix

Added that into the reference list and removed Samara. The adjustment tightened up the standard errors and chi-squared a bit on the first 2-way:

Attachment 141733

pat wt dof chisq p f4rank Sweden_IA_2.SG_VK522.SG Egypt_ThirdIntermediatePeriod.AG_JK2134.AG feasible best dofdiff chisqdiff p_nested
00 0 13 16.755364121649443 0.21073012036400646 1 0.8319910986340255 0.1680089013659744 TRUE NA NA NA NA
01 1 14 30.340556470389895 0.006848638940157966 0 1 NA TRUE TRUE 0 -49.18801990975905 1
10 1 14 79.52857638014895 3.46030652820137e-11 0 NA 1 TRUE TRUE NA NA NA
07-15-2025, 05:50 AM
Gannicus

1 Attachment(s)

Double post.
07-15-2025, 05:19 PM
drb234

https://i.postimg.cc/pTXwhxX7/fafafaff.webp
07-15-2025, 05:36 PM
Gannicus

2 Attachment(s)

This is one of my best runs yet with the same rights as the previous. I used Ireland EBA Rathlin 2, a sample that I match on MyTrueAncestry:Attachment 141739

Those that like to be too particular like to criticize results that produce Z-Scores less than 3. Here in this result the Z-score for Egypt TIP would be 3.25. I've already mentioned though that Z-Scores less than 3 make it into peer-reviewed published genetics papers.

Attachment 141738

pat wt dof chisq p f4rank Ireland_EBA.SG_rath2.SG Egypt_ThirdIntermediatePeriod.AG_JK2134.AG feasible best dofdiff chisqdiff p_nested
00 0 13 18.751626933192213 0.1310081143344856 1 0.7944860039558197 0.20551399604418025 TRUE NA NA NA NA
01 1 14 34.52252862395704 0.001728893048752092 0 1 NA TRUE TRUE 0 -43.68810958949282 1
10 1 14 78.21063821344985 6.067038308893523e-11 0 NA 1 TRUE TRUE NA NA NA
07-15-2025, 08:48 PM
Gannicus

3 Attachment(s)

Trying to isolate Natufian:

Attachment 141743

Chi-squared = 17.91
Degrees of freedom = 9

p-value = 0.0363

pat wt dof chisq p f4rank Russia_Samara_EBA_Yamnaya.AG_I0231.AG Turkey_Central_Pinarbasi_Epipaleolithic.AG_ZBC_IPB 001.B-C0101_Luk2-Pinarbasi.AG Israel_Natufian.AG Luxembourg_Mesolithic.DG_Loschbour.DG feasible best dofdiff chisqdiff p_nested
0000 0 9 17.907109322604683 0.03626665845967619 3 0.4791685725273366 0.3266312318943904 0.08478275718187421 0.10941743839639882 TRUE NA NA NA NA
0001 1 10 59.05707348625523 5.463172094567697e-9 2 0.5803429292627148 0.3827221868253448 0.03693488391194047 NA TRUE TRUE 0 30.604401993030546 0
0010 1 10 28.45267149322468 0.0015271326634493005 2 0.5147056167631247 0.38014783771804256 NA 0.10514654551883273 TRUE TRUE 0 -91.15029459533007 1
0100 1 10 119.60296608855475 6.091319306297955e-21 2 0.5504406348257856 NA 0.32676212869675986 0.12279723647745458 TRUE TRUE 0 -188.55068871897544 1
1000 1 10 308.1536548075302 2.9333895170265636e-60 2 NA 0.7963751322345017 -0.019424186124213403 0.22304905388971172 FALSE TRUE NA NA NA
0011 2 11 67.10922592767723 4.313562300130967e-10 1 0.5965554447330063 0.4034445552669938 NA NA TRUE NA NA NA NA
0101 2 11 197.44040773961413 2.5256739456125005e-36 1 0.6777461704018504 NA 0.32225382959814963 NA TRUE NA NA NA NA
0110 2 11 253.34435098119383 5.568806304414893e-48 1 0.9237112307300359 NA NA 0.07628876926996417 TRUE NA NA NA NA
1001 2 11 349.41902395534134 3.2170424679394866e-68 1 NA 3.3860029662268123 -2.3860029662268123 NA FALSE NA NA NA NA
1010 2 11 330.8205549892451 2.753010405050842e-64 1 NA 0.7897004380171972 NA 0.21029956198280272 TRUE NA NA NA NA
1100 2 11 394.30124856187723 9.914763102157746e-78 1 NA NA 0.7387038103550781 0.26129618964492185 TRUE NA NA NA NA
0111 3 12 285.7497316796714 4.5834266966857024e-54 0 1 NA NA NA TRUE NA NA NA NA
1011 3 12 452.4788427898409 2.810539351885685e-89 0 NA 1 NA NA TRUE NA NA NA NA
1101 3 12 524.2693494803447 1.5070567436541544e-104 0 NA NA 1 NA TRUE NA NA NA NA
1110 3 12 1286.8672327226395 3.3655118335110553e-268 0 NA NA NA 1 TRUE NA NA NA NA

References:

Attachment 141744
Attachment 141745

It's getting somewhat close to passing.

Any suggestions on what to change?
07-15-2025, 11:41 PM
AndreiDNA

1 Attachment(s)

Quote:

Originally Posted by majevica

Karelia is Russia_YuzhniyOleniyOstrov_Mesolithic, the one I used in my model since Samara and Sidelkino already have CHG/Iran_N admix

Karelia also has CHG admix. They even have 4 people with J1 from mesolithic karelia.
Attachment 141747
07-16-2025, 12:46 AM
Gannicus

3 Attachment(s)

So far I've only been able to infer Natufian admix by using populations that have it "baked" in already:

Attachment 141748

pat wt dof chisq p f4rank Latvia_LN_CordedWare.AG Israel_PPNB.AG_I0867.AG Luxembourg_Mesolithic.DG_Loschbour.DG feasible best dofdiff chisqdiff p_nested
000 0 11 14.201740559702879 0.22203070363619803 2 0.5093364971235644 0.3120461889516757 0.17861731392476 TRUE NA NA NA NA
001 1 12 55.526177292551345 1.4562896862833555e-7 1 0.684297592393732 0.315702407606268 NA TRUE TRUE 0 -11.215171823926532 1
010 1 12 66.74134911647788 1.2970190791969662e-9 1 0.8278127622817596 NA 0.1721872377182404 TRUE TRUE 0 -195.33431875347333 1
100 1 12 262.0756678699512 4.1262485394475325e-49 1 NA 0.7093245146422447 0.29067548535775534 TRUE TRUE NA NA NA
011 2 13 88.95895111964444 2.2095624661751116e-13 0 1 NA NA TRUE NA NA NA NA
101 2 13 441.57560449532315 3.602664999464381e-86 0 NA 1 NA TRUE NA NA NA NA
110 2 13 1092.793181336491 2.0157458922755243e-225 0 NA NA 1 TRUE NA NA NA NA

Rights:

Attachment 141749
Attachment 141750
07-16-2025, 01:03 AM
Gannicus

3 Attachment(s)

Quote:

Originally Posted by drb234

https://i.postimg.cc/pTXwhxX7/fafafaff.webp

Here's what I get:

Attachment 141751

pat wt dof chisq p f4rank Scotland_IA.AG_I16504.AG Denmark_IA.SG France_Metz_GalloRoman.SG feasible best dofdiff chisqdiff p_nested
000 0 11 7.219826361226613 0.781012303144483 2 -2.3261454614434283 2.785332857996349 0.5408126034470794 FALSE NA NA NA NA
001 1 12 7.88697960512831 0.7938990014869362 1 -3.5984938615857565 4.598493861585757 NA FALSE TRUE 0 -12.150180270926757 1
010 1 12 20.037159876055068 0.06638628004148951 1 1.9493325459839783 NA -0.9493325459839783 FALSE TRUE 0 -2.3379103343145786 1
100 1 12 22.375070210369646 0.033524961639489155 1 NA 0.6639322060997479 0.33606779390025215 TRUE TRUE NA NA NA
011 2 13 23.834834534897645 0.03268374617987084 0 1 NA NA TRUE NA NA NA NA
101 2 13 29.125875695879426 0.006282002727272953 0 NA 1 NA TRUE NA NA NA NA
110 2 13 39.393485483175716 1.729800161953792e-4 0 NA NA 1 TRUE NA NA NA NA

Rights:

Attachment 141752
Attachment 141753
07-16-2025, 02:26 AM
drb234

Quote:

Originally Posted by Gannicus

Here's what I get:

Attachment 141751

pat wt dof chisq p f4rank Scotland_IA.AG_I16504.AG Denmark_IA.SG France_Metz_GalloRoman.SG feasible best dofdiff chisqdiff p_nested
000 0 11 7.219826361226613 0.781012303144483 2 -2.3261454614434283 2.785332857996349 0.5408126034470794 FALSE NA NA NA NA
001 1 12 7.88697960512831 0.7938990014869362 1 -3.5984938615857565 4.598493861585757 NA FALSE TRUE 0 -12.150180270926757 1
010 1 12 20.037159876055068 0.06638628004148951 1 1.9493325459839783 NA -0.9493325459839783 FALSE TRUE 0 -2.3379103343145786 1
100 1 12 22.375070210369646 0.033524961639489155 1 NA 0.6639322060997479 0.33606779390025215 TRUE TRUE NA NA NA
011 2 13 23.834834534897645 0.03268374617987084 0 1 NA NA TRUE NA NA NA NA
101 2 13 29.125875695879426 0.006282002727272953 0 NA 1 NA TRUE NA NA NA NA
110 2 13 39.393485483175716 1.729800161953792e-4 0 NA NA 1 TRUE NA NA NA NA

Rights:

Attachment 141752
Attachment 141753

I used different rights that included bronze age samples from Britain, Scandinavia and Central Europe
07-16-2025, 02:31 AM
Gannicus

1 Attachment(s)

I haven't had much luck with medieval samples yet.

This is another good model:

Ireland EBA Rathlin2 + Minoan

Attachment 141754

pat wt dof chisq p f4rank Ireland_EBA.SG_rath2.SG Greece_Crete_HgCharalambos_EMBA.AG feasible best dofdiff chisqdiff p_nested
00 0 12 12.973786183094607 0.37094889791715024 1 0.7998928166849076 0.20010718331509242 TRUE NA NA NA NA
01 1 13 28.921362767488024 0.006716122080505627 0 1 NA TRUE TRUE 0 -398.4088686573338 1
10 1 13 427.3302314248218 3.732551633024282e-83 0 NA 1 TRUE TRUE NA NA NA

Same rights as previous post
07-16-2025, 06:28 PM
Gannicus

1 Attachment(s)

This run has a really good p-value:

Attachment 141775

Chi-squared = 4.87

pat wt dof chisq p f4rank Russia_Samara_EBA_Yamnaya.AG Turkey_Central_Catalhoyuk_N.SG Luxembourg_Mesolithic.DG_Loschbour.DG feasible best dofdiff chisqdiff p_nested
000 0 11 4.866129917986197 0.9374759112717503 2 0.44427443972017017 0.4109106704024678 0.14481488987736207 TRUE NA NA NA NA
001 1 12 22.56617198543811 0.03164309064139399 1 0.4636335952541724 0.5363664047458276 NA TRUE TRUE 0 -128.67045269679804 1
010 1 12 151.23662468223614 3.181448605685366e-26 1 0.8518076009691895 NA 0.14819239903081052 TRUE TRUE 0 128.04024231211946 0
100 1 12 23.196382370116694 0.026104235871953096 1 NA 0.7937460664741767 0.20625393352582339 TRUE TRUE NA NA NA
011 2 13 258.8652171497984 9.201495381431885e-48 0 1 NA NA TRUE NA NA NA NA
101 2 13 35.00377467190446 8.455469917767836e-4 0 NA 1 NA TRUE NA NA NA NA
110 2 13 1062.7549585939366 5.763691654010331e-219 0 NA NA 1 TRUE NA NA NA NA
07-16-2025, 08:00 PM
Gannicus

2 Attachment(s)

To quote DJ Khaled, another one :P

Attachment 141777
07-16-2025, 11:36 PM
majevica

Quote:

Originally Posted by AndreiDNA

Karelia also has CHG admix. They even have 4 people with J1 from mesolithic karelia.
Attachment 141747

I see, now my low AG3 in this model makes sense

https://i.postimg.cc/nhKvQxjN/temp-Image-Iak7gj.avif
07-17-2025, 01:12 AM
AndreiDNA

1 Attachment(s)

Attachment 141785
Just gonna drop this here
07-17-2025, 02:12 AM
Gannicus

Quote:

Originally Posted by AndreiDNA

Attachment 141785
Just gonna drop this here

Would that 10% Gorilla be a proxy for the unknown archaic lineage that is found in SSA populations?
07-17-2025, 02:28 AM
Gannicus

3 Attachment(s)

Attachment 141789

Modeled myself with Morocco_MN skh001 which is the sample used to model NUE001 from the Nature article. He was ~80% skh001 and ~20% Mesopotamia_N

Rights:

Attachment 141790
Attachment 141791
07-17-2025, 06:33 PM
AndreiDNA

i found something out about illustrative's qpadm.
kind of embarassing for the illustrative crew.

they have unique datasets for each sample uploaded
not for each user
meaning if you upload a second file, you're creating a second distinct dataset
not just merging your second file with the dataset that contains your first, which is how I would engineer this system if i was in charge
this seems very stupid to me
first of all, this will cost them
because each new upload is another 3Gb of space that they have to pay for
second, people won't be able to model themselves with samples that aren't already on the dataset
and lastly, nothing stops people from uploading more and more and more shit there, continuously eating up the available space on their servers, 3GB by 3GB by 3GB

with my services you can merge as many samples with the dataset as you want (within reason, don't give me too much work like merging 5 samples with the dataset)
so if you want me to model you as your mom + kura araxes, I can
07-17-2025, 10:21 PM
Gannicus

2 Attachment(s)

Did some FST runs:

Attachment 141806

Attachment 141807

Distances in order from closest to furthest:

Code:

Population FST Std. Error ---------------------------------------- Norwegian.DG -0.0049 0.00507 Czech.DG -0.0009 0.00513 English.DG 0.0008 0.00401 French.DG 0.0022 0.00308 Spanish.HO 0.0038 0.00302 Italian_Central.HO 0.0053 0.00311 Scottish.HO 0.0069 0.00339 Italian_South.HO 0.0079 0.00308 Greek_WGA.HO 0.0081 0.00311 Russian.DG 0.0094 0.00306 Basque.DG 0.0110 0.00315 Turkish.DG 0.0125 0.00386 Lebanese_Christian.HO 0.0141 0.00313 Iranian.HO 0.0143 0.00304 Tajik.HO 0.0143 0.00303 Syrian.HO 0.0155 0.00309 Assyrian.HO 0.0161 0.00315 Egyptian.HO 0.0209 0.00303 Saudi.HO 0.0245 0.00330 Balochi.HO 0.0254 0.00310 Moroccan.HO 0.0289 0.00313 GujaratiA.HO 0.0266 0.00337 Turkmen.HO 0.0237 0.00322 Uzbek.HO 0.0243 0.00308 Punjabi.HO 0.0388 0.00328 Uyghur.HO 0.0375 0.00325 Hazara.HO 0.0409 0.00318 Kalash.HO 0.0456 0.00327 Kazakh.HO 0.0488 0.00319 Kyrgyz_Kyrgyzstan.HO 0.0531 0.00337 Eritrea.HO 0.0518 0.00342 Afar_WGA.HO 0.0571 0.00330 Somali.HO 0.0755 0.00316 Mongol.HO 0.0816 0.00331 Thai.HO 0.0943 0.00333 Visayan.HO 0.1091 0.00363 Han.HO 0.1093 0.00336 Korean.HO 0.1096 0.00358 Japanese.HO 0.1104 0.00342 Vietnamese.HO 0.1108 0.00348 Quechua.HO 0.1217 0.00398 Zapotec.HO 0.1365 0.00365 Mixe.HO 0.1630 0.00376 Piapoco.HO 0.1659 0.00401 Pima.HO 0.1677 0.00385 Karitiana.HO 0.2066 0.00398 Surui.HO 0.2273 0.00426 Australian.HO 0.1816 0.00404 Papuan.HO 0.1902 0.00372 BantuKenya.HO 0.1461 0.00337 Yoruba.HO 0.1517 0.00332 Khomani.HO 0.1723 0.00346 Hadza1.HO 0.1890 0.00386 Mbuti.HO 0.2058 0.00353
07-17-2025, 10:59 PM
Gannicus

4 Attachment(s)

And finally, I've got a combination that assigns Natufian admix by itself:

Attachment 141808

Attachment 141809

References:

Attachment 141810
Attachment 141811

Edit: Andrei these are mostly your rights from a few models you made for me. I just added a couple samples and swapped out Pinarbasi for Boncuklu N.
07-18-2025, 02:08 AM
Fistora

1 Attachment(s)

Some of my Genetic distances
https://i.postimg.cc/50fp8FZH/image.png
07-18-2025, 03:03 AM
AndreiDNA

4 Attachment(s)

Attachment 141820 Attachment 141821 Attachment 141822
Attachment 141823

qpadm analysis of Poles from Wielkoposke, sent to me by a user on this forum
07-18-2025, 04:24 AM
Fistora

https://i.postimg.cc/XqhyVpF1/image.png
https://i.postimg.cc/HkG1Md9q/image.png
07-18-2025, 04:31 AM
Gannicus

2 Attachment(s)

Attachment 141824

Swapped out Iran N Wezmeh for Iran N Ganj Dareh
and added these to the references:

Attachment 141825

P value improved quite a bit. Was trying to get the standard errors lower. Oh well.
07-20-2025, 12:14 AM
J. Ketch

I'm a noob. Is there some method or reasoning to choosing the right reference populations?
07-20-2025, 12:27 AM
AndreiDNA

Quote:

Originally Posted by Creoda

I'm a noob. Is there some method or reasoning to choosing the right reference populations?

A. If you want to model using lefts that are real/exist, you select rights that are ancestral to your lefts, or if you can't find those that are ancestral, you have to search for those closely related to your lefts.

An example of that would be
your lefts:
EHG
WHG
ANF
CHG

then your rights have to be:
WHG
ANE (Ancestral to EHG)
ANF/you can break it down into AHG + Natufian + iran N
CHG

If you want to model
lefts:
ANE
WHG
Dzudzuana

You'd have to select rights that are ancestral such as
WHG
Yana (Ancestral to ANE), or alternatively something east eurasian like TY (Which is also ancestral to ANE)
Alternative Dzudzuana right, or
Natufian (Will work since dzudzuana can fit as a intermediate between Natufian, which is a right, and WHG, which is a right)
B. If you want to model with a ghost population, you select rights affinity to which could infer affinity to this ghost population, and the rest of rights you do as in point A.

An example of that would be modeling for basal eurasian admix. Since basal eurasian admix would decrease the affinity to neanderthals and denisovans, those have to be in the rights. Then you follow point A and chose lefts and rights for west eurasians and east eurasians.
07-20-2025, 12:29 AM
Lard

Quote:

Originally Posted by Creoda

I'm a noob. Is there some method or reasoning to choosing the right reference populations?

Hey, I just started out too — I’ve been using the qpAdm interface on Illustrative DNA, which makes it a lot more accessible for beginners.

As for choosing references ("right pops"), it’s a mix of PCA-based reasoning, historical context, and trial-and-error. I break it all down in a thread I posted here:

➡️ �� Stable qpAdm Models of Ashkenazi Jewish Ancestry – Ignoring PCA No More

Would love to chat more over there — I put a lot of effort into the models and framework and it barely got a reply, lol. Feel free to ask anything in that thread. Happy to help however I can.
07-20-2025, 01:21 AM
Gannicus

5 Attachment(s)

My Dzudzuana + ANE admixture:

Attachment 141866

References:

Attachment 141867

I revisited a model AndreiDNA made for me where I was modeled as 3 way mix of Dzudzuana, Italy Epigravettian, and Afontovagora3 (ANE)

Attachment 141868

Chi-squared= 6.6

pat wt dof chisq p f4rank Georgia_Dzudzuana_UP.AG_I11857.AG Italy_Epigravettian.AG Russia_AfontovaGora3_UP.AG feasible best dofdiff chisqdiff p_nested
000 0 7 6.604471680174148 0.47118912498381044 2 0.6190138010869063 0.23446826784160432 0.1465179310714894 TRUE NA NA NA NA
001 1 8 11.519608968168374 0.17395876562151708 1 0.7146053472659971 0.2853946527340029 NA TRUE TRUE 0 -0.6103912843959698 1
010 1 8 12.130000252564344 0.1454971638521046 1 0.7886224100986546 NA 0.21137758990134542 TRUE TRUE 0 -87.06245602662968 1
100 1 8 99.19245627919402 6.242381618212807e-18 1 NA 0.21391679165755226 0.7860832083424478 TRUE TRUE NA NA NA
011 2 9 18.588268854606014 0.0289304686658288 0 1 NA NA TRUE NA NA NA NA
101 2 9 569.2036842919447 8.482679409878293e-117 0 NA 1 NA TRUE NA NA NA NA
110 2 9 103.07209455485025 3.756961650459967e-18 0 NA NA 1 TRUE NA NA NA NA

Attachment 141869

For the result on IllustrativeDNA (First Image), I used AndreiDNA's rights and added Ethiopia 4500 BP (Mota) to the references:

Attachment 141870
07-20-2025, 02:17 AM
Gannicus

2 Attachment(s)

Quote:

Originally Posted by AndreiDNA

A. If you want to model using lefts that are real/exist, you select rights that are ancestral to your lefts, or if you can't find those that are ancestral, you have to search for those closely related to your lefts.

An example of that would be
your lefts:
EHG
WHG
ANF
CHG

then your rights have to be:
WHG
ANE (Ancestral to EHG)
ANF/you can break it down into AHG + Natufian + iran N
CHG

If you want to model
lefts:
ANE
WHG
Dzudzuana

You'd have to select rights that are ancestral such as
WHG
Yana (Ancestral to ANE), or alternatively something east eurasian like TY (Which is also ancestral to ANE)
Alternative Dzudzuana right, or
Natufian (Will work since dzudzuana can fit as a intermediate between Natufian, which is a right, and WHG, which is a right)
B. If you want to model with a ghost population, you select rights affinity to which could infer affinity to this ghost population, and the rest of rights you do as in point A.

An example of that would be modeling for basal eurasian admix. Since basal eurasian admix would decrease the affinity to neanderthals and denisovans, those have to be in the rights. Then you follow point A and chose lefts and rights for west eurasians and east eurasians.

Hey Andrei, I was modeling Morocco_MN skh001 and had some interesting results:

Attachment 141871

References:

Attachment 141872

Takarkori isn't available on IllustrativeDNA. I was wondering if you could try this model with Takarkori instead of Iberomaurusian. And try it with Iberomaurusian absent from the references and in. I want to see if skh001 is going to score around 10% Takarkori in this setup and the 16% or so assigned to Natufian.
07-20-2025, 03:06 AM
Lard
PCA vs. qpAdm: More Similar Than People Think

Fundamentally, autosomal PCA and qpAdm modeling operate on the same basic idea:
They both assess how populations are related based on genome-wide allele frequencies — just using different statistical approaches.
- PCA uses principal components to reduce genetic variance into 2D/3D plots, revealing clustering patterns.
- qpAdm estimates ancestry proportions by modeling a test population as a mix of chosen “left” sources, with “right” reference groups to anchor the model.
But in both cases, your results depend almost entirely on your inputs.
You can cherry-pick populations in either method — and get misleading results that still “look” scientific.

Why qpAdm Isn't Immune to Narrative Bias

There’s a tendency to treat qpAdm as more rigorous than PCA, and while it can be more formalized, it’s just as vulnerable to biased modeling choices.

That’s why so many published studies claim Ashkenazi Jews are “50%+ Levantine” — even though PCA consistently places them overlapping with Southern Italians.

Here’s why:
Most of these studies use deeply flawed European proxies — Northern Italians, French, Sardinians, or even Tuscans — while completely ignoring Southern Italians or Sicilians. This artificially inflates the Levantine component, because these proxies are genetically too distant and don't match PCA placement.

The qpAdm model still returns a “feasible” result — because qpAdm doesn't know your sources are bad.
It just fits whatever you give it.

In other words: Bad source selection in qpAdm doesn’t get corrected by the math — it gets rewarded. The tool will give you a p-value and ancestry breakdown that looks clean, but is fundamentally built on cherry-picked assumptions.

PCA-Informed qpAdm is the Solution

That’s why my models start with PCA — which clearly shows Ashkenazi Jews clustering directly on top of Southern Italians — and then use qpAdm to test models that reflect that structure.

Examples from my thread:
Stable qpAdm Models of Ashkenazi Jewish Ancestry – Ignoring PCA No More

Models such as:
- Italian_South.HO + Lebanese_Muslim.HO
- Italian_South.HO + Israel_Ashkelon_LBA.AG
- Italian_South.HO + Israel_C_o2.AG
All produce stable 2-way models with:
- ~75–78% Southern Italian–like ancestry
- ~22–25% Levantine
- Strong p-values and standard errors
- Full consistency with PCA and historical records
This matches not just the genetic structure, but also the well-documented Roman and Byzantine-era Jewish presence in Southern Italy.

Visual Results

https://i.ibb.co/M0ZfVqC/1.png
Modern Levantine Proxy model

https://i.ibb.co/S4FnRm4s/2.png
Bronze Age Coastal Levantine (Philistine-admixed)

https://i.ibb.co/G43Kn5wk/3.png
Inland Bronze Age Levantine (likely Israelite/Canaanite)

Don’t Let Narrative Modeling Skew the Data

You can go check nearly any study by Behar, Hammer, or Ostrer — they almost all use Northern Italians or French as the European side of the model.
Even YouTubers like Andrei DNA — who is also an Apricity forum member — repeat the same setup and get the same distorted results: Ashkenazi Jews as a “Middle Eastern” population.

But if Ashkenazi Jews sit right on top of Southern Italians in PCA — and they do — then they’re no more “Middle Eastern” than a Calabrian is.

That’s why PCA-aligned qpAdm is essential — and narrative-driven modeling needs to be called out.
https://www.youtube.com/watch?v=5gxr79HfpYQ

Show 40 post(s) from this thread on one page