lauantai 9. kesäkuuta 2018

QpAdm tests, Iron Age Scanian sample RISE174

I've been a long time waiting for ancient Finnish samples and it starts to look  unlikely that I'll ever see them because the Finnish soil is acid and destroys organic material in one millennium.   But don't ever lose hope, it was available all the time from Internet. An old Finnish sample has been available already three years in open data bases.  It was found from the Southern Sweden and was labeled as RISE174, dated 427-611 AD, later Scania_IA.  That is right, we have had three years an ancient Fennoscandinavian sample, common for all Fennoscandinavians and also for many Baltic people, but obviously because it was found from Southern Sweden it's real value was not noticed before.  The reason could also be partly a common prejudice.   I wouldn't be much surprised if Finnish researchers in future will find similar Iron Age samples from Southwestern Finland, if it is possible due to the Finnish soil.  I'll show that present-day Swedes are not much more related to this sample than present-day Finns.  Generally speaking, all qpAdm-results listed below are reliable, although there were many neighboring cultures in the past, being also genetically very similar and in any tests all populations close enough can substitute each other.  

QpAdm results are made using distant ancestral references (the right-file)  to get best possible coverage.  While these results are very reasonable, it is possible to fine-tune all results by using carefully selected less distant references.  So these results are directional.  

All Finnish groups are from the 1000-genomes project and cover full 1MSNP's, as well as all other groups on the second and third plots.  I have seen in my work that testing ancient and modern samples calls for equal coverage of all modern sample groups. 

What kind of Fennoscandinavian was the RISE174/Scania_IA?  Definitely she was not like a present-day Swede, we can see it in following f3-result below and on PCA-plots.  

Next here is a simple PCA-plot, somewhat imperfect one and flat due to a high amount of Asian influence and too few European samples.   This plot shows a new Finnish group, Finnx, and binds it to the Finnish ancestry.  Finnx forms an ordinary Finnish group with less present-day Swedish admixture than typical Southwest Finns, but also less genetic drift than typical East Finns. 

Another plot (vectors 1/2 and 1/3), removed almost all East Asian and Siberian.  There is still a touch East Asian forwarded by South and Central Asians.  It was necessary to remove East and Siberian components to ignore later Siberian admixture in Finland and see the possible Finnx-Scania_IA connection also on PCA.  On the second plot (vectors 1/3) the East Asian is at its least and Scania_IA gives the best match with Finnx. 

 F3-test showing common genetic drift with Scania_IA:


First some European groups in a demonstrative purpose. 

South Italians

Chisq and tail prob:   0.436        0.932651 

Populations:  Beaker_Central_Europe, Iran_LN, Levant_BA
best coefficients:     0.299     0.169     0.531
Jackknife mean:      0.299685849     0.169473438     0.530840713
      std. errors:     0.028     0.032     0.033

North Italians

Chisq and tail prob:   1.738        0.628609

Populations:  Beaker_Central_Europe, Levant_BA, Iran_LN
best coefficients:     0.542     0.416     0.042
Jackknife mean:      0.542422483     0.415568791     0.042008725
      std. errors:     0.019     0.020     0.020


Chisq and tail prob:   1.789        0.774528

Populations:  Beaker_Central_Europe,   Iberia_EN
best coefficients:     0.578     0.422
Jackknife mean:      0.579234989     0.420765011
      std. errors:     0.030     0.030


Chisq and tail prob:   1.101         0.77684

Populations:  Scania_IA, Latvia_MN, Evenk
best coefficients:     0.920     0.092    -0.012
Jackknife mean:      0.918423990     0.093588254    -0.012012243
      std. errors:     0.042     0.042     0.016


Chisq and tail prob:   1.699        0.790857

Populations:  Poltavka, Germany_MN
best coefficients:     0.502     0.498
Jackknife mean:      0.502425062     0.497574938
      std. errors:     0.039     0.039


Chisq and tail prob:   0.642        0.886697

Populations:  Scania_IA, Latvia_MN, Evenk
best coefficients:     0.933     0.052     0.015
Jackknife mean:      0.931374297     0.053427705     0.015197998
      std. errors:     0.039     0.038     0.015


Chisq and tail prob:   0.503        0.973198

Populations:   Sarmatian, Scania_IA
best coefficients:     0.472     0.528
Jackknife mean:      0.475543261     0.524456739
      std. errors:     0.079     0.079


Chisq and tail prob:   1.003          0.9093

Populations:   Latvia_LN, England_N
best coefficients:     0.523     0.477
Jackknife mean:      0.520472837     0.479527163
      std. errors:     0.052     0.052


Chisq and tail prob:    1.647        0.648853 

Populations:    Scania_IA, Latvia_MN, Evenk
best coefficients:     0.851     0.084     0.065
Jackknife mean:      0.849474735     0.085348593     0.065176672
      std. errors:     0.037     0.036     0.014 

Some "what if" analyses:

What if Mordvas are Polish

Chisq and tail prob:    11.941       0.0177937

Populations:    Poltavka, Germany_MN
best coefficients:     0.727     0.273
Jackknife mean:      0.724786064     0.275213936
      std. errors:     0.047     0.047

What if Latvians are Polish

Chisq and tail prob:     9.038       0.0601576

Populations:   Poltavka, Germany_MN
best coefficients:     0.563     0.437
Jackknife mean:      0.562264229     0.437735771
      std. errors:     0.049     0.049

What if Finnx's are Mordva

Chisq and tail prob:    4.728        0.316351

Populations:   Sarmatian, Scania_IA
best coefficients:     0.369     0.631
Jackknife mean:      0.377367376     0.622632624
      std. errors:     0.093     0.093

What if East Finns are Mordvas

Chisq and tail prob:    6.593           0.159

Populations:    Sarmatian, Scania_IA
best coefficients:     0.317     0.683
Jackknife mean:      0.329388055     0.670611945
      std. errors:     0.118     0.118

What if Swedes are Baltic

Chisq and tail prob:    1.867        0.867159

Population:   Scania_IA, Latvia_MN, Evenk
best coefficients,  Jackknife optimisation is negative, 1.000     0.000    -0.000

The results above show the continuum of Scania_IA from Southern Sweden to Russia, to the area where Mordvas live.  Those populations living in eastern and  western areas have however different admixtures, Mordvas lack Baltic middle-neolithic ancestry and show instead late East European Steppe -like admixture, Sarmatians suggesting incursions of the Iranian speakers.  What is also remarkable is that the best Swedish result doesn't include Scania_IA.  This doesn't mean that they have not Iron Age ancestry from Scania and in reality they are really close Scania_IA.  It means only that present-day Swedes are slightly different, own later admixtures and lack of certain older Baltic admixtures.  

torstai 31. toukokuuta 2018

Playing with Eigensoft qpAdm

I've been somewhat cautious in using qpAdm, because it looks much easier to use than it actually is.  You can make admixtures, be happy with them, but are your results as reliable as you think?   I am not going to bore you with details and instead make some examples.   Following three examples are made using same references (right file).

Results picked up from detail rows:

Secondly same results selected by the qpAdm:

The EHG stands for East European Hunter-Gatherer
The Belt_Cave Me stands for  Belt_Cave_Mesolithic_LC, the name was truncated during copy-paste operation. 

You can see that in the result of East European Hunter-Gatherer there is definitely something wrong, although the results picked up from detail rows look great. 

Update 1.6.2018 10:35

My answer to Anynyymi/Anonymous is:  yes it works at least for Latvians, but I won't continue making occasional tests with this left model, because I am more ambitious.  It is just not fair to ask for and show occasional results and let them become forgotten.  We need a larger demo.

1.089        0.779712     0.522     0.233     0.245

Update 1.6.2018 11:30

The example below shows the reason why I need more time and don't want to follow any trains of thought outside of my own head and make detailed tests.  Sorry, if I do, it could mislead readers and I just want to dedicate myself to the point here.

torstai 24. toukokuuta 2018

The devil lurking behind Botai, WSHG and modern population connections in West Eurasia

I've seen that the West-Siberian Hunter-Gatherer ancestry (WSHG) is in several texts connected to modern day Northeast Europeans  (obviously not by original studies) and now later the same theme is again on topics of new Botai samples.  The Botai was firstly presented by  Damgaard at al. 2018.   But have we one or many explanations for these assumed connections between ancient Siberian-like people and present-day population in Asia and Europe?  Was the connection, mostly even subtle, a result of any migrations or are tests only showing some common ancestry between WSHG/Botai and present-day populations?   I have not seen Damgaard et al. due to the payment wall, but Jeong et al. 2018 gives similar results I have got using f3-statistics.   My tests, as well as Jeong, shows extreme high common drift between Botai samples and ANE and WSHG acts very similarly.  ANE is presented by the Mal`ta child, who lived in Siberia about 24,000 years ago.  F3-statistics show that WSHG and Botai shows high common drift also with EHG (Eastern Hunter-Gatherer), which of course means that EHG can act like a WSHG/Botai proxy to West-Eurasian populations and vice versa.  The EHG ancestry is very common everywhere in Central, Eastern and Northern Europe, but highest in Northeast Europe and in Baltic Sea region.

tiistai 22. toukokuuta 2018

F3-analysis of Botai-culture gives contradicting results

The Botai culture was an ancient culture which lived around 3500 BC in Kazakhstan.

Some researches state that horses were domesticated by this culture. The recent study, Damgaard at al. 2018, published three samples of Botai culture.  There has been some discussion about a genetic link between present-day Saami  and Botai people, seen by IBD-segments on public Gedmatch service.   IBD-statistics however has a shortcoming in analyzing ancient samples, because a) IBD-segments become mostly random after around 1500 years and b) ancient samples has usually too low quality to form usable IBD-segments.  However,  my f3-statistics show high common drift for Saamis and Botai samples.  This similarity seems to continue in Finland and Scandinavia,  but in lesser amount  in Mongolia, which probably means an artificial effect between Saamis and Botai samples, generated by a coincidentally similar admixture of Siberian and ancient Steppe cultures in both populations.

 Update 24.5.18 11:30

Here are two updates, the first one listing Botais' top ten of modern populations and the second one the top ten of ancient populations.  I hypothesize of Botais' admixture that is is mainly similar to West-Siberian Hunter Gatherers, with a lump of Eastern Hunter Gatherer admixture, reasoning this by two top matches of ancient populations. Obviously the Eastern Hunter Gatherer admixture ties the result with Fennoscandinavia. The influence between WSHG and EHG is interesting and calls for more study.


torstai 3. toukokuuta 2018

Homozygosity by descent figures in Eurasia

This is not aimed to be an overall test including all Eurasian populations. Basically I have included only some samples covering over 400000 markers.   In some cases sample sizes were limited by too few individual per population and I dropped all population including less than 3 samples.  Even this is absolutely too little to get steady results.  It is also possible that some sample groups are poorly gathered and don't represent given population ideally.  Anyway this test gives a general view about existing homozygosity by descend in Eurasia.  The definition of Homozygosity by descendPossessing two genes at a given locus that are descended from a single source, such as may occur in inbreeding or in consanguineous mating.  

If my readers are interesting in seeing larger data and more populations, for example East Asians, Africans and Americans, I can rerun this test.   

Results are generated by Beagle software.

Update 3.5. 19:15   Fixed Turkish sample size and LOD-scores
Update 3.5.  20:47  Removed Turkish results, because some individuals showed excessive homozygosity


tiistai 24. huhtikuuta 2018

The naked truth about Dstat and Nganasan ancestry in Europe, updated

Maybe you are already bored of this issue, but I have to return to it, because it is so crazy and studies still use Nganasans to prove historical events in Europe.  It is also important to correct false conclusions.

In following tests I compare Nganasans, Native Americans, Europeans and some Asian populations and I prove that tests using Dstat and Nganasans are the craziest folly I have seen in population genetics.    My results imply that Nganasans went through a genetic drift during thousands years losing almost all ANE (Ancient North Eurasian) ancestry.  ANE is commonly found everywhere in Western and Northern Eurasian  and Native American people.   It is possible that they later got some genes from early Uralic speakers and albeit of the all  homogeneity they show in simplified comparisons the latest admixture event they have.

Update 26.4.18 10:20

After testing all available Siberians and their ANE/ENA ratios I found that actually Ket-people share highest drift with Finns, much more than Nganasans.  What probably is misleading people with Nganasans is that when comparing to other Siberians, the Finns share proportionally more drift with Nganasans than for example Lithuanians.  Still Lithuanians share more drift with Ket people than Finns with Nganasans.   My reckoning about this is that we (Finns) have received two Siberian migrations, the first one occurred when the Uralic language was brought to the Baltic Sea region and the second one via northeastern route and was forwarded by Saami people.

sunnuntai 15. huhtikuuta 2018

Is the Nganasan ancestry of Uralic speakers proven?

Lamnidis et al. 2018 preprint shows that all tested Uralic speakers except Hungarians show highest genetic drift with Nganasans.  Maybe, but it is not a piece of cake to prove, because Nganasans actually act like a proxy population for all North and East Asians.   Here is the original Dstat-table from the preprint

We see that Saamis, Finns, Russians (obviously North Russians) and Mordovians show positive drift with Nganasans.   But if we compare Nganasans to Beijing Chinese we see that also Altaians, Bengalis, Mongolians and Hazaras have more drift with Nganasans than with Chinese people. 

It is easy to find out the problem,  we have serious problems in using such a small inbred population as a reference, for example Hazaras should have Mongolian ancestry, but statistics  shows more Nganasan than Mongolian drift.

 Lithuanian     Hazara    Mongola      Mbuti     -0.0618   -21.006  50384  57016 890006
 Lithuanian     Hazara    Ngan           Mbuti     -0.0624   -17.118  50494  57211 890006

Nevertheless, in my opinion the Asian admixture in Finland is North Asian, but wrong conclusions are always possible when we have more that two variables.