Another dataset of high coverage shotgun genomes, courtesy of Ranis

[ChrisR]

@TanTin @kolompar thx for your answers.

When looking at an old Francioli, Menelaou et al 2014 Figure I still hope for Europe or West Eurasia there is a chance using all existing variation could give a chance to separate countries/populations rather nicely - not sure how well 1000G EUR represented the Netherlands neighboring SNP variation but if a rather small country like Netherland according to GoNL has 7.6 M "custom SNPs" this seems like some potential.
Obviously we would need both modern and ancient DNA in HQ and in proper numbers.
And then there is the problem if there is any chance consumer computers can handle this in timely/fluid manner...


Is there any more recent and much more comprehensive similar analysis for Europe comparing a smaller country or region to the rest of the continent?

[kolompar]

May I ask if for a Postglacial to pre IA analysis, maybe for Western Eurasia, are there enough samples?

My personal rather subjective impression is for Central and Western/South Europe the multiple waves of incoming East/Northern admixture (Steppe+EEF+WHG) are near impossible to separate by the existing sub 2M SNP admixture analysis. Speaking about cultures anything from BB, Gauls, Italics to Germanic and Slavic influences make many modern populations from the Alps to Iberia and beyond very similar to each other even if in the last 40 to 60 generations there was no substantial genetic exchange.
To overcome this from my limited understanding HQ WGS results from all involved populations/cultures up to the Medieval Slavs are needed. Not sure if this dataset already allows this, but it seems it could at least substantially refine the pre Holocene era.

Your question is excellent, but I am afraid this data will also fail to provide us such answer. 
And there are some reasons why. Many such reasons.

To separeate genetically  BB, Gauls, Italics to Germanic and Slavic ?
Do you know how many thousand years of sepparation are needed to have such genetic difference accumulated ?

To have some idea:  for the Y-chromosome - you need near 1000 years to have 1 snip modified on the Y-chromosome.

That can't be right because we have haplogroups branching within hundreds of years like F or R1b-M269. Actual rate seems to be around 1 mutation per generation.
And the Y-chromosome shows Slavs and such were relatively small founding populations so autosomes should work too. Even G25 with its few hundred thousand SNPs can tell them apart relatively well.
There's 7 million SNPs here, if you take chimp-human split that's like 1 SNP per year to work with.

[TanTin]

I'd like to share with you this report for CHR17.

Here is an example for Denisova/Neandertal variants among AMH.
There are 2 options only:  people have the ancient ancestral segment matching to D/N / Chimp.
Or people have the modern variant, which is also matching to some modern Africans (Yoruba)..

Apparently this segment did not occure OOA.  It existed long before OOA events or may be near the time when OOA happened.
The fact there is such long segment is telling us such non_Denisova / Non-Neandertal modern population already existed.
So we can see the Papuans also have 50/50  . However a small portion of this segment was broken in the Papuans and for that small portion the Papuans have 100% the D/N version.
The selection of these snips is result of several filters. This segment may be already broken in smaller pieces.
I present the result after I applied my filters.
I will provide the position number of snips for this report.  So this is generally a Denisova/Neandertal segment present in AMH.  Near 50% of the people will have it, the other 50% should have the opposite variants.

The position numbers are provided in  "genome-wide shotgun data"  or  Ranis_7M .

[TanTin]

I did additional verifications on " high coverage shotgun genomes" or Ranis "genome-wide shotgun data" ...
https://edmond.mpg.de/dataset.xhtml?pers...7/3.EGKV28

I applied some specific filters. The same criteria was applied to both datasets for chromosome N 15.

For Ranis " high coverage shotgun genomes" ( 7M ) I have 1257 snips which passed the filter criteria..

Applying the same filter for 1240k Dataset ( D. Reich dataset): I have 708 snips which pass the filter.

So my expectation was that most of these 708 or at least 50% if them will overlap with 7M dataset results..
However that's not the case at all !

Only 5 snips were present in both datasets . These are the overlap snips.

V1 V2 V3 V4 V5 V6 V7
1 15 rs11161209 0.241354 25952220 T C 9
2 15 rs4924592 0.561048 42270730 C T 9
3 15 rs4777979 1.138900 92965421 A G 9
4 15 rs6497126 1.190330 94285169 C T 9
5 15 rs10520808 1.256360 96647766 C A 9

That means the 1240k dataset is covering completly different scope of snips, compared to the " high coverage shotgun genomes" Ranis 7M dataset.

The overlap between both datasets could be somewhere 5% - 10 % .. Will need to check in more details on this.

[TanTin]

Checking for all snips on CHR15:
dataset Ranis: 215k
dataset 1240k: 36k

Shared snips: 18.7k

The number of shared snips is 50% of dataset 1240k or ~9 % of dataset Ranis .
However when we apply more strict criteria for filtering the number of shared snips may drop to almost 0 %.
Dispite the significantly higher number of snips in Ranis dataset, half of the 1240k snips were not included here.