They only scored 20% Cushitic on qpAdm. Think G25 elevates Cushitic ancestry as it does ANF
No offence, sxb, But that is nonsense. The results undercuts Cushitic ancestry.
I know how to roughly control for this. It is, Nilo-Saharan DNA has exessive enriched Shum-Laka type signal.
I figured that the Iron Age Pastoral sample average had ~40% Dinka-like DNA on top of a Somali-like profile. It had 8.2% West-central African hunter-gatherer ancestry, something I ascertained through a model I devised.
Then, by the assumption that Nilo-Saharan ancestry from Wadi Howar would essentially have the same signal, that could reflect a proxy for a relative measure. So I needed to test this. 8.2% represents 40% Dinka-like ancestry. Right? So I went to check how much Shum Laka appeared for the Danagla sample. It gave me, 5.2%.
What is 5.2 of 8.2? It is 63.4%. What is 63.4% of 40? It is 25.
Thus we should expect around 25% Nilo-Saharan to be the result for the Danagla sample.
Well, it checks. Lets do the same for Mahas.
skipping the stuff, it is 21.4%.
Yet again, it checks.
The third one, Halfawi.
4.2%.
It should be at 20%.
Yet it gets 16%....
Why is this the case? I suspect it is because it got minor West African admixture or because the NS has a HG skew, probably sourcing the NS from Saharans from further West.
By the way, the AEA Somalis have 0% of that signal. That is why this is a near-perfect benchmark to delineate NS ancestry from the Cushitic AEA link.
We're supposedly 60% AEA but have zero of the HG signal.
Now, to the conversation, the Christian Nubians were less NS than modern Nubians.
They on average score less than modern Nubians, and they have less than 20% on average:
The average from R cemetery.
This is 14.6% NS.
Checks again.
Then the S Cemetery:
11.68% NS.
Initially, it gave a high number, 18%. And this is a discrepancy. Then I wondered again, similar to the Halfwai, maybe it prefers something different with a stronger shift. I picked the samples from a Nilotic girl who shared her results, and she fit the bill:
I say this to say, the tool the guy used to infer that NS should be more enriched and Cushitic reduced does not hold water. I know this tool fairly well and one can figure out if something is an overfit or disproportionate skew.
Nubians are actually substantially Cushitic. The Christian period averages are lower than the modern averages we have. Most of the time, I can replicate results from qpAdm if it is sound. Many times it simply is not. You will not get the correct results if one does not model right, no matter if you use qpAdm or G25.
