qpAdm and the genetic cues on the ancient origins of feylis part 1
I tried modeling the Feyli samples using the following base populations referenced in Lazaridis 2016 with qpAdm:
1- Iran Neolithic
2- Anatolian Neolithics
3- Levant Neolithics
4- EHG IO061
The same outgroups were used as in Lazaridis 2016.
The majority of the Feyli samples received mediocre to poor fits, thus indicating that some other population(s) were involved in their histories. I did not use Iran Chl as it was shown to be a CHG-Iran N composite in Lazaridis 2016, and thus would hide additional Iran N admixture.
I used EHG, since the more recent steppe groups such as Andronovo, Potapovka, and Sintashta are themselves W Asian farmer admixed and thus their non farmer true steppe geneflow (EHG) would become clouded. Nonetheless, it would be useful to use those W Asian farmer admixed steppe groups to get a model more reflective of realistic admixture events for Feylis, in which case Iran Chl can also be used, if the goal is not to accurately determine geneflow attributable to the aforementioned base groups, but rather to determine contributions from populations that interacted in SC Asia or the Iran area. I will attempt such a model later.
Any fits shown below good, have a significantly higher probability of error.
EXCELLENT FIT SAMPLE IRAN NEOL ANATOLIA NEOL LEVANT NEOL EHG HAN TAIL PROBABILITY CHISQ Kurd_F7 46.5% 37.9% 0.0% 10.9% 4.7% 99.7% 0.34
GOOD FIT SAMPLE IRAN NEOL ANATOLIA NEOL LEVANT NEOL EHG HAN TAIL PROBABILITY CHISQ Kurd_F4 39.6% 43.3% 0.0% 11.9% 5.3% 85.2% 1.98
MEDIOCRE FITS SAMPLE IRAN NEOL ANATOLIA NEOL LEVANT NEOL EHG HAN TAIL PROBABILITY CHISQ Kurd_F6 61.3% 0.0% 19.3% 17.8% 1.6% 67.2% 3.18 Kurd_F2 45.6% 38.0% 0.0% 8.3% 8.1% 65.9% 3.26 Kurd_F5 50.4% 0.0% 23.8% 25.8% 0.0% 58.4% 4.69
POOR FITS SAMPLE IRAN NEOL ANATOLIA NEOL LEVANT NEOL EHG HAN TAIL PROBABILITY CHISQ Kurd_F8 32.1% 0.0% 39.3% 23.2% 5.4% 29.2% 6.15 Kurd_F1 50.2% 0.0% 29.0% 20.8% 0.0% 26.4% 7.66 Kurd_F3 45.9% 17.1% 16.0% 21.0% 0.0% 14.7% 8.2
A quick comparison of the Feyli samples F7 and F4, with northern Iraqi Kurd samples C1 and C3, which also received excellent fits, indicates that Kurd C1 and Kurd C3 received about the same amount of geneflow attributable to Iranian Neolithics, however, the steppe contribution attributable to EHG is almost double.
SAMPLE MIXING PROPORTIONS
SAMPLE IRAN NEOL ANATOLIA NEOL LEVANT NEOL EHG HAN TAIL PROBABILITY CHISQ Kurd C1 44% 33% 4% 19% 0% 97% 0.9 Kurd C3 42% 37% 0% 21% 0% 94% 1.8
Using Andronovo in lieu of EHG, and keeping everything else the same, produced the best fit I have ever seen with qpAdm for Feyli sample F7; Tail probability of 99.9% and a super low chisq of 0.08. As expected the steppe contribution for both Kurd F7 and Kurd C3 increased, since Andronovo is itself less EHG. However, Andronovo is likely a more realistic scenario than EHG, if the goal is not to determine actual EHG geneflow.
Feyli F7 may not represent an average for the feyli samples, but is shown because its fit has the lowest probability of error for all the feyli samples.
EXCELLENT FIT SAMPLE IRAN NEOL ANATOLIA NEOL LEVANT NEOL ANDRONOVO HAN TAIL PROBABILITY CHISQ Kurd_C3 35.5% 24.2% 0.0% 40.4% 0.0% 98.3% 1 Kurd_C3 33.7% 26.0% 0.0% 39.2% 1.1% 96.4% 0.98 Kurd_F7 42.5% 13.6% 12.7% 25.9% 5.3% 99.9% 0.08 Kurd_F7 43.0% 0.0% 21.7% 30.1% 5.3% 99.8% 0.27
EDIT: Iran Chl was used but produced a significantly worse fit than using Iran N. The fit can be improved by using a combination of Iran Chl and Iran N, but this scenario would not be informative.
EDIT 2: The following steppe surrogates were also used for the above, in lieu of Andronovo:
The fits were good using Afansievo, Steppe EMBA / MLBA, but were worse than using Andronovo.
Fits using Sintashta were significantly worse than all the above.
Last edited by Kurd; 07-20-2016 at 01:12 PM.
Ran over the 10 samples and noticed all of them have twice as much South Asian, whats the reason for that? I have barely 5,5% while many of the others have around 10 - 12%.
1 Caucasus_HG 44.28
2 Anatolian_NF 26.35
3 Near_East 13.43
4 European_HG 6.29
5 South_Asian 5.48
6 Siberian 1.47
7 Sub-Saharan 0.95
8 East_Asian 0.91
9 Oceanian 0.83
Last edited by Zoran; 08-19-2016 at 04:35 AM.
Last edited by Hirontumen; 01-16-2017 at 01:23 PM.