eurasiadna
Junior Member
- Messages
- 7
- Reaction score
- 1
- Points
- 0
Hi,
This is pretty crazy but I do believe in it. I've been looking at aDNA stuff for quite a few years and so I am quite competent at analysing data. I mostly use fstats and pcas for analysis.
It's really profound, the academics have most things back-to-front, the wrong way round.
They are confusing drift with admixture.
Steppe DNA is not a result of admixture of different sources (EHG/ANE and CHG/Iran_N) but is the ORIGIN of most Eurasian DNA.
So if you imagine that Steppe DNA consists of a large number of Nomadic peoples, they are moving around and settling in different places over time. So Steppe DNA is like the root of a tree of Eurasian DNA, then Iran_N, CHG, Anatolian, WHG, EHG, ANE they are all branches separating off Steppe DNA. When these samples are placed in ADMIXTURE it gives a unique DNA profile to the most drifted samples (EHG, CHG etc) and gives SteppeDNA a combination of some of them. So scientists are incorrectly interpreting this as SteppeDNA being a mixture of EHG and CHG/Iran_N but really EHG and Iran_N are derived from SteppeDNA through drift, and EHG and Iran_N dont share alleles not shared with SteppeDNA because they are going in opposite geographic relative to SteppeDNA, so it is incorrectly interpreted that SteppeDNA is a mix of the two.
People will say that SteppeDNA is much newer than Iran_N and CHG and EHG so this cannot be true. But we simply havent sampled SteppeDNA earlier than the Eneolithic, but it would have existed earlier. The ancestor of SteppeDNA living in 7,000BC or whatever would not have been so different to SteppeDNA that we have sampled.
There are various arguments I can make for this:
1. The mixing event with EHG and CHG/Iran_N has not been found in the steppe, and will never be found. CHG that could have mixed with EHG will never be found in the PC steppe.
2. It cannot be determined with certainty whether the Southern component was CHG or Iran_N. This is because both of these have inherited SteppeDNA instead of being a source.
3. Steppe DNA is everywhere. Anatolia_N has too much and TepeHissar has way too much.
But this is the proof.
Anatolian_N is too close genetically to SteppeDNA for the current model. Steppe DNA is too close to every Eurasian group to be anything other than Central in the phylogenetic tree.
f3 A B C
1 207. Anatolia_N Russia_Afanasievo Chimp.REF
2 205. Anatolia_N Russia_Sidelkino_HG.SG Chimp.REF
3 204. Anatolia_N Iran_GanjDareh_N Chimp.REF
4 202. Anatolia_N Russia_MA1_HG.SG Chimp.REF
5 194. Anatolia_N Luxembourg_Loschbour Chimp.REF
6 0 Anatolia_N Anatolia_N Chimp.REF
7 206. Iran_GanjDareh_N Russia_Afanasievo Chimp.REF
8 204. Iran_GanjDareh_N Anatolia_N Chimp.REF
9 203. Iran_GanjDareh_N Russia_Sidelkino_HG.SG Chimp.REF
10 201. Iran_GanjDareh_N Russia_MA1_HG.SG Chimp.REF
11 188. Iran_GanjDareh_N Luxembourg_Loschbour Chimp.REF
12 0 Iran_GanjDareh_N Iran_GanjDareh_N Chimp.REF
13 197. Luxembourg_Loschbour Russia_Sidelkino_HG.SG Chimp.REF
14 195. Luxembourg_Loschbour Russia_Afanasievo Chimp.REF
15 194. Luxembourg_Loschbour Anatolia_N Chimp.REF
16 192. Luxembourg_Loschbour Russia_MA1_HG.SG Chimp.REF
17 188. Luxembourg_Loschbour Iran_GanjDareh_N Chimp.REF
18 0 Luxembourg_Loschbour Luxembourg_Loschbour Chimp.REF
19 212. Russia_MA1_HG.SG Russia_Sidelkino_HG.SG Chimp.REF
20 208. Russia_MA1_HG.SG Russia_Afanasievo Chimp.REF
21 202. Russia_MA1_HG.SG Anatolia_N Chimp.REF
22 201. Russia_MA1_HG.SG Iran_GanjDareh_N Chimp.REF
23 192. Russia_MA1_HG.SG Luxembourg_Loschbour Chimp.REF
24 0 Russia_MA1_HG.SG Russia_MA1_HG.SG Chimp.REF
25 212. Russia_Sidelkino_HG.SG Russia_MA1_HG.SG Chimp.REF
26 211. Russia_Sidelkino_HG.SG Russia_Afanasievo Chimp.REF
27 205. Russia_Sidelkino_HG.SG Anatolia_N Chimp.REF
28 203. Russia_Sidelkino_HG.SG Iran_GanjDareh_N Chimp.REF
29 197. Russia_Sidelkino_HG.SG Luxembourg_Loschbour Chimp.REF
30 0 Russia_Sidelkino_HG.SG Russia_Sidelkino_HG.SG Chimp.REF
f3(C='Chimp.REF', A=c('Russia_Sidelkino_HG.SG', 'Russia_MA1_HG.SG', 'Iran_GanjDareh_N', 'Anatolia_N', 'Luxembourg_Loschbour'), B=c('Russia_Afanasievo','Russia_Sidelkino_HG.SG', 'Russia_MA1_HG.SG', 'Iran_GanjDareh_N', 'Anatolia_N', 'Luxembourg_Loschbour')), data=snps, params=list(outgroupmode="YES"), outdir="f3/")
So how come they got it all wrong? When they first ADMIXTURE or qpadm, the program always picks EHG as the best source for Steppe because EHG is closest to SteppeDNA. Then because EHG shares more drift with Anatolian compared with CHG/Iran_N (relative to Steppe DNA) the second source will be CHG/Iran_N. This is because the SteppeDNA alleles lost via drift going Steppe -> EHG are best found in the population sharing least drift with EHG but still closer to Steppe, which is CHG/Iran_N.
So to test this we can run a qpadm with EHG as a source and another with ANE as a source, and see how the program selects the other component.
target Russia_Sidelkino_HG.SG Iran_GanjDareh_N Anatolia_N
Russia_Afanasievo 0.501 0.434 0.065
target Russia_MA1_HG.SG Iran_GanjDareh_N Anatolia_N
Russia_Afanasievo 0.544 0.235 0.221
qpAdm(target=c("Russia_Afanasievo"), sources=c("Russia_Sidelkino_HG.SG", 'Iran_GanjDareh_N', 'Anatolia_N'), outgroups=c('ONG.SG', 'Kalash', 'Luxembourg_Loschbour','Russia_MA1_HG.SG', 'Russia_Kostenki14.SG', 'Kazakhstan_Eneolithic_Botai.SG', 'Russia_UstBelaya_Angara', 'Israel_Natufian', 'Morocco_Iberomaurusian'),data=snps, outdir="qpadm/")
I know this is not perfect (dates of samples etc) but I am sure this will work even if I use Botai instead of MA1. You can see how using MA1 the program now picks a lot of Anatolian. So we can model SteppeDNA with multiple different sources actually, Anatolian, CHG, Iran_N, EHG, MA1. Some runs will be better than others. But how the others sources are picked when we change the first from EHG to MA1 really tells us whats going on and why academics have it the way they do.
This is pretty crazy but I do believe in it. I've been looking at aDNA stuff for quite a few years and so I am quite competent at analysing data. I mostly use fstats and pcas for analysis.
It's really profound, the academics have most things back-to-front, the wrong way round.
They are confusing drift with admixture.
Steppe DNA is not a result of admixture of different sources (EHG/ANE and CHG/Iran_N) but is the ORIGIN of most Eurasian DNA.
So if you imagine that Steppe DNA consists of a large number of Nomadic peoples, they are moving around and settling in different places over time. So Steppe DNA is like the root of a tree of Eurasian DNA, then Iran_N, CHG, Anatolian, WHG, EHG, ANE they are all branches separating off Steppe DNA. When these samples are placed in ADMIXTURE it gives a unique DNA profile to the most drifted samples (EHG, CHG etc) and gives SteppeDNA a combination of some of them. So scientists are incorrectly interpreting this as SteppeDNA being a mixture of EHG and CHG/Iran_N but really EHG and Iran_N are derived from SteppeDNA through drift, and EHG and Iran_N dont share alleles not shared with SteppeDNA because they are going in opposite geographic relative to SteppeDNA, so it is incorrectly interpreted that SteppeDNA is a mix of the two.
People will say that SteppeDNA is much newer than Iran_N and CHG and EHG so this cannot be true. But we simply havent sampled SteppeDNA earlier than the Eneolithic, but it would have existed earlier. The ancestor of SteppeDNA living in 7,000BC or whatever would not have been so different to SteppeDNA that we have sampled.
There are various arguments I can make for this:
1. The mixing event with EHG and CHG/Iran_N has not been found in the steppe, and will never be found. CHG that could have mixed with EHG will never be found in the PC steppe.
2. It cannot be determined with certainty whether the Southern component was CHG or Iran_N. This is because both of these have inherited SteppeDNA instead of being a source.
3. Steppe DNA is everywhere. Anatolia_N has too much and TepeHissar has way too much.
But this is the proof.
Anatolian_N is too close genetically to SteppeDNA for the current model. Steppe DNA is too close to every Eurasian group to be anything other than Central in the phylogenetic tree.
f3 A B C
1 207. Anatolia_N Russia_Afanasievo Chimp.REF
2 205. Anatolia_N Russia_Sidelkino_HG.SG Chimp.REF
3 204. Anatolia_N Iran_GanjDareh_N Chimp.REF
4 202. Anatolia_N Russia_MA1_HG.SG Chimp.REF
5 194. Anatolia_N Luxembourg_Loschbour Chimp.REF
6 0 Anatolia_N Anatolia_N Chimp.REF
7 206. Iran_GanjDareh_N Russia_Afanasievo Chimp.REF
8 204. Iran_GanjDareh_N Anatolia_N Chimp.REF
9 203. Iran_GanjDareh_N Russia_Sidelkino_HG.SG Chimp.REF
10 201. Iran_GanjDareh_N Russia_MA1_HG.SG Chimp.REF
11 188. Iran_GanjDareh_N Luxembourg_Loschbour Chimp.REF
12 0 Iran_GanjDareh_N Iran_GanjDareh_N Chimp.REF
13 197. Luxembourg_Loschbour Russia_Sidelkino_HG.SG Chimp.REF
14 195. Luxembourg_Loschbour Russia_Afanasievo Chimp.REF
15 194. Luxembourg_Loschbour Anatolia_N Chimp.REF
16 192. Luxembourg_Loschbour Russia_MA1_HG.SG Chimp.REF
17 188. Luxembourg_Loschbour Iran_GanjDareh_N Chimp.REF
18 0 Luxembourg_Loschbour Luxembourg_Loschbour Chimp.REF
19 212. Russia_MA1_HG.SG Russia_Sidelkino_HG.SG Chimp.REF
20 208. Russia_MA1_HG.SG Russia_Afanasievo Chimp.REF
21 202. Russia_MA1_HG.SG Anatolia_N Chimp.REF
22 201. Russia_MA1_HG.SG Iran_GanjDareh_N Chimp.REF
23 192. Russia_MA1_HG.SG Luxembourg_Loschbour Chimp.REF
24 0 Russia_MA1_HG.SG Russia_MA1_HG.SG Chimp.REF
25 212. Russia_Sidelkino_HG.SG Russia_MA1_HG.SG Chimp.REF
26 211. Russia_Sidelkino_HG.SG Russia_Afanasievo Chimp.REF
27 205. Russia_Sidelkino_HG.SG Anatolia_N Chimp.REF
28 203. Russia_Sidelkino_HG.SG Iran_GanjDareh_N Chimp.REF
29 197. Russia_Sidelkino_HG.SG Luxembourg_Loschbour Chimp.REF
30 0 Russia_Sidelkino_HG.SG Russia_Sidelkino_HG.SG Chimp.REF
f3(C='Chimp.REF', A=c('Russia_Sidelkino_HG.SG', 'Russia_MA1_HG.SG', 'Iran_GanjDareh_N', 'Anatolia_N', 'Luxembourg_Loschbour'), B=c('Russia_Afanasievo','Russia_Sidelkino_HG.SG', 'Russia_MA1_HG.SG', 'Iran_GanjDareh_N', 'Anatolia_N', 'Luxembourg_Loschbour')), data=snps, params=list(outgroupmode="YES"), outdir="f3/")
So how come they got it all wrong? When they first ADMIXTURE or qpadm, the program always picks EHG as the best source for Steppe because EHG is closest to SteppeDNA. Then because EHG shares more drift with Anatolian compared with CHG/Iran_N (relative to Steppe DNA) the second source will be CHG/Iran_N. This is because the SteppeDNA alleles lost via drift going Steppe -> EHG are best found in the population sharing least drift with EHG but still closer to Steppe, which is CHG/Iran_N.
So to test this we can run a qpadm with EHG as a source and another with ANE as a source, and see how the program selects the other component.
target Russia_Sidelkino_HG.SG Iran_GanjDareh_N Anatolia_N
Russia_Afanasievo 0.501 0.434 0.065
target Russia_MA1_HG.SG Iran_GanjDareh_N Anatolia_N
Russia_Afanasievo 0.544 0.235 0.221
qpAdm(target=c("Russia_Afanasievo"), sources=c("Russia_Sidelkino_HG.SG", 'Iran_GanjDareh_N', 'Anatolia_N'), outgroups=c('ONG.SG', 'Kalash', 'Luxembourg_Loschbour','Russia_MA1_HG.SG', 'Russia_Kostenki14.SG', 'Kazakhstan_Eneolithic_Botai.SG', 'Russia_UstBelaya_Angara', 'Israel_Natufian', 'Morocco_Iberomaurusian'),data=snps, outdir="qpadm/")
I know this is not perfect (dates of samples etc) but I am sure this will work even if I use Botai instead of MA1. You can see how using MA1 the program now picks a lot of Anatolian. So we can model SteppeDNA with multiple different sources actually, Anatolian, CHG, Iran_N, EHG, MA1. Some runs will be better than others. But how the others sources are picked when we change the first from EHG to MA1 really tells us whats going on and why academics have it the way they do.