Avaria

https://en.wiktionary.org/wiki/avaria#Italian

Some time ago one of the Avar period samples from Hungary reported initially as a female (V37.2: released Feb 2019):

AV1	AV1	AV1	petrous	..	..	1240K	AmorimNatureCommunications2018	1361	541-637 calCE (1487±26 BP)	Hungary_Avar	Szólád	Hungary	46.28333333	17.85	F	X2m'n	1472±27	0.325459145	2.76426	802843	half	All	PASS (literature)	..	..	0.325459145	0.0874	X2m'n	0.98	2018

Gained male Y-chromosome (V42.4: released Mar 2020):

5697	AV1	AV1	AV1	petrous	2018	AmorimNatureCommunications2018	AmorimNatureCommunications2018	1361	541-637 calCE (1487±26 BP)	Hungary_AvarPeriod	Szólád	Hungary	46.28333333	17.85	1240K	..	2.573859	802843	U	R1b1a1a2a1	..	X2m'n	..	..	..	..	..	..	..	half	..	PASS (literature)

This erroneous information possibly have come from the authors themselves as even folks from the Reich's Lab were bamboozled for a while (although not completely, as the sex of the sample was changed to Unknown). Fortunately since then they've greatly improved their process of assessing uniparental markers and current version of their dataset was released with corrected metadata ([Wed Jan 20 22:51:34 EST 2021]: V44.3 release):

11218	AV1	AV1	AV1	petrous	2018	AmorimNatureCommunications2018	AmorimNatureCommunications2018	Direct (WARNING MISSING LAB CODE): 95.4%; IntCal20, OxCal v4.4.2 Bronk Ramsey (2020); r:5; Atmospheric data from Reimer et al (2020)	1360	25	549-640 calCE (1487±26 BP)	Hungary_Avar_5	Szólád	Hungary	46.28333333	17.85	1240K	..	2.573859	802843	..	F	Hungary, Szólád Family A (2 members) (AV2-AV1 have a mother-daughter relationship)	..	10.47	X2m'n	..	Haplofind	..	0.0874	..	n/a (female)	n/a (female)	n/a (female)	n/a (female)	-0.015	0.006	None	[0,0]	ds.half	..	PASS (literature)

Could we have somehow avoided these shenanigans?

Below you can see a full (sic!) output from Yleaf:

chr	pos	marker_name	haplogroup	mutation	anc	der	reads	called_perc	called_base	state
chry	14730165	FGC26351	A00	T->A	T	A	1	100	T	A
chry	6964065	M6857	B2b1b1b~	A->T	A	T	1	100	A	A
chry	18581720	CTS8960	D1a2a3a1a	A->G	A	G	1	100	A	A
chry	22770207	CTS10774	E1b1a1a1a1c1a1a3a1d	G->A	G	A	1	100	G	A
chry	22770232	CTS10775	E1b1a1~	G->T	G	T	1	100	G	A
chry	7873617	Y6168	E1b1b1b2a1a1a1a1e1~	C->A	C	A	1	100	C	A
chry	8272388	Z40155	G2a1a1b2	C->T	C	T	1	100	C	A
chry	8272382	FT40739	G2a2b1a1a1a2b2~	G->A	G	A	1	100	G	A
chry	23987375	BY200867	G2a2b2a1a1b1a1e2a	T->A	T	A	1	100	T	A
chry	17544312	Z13554	H3	G->T	G	T	2	100	G	A
chry	16202490	L623	I2a1b1a2b2	A->T	A	T	1	100	A	A
chry	13960753	Z28245	J1a2a1a2d2b2b1a~	G->T	G	T	1	100	G	A
chry	16202478	ZS6620	J1a2a1b1a1~	A->G	A	G	1	100	A	A
chry	9990070	Z8257	J2b2a~	C->G	C	G	2	100	C	A
chry	17411756	AM01369	J2b2a~	C->T	C	T	1	100	C	A
chry	14858930	Z42488	L1a1b3a1a1a~	G->A	G	A	1	100	G	A
chry	22567345	M2106	N1a1a1a1a	C->G	C	G	1	100	C	A
chry	14513977	YP5780	N1a2b2b	G->A	G	A	1	100	G	A
chry	9953214	Z25708	O2a2b2a1	G->A	G	A	1	100	G	A
chry	16202425	F2086	Q1a1a	T->C	T	C	1	100	T	A
chry	17544259	YP819	Q1a2a1~	T->G	T	G	2	100	T	A
chry	8502236	L51	R1b1a1b1a	G->A	G	A	1	100	A	D
chry	21154383	S22641	R1b1a1b1a1a1c3	G->A	G	A	1	100	G	A
chry	13204207	BY15390	R1b1a2	T->A	T	A	1	100	T	A

The proportion of reads mapped to X and Y chromosomes leaves no illusions (Skoglund et al., 2013):

chrX	289426	3.86%
chrY	1732	0.02%

If checking this was so easy, why didn't all these genetic experts out there do this instead of babbling about some "Celto-Germanic father"? Were they trying to deceive you, dear reader?

I hope that you now know who not to trust.

Comments

ambron said…
Arza, I must admit that I don't understand anything anymore... Was or wasn't Av1 R1b? Was Av1 a girl or a boy?
Arza said…
It's a female. Daughter or mother of Av2. And it always was... I pointed this out a long time ago.
ambron said…
I understood correctly, but preferred to be sure.
ambron said…
It would be good to throw into this thread the Av1 father's PCA, on which he groups with modern Hungary.
Arza said…
https://i.postimg.cc/LXVcv784/avaria.png

Green is more probable because it better fits the whole G25 dataset, but who knows.
ambron said…
So if someone claims that Western Slavs are a mixture of individuals such as Av2 and the father of Av1, then he must realize that it is a mix of Masovians and Hungarians.
Arza said…
This is how such extrapolated "Celto-Germanic" father would look like:

Target: Av0_Pannonian
Distance: 1.9282% / 0.01928151 | R3P
52.0 Polish
24.2 French_Occitanie
23.8 Greek_Peloponnese

Target: Av0_Pannonian
Distance: 1.6694% / 0.01669411
15.8 Lithuanian_VZ
15.0 French_Occitanie
14.2 Lithuanian_SZ
12.0 Greek_Central_Macedonia
11.4 Polish
8.6 Dutch
7.0 Basque_French
6.6 Serbian
4.6 BedouinB
1.8 Greek_Peloponnese
1.2 Irish
1.0 Spanish_Galicia
0.8 Yemenite_Mahra

Av0_Pannonian,0.132035,0.137097,0.050534,0.033269,0.043393,
0.015338,0.00799,0.011769,0.003068,-0.004556,
0.001136,-0.009291,0.00446,0.009083,-0.009771,
0.014453,-0.016559,-9.99999999999266E-07,-0.003897,0.003752,
-0.004491,-0.002968,0.001972,-0.006387,-0.000958
Ryukendo K said…
Hi Arza,

Thank you for your great work on this blog! I have a couple of questions which--if you haven't already looked into it--I hope you could investigate;
1) Given that Baltic_BA's HG ancestry did not come from the Baltic region, where did it come from? Did you see the recent poster from a conference here: https://twitter.com/GerberDniel2/status/1400010765445353475/photo/1

Using the current HG-rich samples from Hungary, can you show in some way that the HG ancestry in Baltic_BA came from Hungary and surrounds?

2) You showed in the previous article that, after controlling for Baltic_BA and Yamnaya ancestry, the remaining EEF ancestry of Slavs was very Balkan-like. Interestingly, you show this is true also for modern Balts. I think the question of when this Balkan-like EEF signal reached Baltic populations is very interesting, because it may show that Balts reached the region fairly late, after Baltic_BA type ancestry did. The Baltic_BA grouping contain samples from 1200BC to near the 0s AD, and there are also later samples like LTU_BA from the Baltic region in the historical period. Could you check if later samples from the Baltic region show any sign of Balkan EEF introgression compared to the earliest samples in the Baltic_BA cluster?

Much appreciated - Ryukendo
Arza said…
Hi Ryukendo, good to see you here!

I'll answer your questions by tomorrow. I did not include later Baltic samples because to do so I need to account for possible Uralic-related ancestry. BTW Few hours ago I've found out that in the newest .anno file one of the Estonia_BA samples is described as an outlier:

s19_X09_1.SG Estonia_BA_o.SG QUESTIONABLE (literature, popgen.outlier)

Unfortunately numsnps is only 16729. Right now I'm running dstats and so far the only Z-score > 3 is this one:

Chimp.REF Hungary_Maros_EBA.SG Estonia_BA.SG Estonia_BA_o.SG 0.0113 0.00347 3.26 0.00111 6297

A smoking gun, isn't it?

Re: BR2 haplotype sharing

Target: HUN_LBA:I1504
Distance: 1.7473% / 0.01747324
68.8 Sorb_Niederlausitz:Niederlausitz1
31.2 POL_Globular_Amphora:I2433

He's almost exactly between one of the GAC samples and that single Sorb.

He's probably not ancestral to the Slavs. Likely it's the opposite - he has the "Slavic" ancestry plus extra neolithic on top of that.

2/3 of his ancestry is like modern West Slavs - probably that's why the haplotype sharing in Cassidy et. al (but also check ChromoPainter results from the supplement of Margaryan et al.) tops among Poles and drops among the rest of the Balto-Slavs who have different proportions of common ancestral components.

The rest of his ancestry can explain why there is a high haplotype sharing with Wales. It can be mediated via neolithic substrate common to him and e.g. one of the Celtic groups that ended up in Wales.
ambron said…
Is it those Nitran group samples rich in HG and Z280+?
Arza said…
According to the presentation made by Anna Szecsenyi-Nagy Nitra was R1a-rich and similar to CWC (in terms of EEF-Yamnaya-IronGates relatedness). Although in a large sample some outliers with recent and local HG-rich ancestry are rather expected.
Arza said…
@ Ryukendo

Re: 2)

Most of the samples have strong preference for HG, but at the same time they pick rather WHG-poor neolithic pops. And now I'm not sure if I should treat them separately, or as a complex proxy that tries to compensate for different things because something is missing e.g. on the Uralic side.

We can eventually try this approach with Baltic_LTU_BA (because here we don't have to worry about poorly referenced Asian ancestry skewing the model). Turlojiske3 has a nice distance and as an extra neolithic it picks up LBK.

Target: Baltic_LTU_BA:Turlojiske3
Distance: 1.3927% / 0.01392702
54.6 Baltic_LVA_BA
16.0 Baltic_EST_BA
10.2 Corded_Ware_POL_early
9.0 Corded_Ware_DEU
7.8 DEU_LBK_SMH
2.4 Yamnaya_RUS_Samara

Turlojiske1 is more complex. EEF sources are WHG-poor, but in the original model it picks up Iron_Gates. Latvia_HG cannibalizes it completely (assimilation of some HG leftovers?):

Target: Baltic_LTU_BA:Turlojiske1
Distance: 2.1177% / 0.02117722
58.8 Baltic_LVA_BA
11.8 Corded_Ware_DEU
10.0 Yamnaya_RUS_Kalmykia
6.8 Baltic_LVA_HG
5.6 TUR_Barcin_N
3.6 Baltic_EST_BA
2.0 HUN_Tiszapolgar_ECA
1.4 DEU_LBK_HBS

https://i.postimg.cc/bydTHKWr/Baltic-LTU-BA.png
(Iron Gates is not subtracted, but Latvia_HG is, hence the shift)

So it seems that such "southern" ancestry was in the Baltics from the very beginning (of the existence of the Baltic_BA cluster). Although I wouldn't treat this as an ultimate proof, because samples from Turlojiske behave slightly... weird. It's either bad quality or some admixtures from different sides. Or both. Generally I avoid using them in my models.
Arza said…
@ Ryukendo

Re: 1)

Carpathians. Previously I thought that this kind of ancestry survived somewhere north of Ukraine and east of the Baltics, but after the publication of the Volosovo and Fatyanovo samples it became clear that there were just "regular" HGs of the Latvia_MN type.

On the other hand area around Carpathians was (and still is) yielding Baltic_BA-admixed pops one after another. I think that the most important samples are:

BR2 - GAC-BR2 vector points directly at the Balto-Slavic cline. It's unlikely that it's just an accident and probably at least one of the grandparents was on this cline which wouldn't exist without "Baltic_BA". Such population had to be somewhere there. (haplotype sharing is a nice bonus)
_________

Bell_Beaker_HUN_EBA:I3528 (2562-2299 calBCE)

Ancients without VK2020:

Target: Bell_Beaker_HUN_EBA:I3528
Distance: 1.7266% / 0.01726640
19.0 DEU_LBK_SMH
17.2 SRB_Iron_Gates_HG
16.0 Baltic_EST_BA
10.4 ITA_Collegno_MA
6.2 TUR_Alalakh_MLBA
5.0 KAZ_Golden_Horde_Euro
4.8 FRA_Grand_Est_MN
4.6 DEU_Tollense_BA
3.4 FRA_Nouvelle_Aquitaine_Meso
3.2 IRN_Tepe_Hissar_C
3.0 ITA_Sicily_EBA
2.8 DEU_Wartberg_MN
2.4 Baltic_EST_MA
1.2 ROU_C_o
0.8 VUT_2300BP_all

There is enough "drifted" ancestry to reconstruct one of the grandparents as almost pure Baltic_BA (Tollense here is just the WEZ56). Note how early he is.

_______

ROU_C_o:GB (3512-3350 calBCE)

Target: ROU_C_o:GB

Distance: 1.5695% / 0.01569462
33.4 SRB_Iron_Gates_HG
17.6 DEU_Wartberg_MN
15.8 Baltic_EST_BA
9.2 HUN_ALPc_III_MN
8.6 Iberia_Central_CA
7.2 TUR_Pinarbasi_HG
3.6 Baltic_LVA_BA
2.6 HUN_Koros_N_HG
1.6 UKR_Trypillia
0.4 HUN_ALPc_Szakalhat_MN

Nearly 20% of Baltic_BA, despite that this sample rather doesn't have any steppe ancestry, which in turn means that the Baltic_BA acts as a proxy for a hypothetical WHG population with the "Balto-Slavic" drift. Matt on Eurogenes posted once a ghost of this pop that was spot on. When I added it to the mix it replaced Baltic_BA completely. I'll re-run this model if I find these coordinates.

To sum things up:

by 1250 BCE we have a proof of the Balto-Slavic cline developing somewhere in the vicinity of Hungary
by 2400 BCE we have a proof of a population similar to Baltic BA somewhere in the vicinity of Hungary
by 3400 BCE we have a proof of the "Balto-Slavic" drift in a HG-rich sample with zero steppe ancestry in central Romania.
Arza said…
Re: poster

Yeah, I saw it. I also saw the presentation of Anna Szecsenyi-Nagy and I have no doubts how these outliers will look like and what kind of "drift" they'll have.

The only question is if we will see among them any Baltic_BA-like sample... or will we have to wait a little longer for it. At this point finding such specimen is just a matter of time.
ambron said…
Arza, I am reminded of Stanisław's old disputes with Michał about whether the Slavic HG is local or steppe.
Arza said…
@ Ryukendo

After reading your posts on AG I think I misunderstood you. If you were asking for a model where Baltic_BA is the target then it's not possible (and especially not in qpAdm, as besides the proper source we are lacking also a population that we could put in the right pops).
ambron said…
Arza, I argue with Altvred on AG whether the population represented by Wez56 could or could not have contributed to the genesis of Poles. Can this problem be solved with the tools you use?
Arza said…
Proving or disproving this would require tools that look at IBD, rare haplotypes and so on.
ambron said…
Arza, have you made any progress with Krakauer Berg?