search this blog

Monday, June 22, 2015

Oase 1: An early modern human from Romania with a recent Neanderthal ancestor


The Y-chromosome belongs to macrohaplogroup F and the mtDNA to macrohaplogroup N. For details see the supp info PDF here.

Abstract: Neanderthals are thought to have disappeared in Europe approximately 39,000–41,000 years ago but they have contributed 1–3% of the DNA of present-day people in Eurasia1. Here we analyse DNA from a 37,000–42,000-year-old2 modern human from Peştera cu Oase, Romania. Although the specimen contains small amounts of human DNA, we use an enrichment strategy to isolate sites that are informative about its relationship to Neanderthals and present-day humans. We find that on the order of 6–9% of the genome of the Oase individual is derived from Neanderthals, more than any other modern human sequenced to date. Three chromosomal segments of Neanderthal ancestry are over 50 centimorgans in size, indicating that this individual had a Neanderthal ancestor as recently as four to six generations back. However, the Oase individual does not share more alleles with later Europeans than with East Asians, suggesting that the Oase population did not contribute substantially to later humans in Europe.

Citation...

Qiaomei Fu et al., An early modern human from Romania with a recent Neanderthal ancestor, Nature (2015) doi:10.1038/nature14558

Update 05/07/2015: This is how Oase 1 comes out in the Eurogenes K15. Forcing ancient genomes into modern variation like this isn't the ideal way to analyze them, but in this case the relatively low affinity of Oase 1 to present-day Europe and the Near East makes sense.

North_Sea 8.12
Atlantic 13.45
Baltic 9.65
Eastern_Euro 0
West_Med 4.51
West_Asian 0.4
East_Med 0
Red_Sea 0
South_Asian 26.1
Southeast_Asian 7.65
Siberian 0
Amerindian 0
Oceanian 9.36
Northeast_African 4.16
Sub-Saharan 16.6

36 comments:

Mike Thomas said...

F !
Wasn't my guess, but I think others got it !
Let's read...

Mike Thomas said...

Maybe I'm missing it, but I don't think its free ?!

Davidski said...

Give it time to load...

http://www.nature.com/articles/nature14558.epdf?referrer_access_token=eBNwzv2b44jxLSRkDN7gN9RgN0jAjWel9jnR3ZoTv0MjYp5MQz8mXGCmEqqZEWLuPVa54FDKgGTfdqyW2mt9g4eGfhFHdwEbJSmri7eCRb1eQpzSNPlysVZVMFZRwxMnCNSTEkAZU9Vi2vz7qY7q4Wkjjr7ePp5TrM4r22ZaA4RSQOuwj9ty8dwAU7m_BKq5FeEvjDTN-INTld6if5NjwF_dm3wYyKA5zkN_7ohqms4nWKnELkDUWS07-OKcCj_6Ggh8tI4A2QcaREBFbU8VIQ%3D%3D&tracking_referrer=www.nature.com

Mike Thomas said...

Yep I see
All the juice is in the suppl PDF anyway

I know it's very ancient, Dave any chance of a K15 (for history's sake) ?

terryt said...

"The Y-chromosome belongs to macrohaplogroup F and the mtDNA to macrohaplogroup N".

Very interesting, but makes sense. Is anyone able to narrow it down further?


Matt said...

It's interesting to know that Neanderthal admixture happened, and confirms the value of the old craniometry, although Oase1 seems to have left only statistically indistinguishable echoes in even Mesolithic Europeans if any.

From the point of view of later genetics and not being too concerned about Oase1 I guess this paper adds to the Basal Eurasian saga, where again K14 is as related to the Oase1 as East Asians and European HG, like it is to Ust Ishim, despite K14 being less related to East Asians than other pre-Neolithic Europeans are.

Ebizur said...

According to Extended Data Figure 1, the mtDNA of the Oase specimen appears to be an extremely basal (even more basal than Australian aboriginal) sort of N when only "deaminated"/"damaged" fragments are considered, while it appears to be a European sort of R (H39) when "all" fragments are considered. The authors have claimed that the phylogenetic position of the Oase 1 mtDNA consensus sequence (with "all" fragments considered) is attributable to contamination from a recent European human; supposedly, around two thirds of the mtDNA fragments may be attributed to modern European contamination, whereas only 16.7%-29.9% (95% CI) of the nuclear DNA may be attributed to modern European contamination.

Curiously, they now have placed the mtDNA of the Ust'-Ishim specimen on what appears to be a pre-B branch.

I note again the apparent difficulty of resolving the polytomy of L3-M-N; their tree (like several other recently published phylogenetic trees of human mtDNA) has M and (what appears to be) L3 grouped together and opposed to N, but the bootstrap values in this area of the tree are low. Does this imply that the founding population of Out-of-Africa modern humans must have had a large enough female effective population size to maintain several different proto-L3'M'N variants that have suffered a few back-and-forth mutations resulting in unresolvable reticulation?

The assignment of the mtDNA to basal N seems a bit tenuous to me; a majority of the mtDNA fragments seem to point to a sort of H, but the authors have written that fact off as a result of contamination from a recent European, and have claimed that an excess of contamination for mtDNA over the amount of contamination estimated for nuclear DNA is "not necessarily a contradiction."

Ebizur said...

As for the Y-chromosome:

"The Y chromosome of Oase 1

Data processing and sex determination

Panels 1-3 included targets on both chromosome X and Y, so we could use these fragments to determine the sex of Oase 1. Because of the evidence of contamination documented in Supplementary Note 1, we restricted to deaminated fragments (Methods)." (emphasis mine)

"We determined the Y-haplogroup based on the ISOGG database version 10.14, which gives haplogroup assignments for a subset of SNPs on the Y chromosome (http://www.isogg.org/tree). There are 754 SNPs (out of the 15,102 in this version of the ISOGG database) that are covered at least once in Oase 1. We used these to determine the position of the Oase 1 Y chromosome in the tree based on where it carried the derived or ancestral allele. This allowed us to define Oase 1 as belonging to macrohaplogroup F (positions given in hg19 coordinates).

Assignment to F:
P187 (9108252 G→T); P158 (17493513 C→T)

Assignment to CF (which contains CT):
CTS6376 (16863259 C→G)

Assignment to CT (which contains F and CF):
PF38 (3396403 C→T); M5612 (7782393 C→T); M5631 (8396636 G→A); M5632 (8526565 G→A); Z17706 (9989244 G→T); Y1525 (14074463 C→T); L957 (14079528 C→T); Y1526 (14472971 C→T); CTS3662 (15097073 G→A); CTS8542 (18077583 T→C); M5760 (18974195 C→T); L1480 (19212465 A→G); M5783 (21429988 A→G); M5786 (21650381 A→G); Z17721 (22477665 G→C); M5809 (23090404 G→A); M5812 (23105586 C→A); M5823 (23567930 C→T)

We found no evidence that Oase 1 belongs to specific sub-haplogroups of F, as it carries the ancestral allele at all previously described diagnostic mutations for these haplogroups. However, we cannot rule out the possibility that Oase 1 is derived at other SNPs that are diagnostic for these sub-haplogroups but for which Oase 1 has no coverage.

No evidence of membership in G:
S8863 (4179056: G→A); M3485 (8563874 C→T); M3486 (8600158 A→T); L154 (8614138 T→G); M3496 (9850420 C→A); M3497 (9850423 C→A); Z3248 (13460729 G→A); CTS5317 (16203361 G→C); Z3400 (18744995 T→C); M3569 (18744996 C→T); PF3083 (22272581 T→C); PF3087 (22472842 A→C); CTS10945 (22848965 A→G)

No evidence of membership in H:
Z13965 (24523481 C→G)

No evidence of membership in IJ:
P127 (8590752 C→T); PF3526 (8590752 C→T)

We could not test whether Oase 1 is part of macrohaplogroups GHIJK, HIJK, K or K(xLT), as none of the SNPs diagnostic for them are covered by deaminated fragments in Oase 1."

So at the current stage, the Y-DNA of Oase 1 appears to belong to haplogroup F, but there is not yet sufficient information to label it as F(xK).

Grey said...

Matt

"although Oase1 seems to have left only statistically indistinguishable echoes in even Mesolithic Europeans if any."

Yes but it might only be a handful of genes that made an archaic better suited to a particular environment e.g. the high altitude genes in Tibet, so it seems to me the crossover of a handful of archaic genes (or in some cases maybe even just one) might have a disproportionate effect if they solved a particular adaptation problem in a particular region.

(I wonder about this with ANE and admixture software i.e. whether the software weights population clusters by quantity of dna times divergence so a very small amount of very divergent dna can still create a strong signal.)

Maju said...

@Ebizur: thank you very much for your detailed synthesis. There's a minor error: "CF (which contains CT)" should read "CF (which contains F)" but otherwise it is tremendously informative. I know I've told you this many times before but I have to insist: you should write your own blog.

Anyways, so we are before an early UP (probably Aurignacian or similar) European who does not just have local Neanderthal admixture but who could also be carrying modern-like European haplogroups (just that we cannot be sure because of poor DNA quality). That's fascinating even if we are left with the mystery of not knowing for sure.

"Does this imply that the founding population of Out-of-Africa modern humans must have had a large enough female effective population size to maintain several different proto-L3'M'N variants that have suffered a few back-and-forth mutations resulting in unresolvable reticulation?"

I don't think this can be the case: there is a lot of literature on mtDNA phylogeny and, when using coding region markers, there is absolutely no doubt about it: M and N are clearly downstream of L3, by several coding region SNPs (3 for M, 5 for N), i.e. the founder OoA population carried only (that we can discern) L3, which eventually evolved into to M and N (they may have also carried other L but does not seem to have got beyond Arabia, so pretty much pointless). Did they use coding region or just limited themselves to HVS-I, which is not sufficiently informative in most cases? Or are the problems caused by DNA degradation and possible contamination? I don't really understand why you raise this question.

G Horvat said...

With regards to the mtDNA, I don't think we have been provided with all of the mutations and so I can not determine which classification it should be but it doesn't look like it belongs to any H, R or N clusters.

From the article:
"Oase 1 carries the following substitutions that define the N macrohaplogroup: 73G, 263G, 750G, 1438G, 2706G, 3107d, 4769G, 7028T, 8860G, 11719A, 12705T, 14766T, 15326G, 16223T"

It has 14 differences from the CRS, a haplogroup H haplotype, then. Based upon these differences from it, the sequence does not belong to H and based upon the 12705T and 16223T, it does not belong to any R haplogroups.

"Oase 1 does not share derived alleles at positions 8701 (G in Oase 1) and 9540 (C in Oase 1)"

8701G and 9540C are the original forms of these nucleotides. They are found in the M & L sequences. If the M-specific mutations are missing, then they should look at the possibility that the sequence belongs to L3.

Matt said...

@ Grey, that's possible, especially if European ADMIXTURE went something like

Oase1->Oase1+Kostenki_hybrid->(Oase1+Kostenki_hybrid)+SomethingElse_Hybrid
etc. which is optimal for "introgression".

But with the ultra small populations of ancient Eurasia, when AMH had small populations generally and particularly for Eurasia maybe weren't as well adapted to it as they would become in the future, and where Europe is itself particularly probably a bit more prone to population crashes, I wouldn't count on it.

Re: the small amount of very divergent and ANE, I'm assuming this is to do with your "yeti" (e.g. robust / tall / cold adapted divergent archaic humans) thing?

Looking sympathetically at this idea, archaic hominin introgression is not a priori wrong, as we now know. The thing with looking at ANE as having a very small amount of very divergent dna is that a) ANE is definitely in a clade with WHG branch, b) it seems no further from ENA than WHG is, by the phylogeny, which would be strange if it was on the WHG branch plus extra archaic c) populations with extra archaic admixture like the Papuans seem to get a pretty clear signal of admixture from a population that is basal (prior) to the African->Eurasian split in various models (which ends up visually falls close to the base of the chimpanzee outgroup on e.g. treemix). So this all makes me extremely skeptical.

Grey said...

Matt

"I wouldn't count on it."

Sure it's just an idea, the main point is simply that a very small amount of surviving introgression doesn't mean it couldn't have been critical - like the Tibet high altitude genes for example. I assume those genes only make up a tiny percentage of Tibetan dna and yet without it they wouldn't be living there at all.

capra internetensis said...

I don't see a problem with the mtDNA assignment. The mt hg based on all fragments clusters deep in modern European DNA on a long branch, which makes no sense if it isn't contamination. And contamination is expected.

Using deaminated fragments the result is pre-N, and why shouldn't an archaic dude with no living descendents carry pre-N?

I'm not sure what if anything this tells us about the history of N.

Maju said...

@G Horvat: "based upon the 12705T and 16223T, it does not belong to any R haplogroups".

Absolutely! So far L(xR).

"8701G and 9540C are the original forms of these nucleotides".

True again. So L(xN).

As you say well, it could be L3(xM,N) because I see nowhere any of the M-defining mutations (T489C, C10400T, T14783C, G15043A). They could still be in the missing or "contaminated" segments, I can't judge so much. I also do not see any of the L3 root mutations listed anywhere.

I also do not understand why they still work with CRS as reference instead of the HRS one (i.e. "mitochondrial Eve"), it makes it all much more difficult to parse, really.

G Horvat said...

@Maju
"I also do not understand why they still work with CRS as reference instead of the HRS one (i.e. "mitochondrial Eve"), it makes it all much more difficult to parse, really."

That is indeed the root of the problem.

terryt said...

"Curiously, they now have placed the mtDNA of the Ust'-Ishim specimen on what appears to be a pre-B branch".

An eastern branch. That fits perfectly when we consider Y-DNA K2a* too as being eastern.

"the mtDNA of the Oase specimen appears to be an extremely basal (even more basal than Australian aboriginal) sort of N when only 'deaminated'/'damaged' fragments are considered, while it appears to be a European sort of R (H39) when 'all' fragments are considered".

An extinct N line actually makes more sense to me. There is plenty of western basal N whereas R is more likely to be more recent in the west than is N.

"This allowed us to define Oase 1 as belonging to macrohaplogroup F (positions given in hg19 coordinates)".

An extinct subgroup of F makes complete sense as, in spite of what many people believe, I think it is doubtful that the haplotype expanded from South Asia. Although G, H and IJ have been eliminated as candidates the three seem primarily Anatolian or Iranian and so some relation could easily have reach Romania by 37-42 kya.

"If the M-specific mutations are missing, then they should look at the possibility that the sequence belongs to L3".

From the information Ebizur provided it seems the mt-DNA line possibly branched from N before it was fully formed. In other words it is partway between L3 and N. As Capra says:

"Using deaminated fragments the result is pre-N, and why shouldn't an archaic dude with no living descendents carry pre-N?"

Exactly.

"it seems to me the crossover of a handful of archaic genes (or in some cases maybe even just one) might have a disproportionate effect if they solved a particular adaptation problem in a particular region".

I suspect it is likely to be just one more random hybrid formation. However its presence does show that hybrids must have been able to have offspring reasonably easily.

Mike Thomas said...

It's really interesting
So the two ealey specimens (Oase and Kostenki) and no Y Hg I yet.

? Arrived with "Gravettians"?

terryt said...

"? Arrived with "Gravettians"?"

Could be. I is fairly well downstream. It is post IJK and post IJ. That could mean it developed reasonably late. By the way Mike. One of your comments here appeatred in my emails but I can't find it here. It was just confirming pre-F and pre-N haplogroups.

Mike Thomas said...

Yes Terry
I deleted it . It was a reply to your comments "is anyone able to narrow this down ". But I realized that my reply was tangential

Nirjhar007 said...

David, Check This Shit-
https://genetiker.wordpress.com/2015/06/24/analyses-of-the-oase-1-genome/#comments

terryt said...

"David, Check This Shit-"

Quote:

"The Y-SNP calls for Oase 1 show that he belonged to Y haplogroup K*. He had a positive call for the SNP PF3500 at position 21571895, which is at the level of IJK in the Y chromosome tree. He also had a positive call for the SNP PF5495 at position 15842844, which is at the level of K. He had negative calls for SNPs defining haplogroups LT, P, N, and O".

Doesn't fit the information Ebizur provided above.

"We found no evidence that Oase 1 belongs to specific sub-haplogroups of F, as it carries the ancestral allele at all previously described diagnostic mutations for these haplogroups. However, we cannot rule out the possibility that Oase 1 is derived at other SNPs that are diagnostic for these sub-haplogroups but for which Oase 1 has no coverage".

andrew said...

"It's really interesting
So the two ealey specimens (Oase and Kostenki) and no Y Hg I yet.

? Arrived with "Gravettians"?"

My money would be on Y-DNA I1 and mtDNA U5 emerging as the dominant hgs of the Franco-Cantrabrian refugium, and then mostly beating other LGM survivors in the race to repopulate Europe.

capra internetensis said...

@Terry

Fu et al only looked at SNPs listed on the ISOGG tree version 10.14 (from February of this year). There are quite a lot of known SNPs that have not made into onto ISOGG's tree yet, either because they don't meet ISOGG's specifications or simply because they are swamped by the sheer workload.

Mike Thomas said...

Andrew

Not I1. Goes against the evidence

Terry / Capra

So if Genetiker is right, I think he's implying Oase was haplgroup K ?!

terryt said...

"So if Genetiker is right, I think he's implying Oase was haplgroup K ?!"

That's what I understand to his view.

capra internetensis said...

@Mike

I guess Andrew meant I2.

Yeah, it looks like he is K* - could be a false positive, but I doubt it (K is a short branch and IJK is very short branch, so it would be surprising to get false positives on both by chance). He is negative for a number of LT SNPs and P SNPs, so he doesn't belong to either of those branches (unless at the very base), but he isn't negative for any NO SNPs - actually in Genetiker's list he only has a call for one NO SNP, and it is positive, but since it's a recurrent mutation it doesn't have much value.

Since Ust' Ishim man was K2a (i.e. pre-NO), it is not too shocking to find Oase-1 was also K (and possibly even K2a as well), though I must admit I was not expecting it.

Mike Thomas said...

Yes I'm very surprised I have to say
Seems like Terry and others might be right about the early routes taken ...
But I'm not sure How to take genetiker's results. He seems controversial, and I'm not sure how he is getting the resilution others arent

capra internetensis said...

@Mike Thomas

Genetiker is the only one who is actually going over the Y chromosome data as far as I can tell, and he is using more SNPs than Fu et al did. He has some iffy pet theories but as far as I know he is honest and does solid work.

Does anyone know anything about the relevant archaeology, BTW? I have looked into it a little but there seems to be a bewildering array of cultures with little agreement on how they are related - what is or isn't Aurignacian, what was made by modern humans vs Neanderthals, where the Upper Paleolithic came from, etc.

Maju said...

The relevant archaeology of Oase 1 is that there is no relevant archaeology, i.e. it's a remain without cultural context. However from the age, it should be Aurignacian (full Aurignacian, not merely "Aurignacoid" - although there were still pockets of Mousterian-using Neanderthals for example in Croatia).

Mike Thomas said...

Thanks Capra

Yep, Majus xomments are correct.

Davidski said...

Genetiker's analyses of genome-wide data are at the level of a 12-year-old.

If that's anything to go by, then it's safer just to ignore everything else he does.

And no, he's not the only one going through the Y-chromosome data.

terryt said...

"Genetiker's analyses of genome-wide data are at the level of a 12-year-old".

Perhaps a bit strong, but I admit I had doubts about K. To go back to Ebizurs' comment:

"We found no evidence that Oase 1 belongs to specific sub-haplogroups of F, as it carries the ancestral allele at all previously described diagnostic mutations for these haplogroups".

But he did add a rider:

"However, we cannot rule out the possibility that Oase 1 is derived at other SNPs that are diagnostic for these sub-haplogroups but for which Oase 1 has no coverage".

As a result I was prepared to consider Genetiker's idea.

"Since Ust' Ishim man was K2a (i.e. pre-NO), it is not too shocking to find Oase-1 was also K (and possibly even K2a as well), though I must admit I was not expecting it".

To say the least. If it is the same as Ust-Ishim Y-DNA it would require a huge expansion of K2a, all the way from East Asia to western Europe, followed by a complete collapse and disappearance of the haplogroup.

capra internetensis said...

@Davidski

Is he calling Y SNPs wrong? That is what we are talking about.

If someone else is doing it, and they are reaching different conclusions, then I hope they will let us know.

Mike Thomas said...
This comment has been removed by the author.
arch said...

Ebizur-"the mtDNA of the Oase specimen appears to be an extremely basal sort of N. The assignment of the mtDNA to basal N seems a bit tenuous to me; a majority of the mtDNA fragments seem to point to a sort of H"

Well its pretty simple, it cant be both MTdna Hg from the same individual without contamination, so the deaminated recovered DNA Hg which is non-extant in modern humans, and consistent with a ancient sample is logical. modern 'H' that is carried by half the modern population of Europe and KNOWN handlers of OASE 1 remains is not logically ancient material.

I think the real problem is that they are using these processes for a purpose not actually fully suited, and it results in what I think is a skewed estimate. Below is excerpted what appears to me to be the work that the estimates proffered for OASE 1 are actually derived from. In that paper you note a ancient sample, Otzi, has only a pittance of its genome that would not be discarded from further proccesses under the methods used for OASE 1.

"For example, only 3.5% of DNA sequences obtained from Ötzi, a 5300-yr-old mummy found in glacier ice in the Alps, show deamination-induced C to T substitutions at their 5′ end (Keller et al. 2012)

Since uracil originates from cytosine, U selection slightly increases the sequence representation of GC-rich regions of the genome. However, since single-stranded library preparation leads to an overrepresentation of AT-rich sequences, the overall GC-content{}moves closer to the genome average. More data will be needed to evaluate in detail .."

What this work DOES and is well suited for is screening out contamination. What its being applied to do is something much different in the case of OASE 1, which is to give genome-wide percentages of alleged 'neandertal' genetic ingress. The reason I dont believe this process is genuinely applicable to that purpose are obvious. I think that they are targeting specific 'Neandertal' rich sites, tossing out material that does not meet a artificial screening criteria, which as in the case with Otzi may be in fact be ancient yet non-deaminated, and this process which was created to rigourously screen out contamination throws major discrepancies when used for other purposes.

The biggest hole I immediately note in this paper is its own admission that among S/S African Dinka, who the authors cite as having "little or NO 'Neandertal' ancestry", they still find 485 alleles matching what they term 'Neandertal-like' alleles. For Europeans they find 1,033 alleles, and for Chinese they find 1,322 'neandertal-like' alleles. IF the Dinka S/S African population has "Little or NO Neandertal ancestry", yet shares nearly 50% of the ancestry that Europeans do, its a very telling observation to me that overturns almost all the conclusions.

The Dinka dont share 50% of the Euro pct of Neandertal-like alleles by mere random mutation, the amount is simply too high. THIS IS ACTUALLY THE SMOKING GUN OF THIS STUDY. What this indicates to me is what we have already long known - S/S Africans share the same ancestral material from a common ancestor along with the rest of humanity, however they separated from other populations earlier, and all other groups are more closely genetically related to one another.

Euros and Asians and Oceania dont share MOST of what is being termed 'neandertal-like' segments in common due to grand-pappy neandertals, they share them because this is ancestral to a common ancestor that is closer in time between them than it is to the Dinka, and other S/S Africans. MOST of what is being termed "neandertal-like" in this paper does not have anything to do with Neandertal ingress in my observation, and it never did. The authors note that they toss out Dinka Alleles and subtract them from consideration in other populations when these "Neandertal-like" alleles are found across all populations (in considerable numbers no less)