PDA

View Full Version : Comparing the tree branching shapes of the main P312 clades (and U106)



alan
05-05-2014, 02:01 PM
What I am interested in is comparing the early generation branching patterns of the main P312 divisions and also U106. Basically a comparison of branching in terms of SNP generations which I understand are under debate in terms of length. Even if we dont know the exact time between SNPs it still interesting to see if there are any contrast between clades. I think this deserves a thread on its own.

I would ask if other posters could link of post to the most up to date detailed trees for those clades. Rich R- can you post your U152 table or better still a link to it here. I have seen a few L21 trees recently but just to be sure can someone post the most detailed up to date one. As for DF27 - I am not sure what the best tree is. I have seen a good tree for U106 but again I am not sure what the best is. If we could get all the most up to date table together in the first few posts in this thread it would form the basis for discussion/

As a starting point I am assuming the actual SNPs L21, U152 and DF27 are roughly contemporary as they are all one SNP down from P312. I am assuming that U106 and P312 still share the L11 ancestor just one SNP upstream. That to means it doesnt make sense in excluding it.

To give an example of the sort of inferences I am looking at is for example it seems to me that U152 significantly branches immediately below that SNP. L21 on the other hand only undergoes significant branching after DF13 - an extra SNP down the chain. This might be interpreted as L21 only expanding perhaps a century later. On the other hand my impression - forgive me if this is a false one - is that L21 had a rather steady growth for a long period while U152 branching looks somewhat different and less constant.

I think we really might learn something by comparing the trees.

razyn
05-05-2014, 02:33 PM
What I am interested in is comparing the early generation branching patterns of the main P312 divisions and also U106.
So, in effect this thread is comparing the tree branching shapes of L11.


As for DF27 - I am not sure what the best tree is.
I'm not, either, and I'm nominally one of the guys who should. I think the best graphic image of it is the one Mike W made for the introductory pages of the R1b-DF27 project, and its parallel Yahoo group. But he hasn't updated it since last summer, and we've continued to learn stuff. Mike has L21 around his neck, so I feel his pain -- but I don't have the knowhow to do images like that. I like Chris Morley's data a little better than Mike's (and a lot better than FTDNA's, or ISOGG's, or YFull's); Morley has updated it several times, but at the moment it's three months old. Jim Wilson's release of Chromo2 data is newer than that, as are almost all results from FGC, and all from Big Y testing. And all of these have made some discoveries (not in the Feb. 7 version of Morley's tree) pertinent to DF27.


As a starting point I am assuming the actual SNPs L21, U152 and DF27 are roughly contemporary as they are all one SNP down from P312. I am assuming that U106 and P312 still share the L11 ancestor just one SNP upstream.

I think the operative word is "roughly," and it could be used more than once in those two sentences. But in principle, I roughly agree. Several of us suspect that there are more of these higher-level SNPs (or different, better "equivalent" choices) that we just haven't yet identified. The more SNPs we find, the more we wonder what we haven't found. And that has some bearing on the SNP-counting method of estimating clade ages.


I think we really might learn something by comparing the trees.

I totally agree.

Wing Genealogist
05-05-2014, 02:40 PM
Preliminary U106 tree can be found online at: https://app.box.com/s/afqsrrnvv2d51msqcz2o

alan
05-05-2014, 03:52 PM
I am a bit poor in knowledge about it but I understand that the approach has been to try and get someone from each major cluster/clade to do the big genome testing projects to find new SNP. I am just guessing but I assume the real problem in getting a full picture is the unclustered people within a clade - unless each one did this kind of testing it wont be clear how they branched. I guess given its expensive that a large number of this very small lines will not have their relationships resolved any time soon. However, I assume this sort of genome testing at least is giving us an idea of some of the first and second generation branching just below the big clade defining SNP - L21, U152, DF27 and U106. Its probably still fair to say that at or just under those SNPs surviving branching happened that contrasts with the very limited branching from L11, L51 etc so there does seem to be some sort of kicker to branching that happened around or just after L21, U152, DF27 and U106. The one thing that is obvious is L21 did not do much until one SNP down under DF13, perhaps indicative of a slight delay of a century or more in L21 finding fertile ground for significant branching.

R.Rocca
05-05-2014, 04:04 PM
A couple of potential pitfalls with comparing trees in this manner:

1. Certain trees are at a more mature level in their definition than others. If razyn is looking at a DF27 tree that does not have Big-Y or Full Genoms data, it is likely not as well defined as the U152 and/or U106 trees.

2. Testing bias. I suspect that when the "Genome of the Netherlands" project finally releases their data, the U106 tree will explode. We saw the same thing happen when Francalacci et al. published their Sardinia paper, and the I2a tree went from just a few downstream branches to hundreds overnight.

razyn
05-05-2014, 04:36 PM
1. I look at a lot of "trees," including Big Y and Full Genomes ones that are currently looking more like combs than trees. But I have high hopes that the actual branching points are getting discovered, gradually.

2. Genomes of the Netherlands will tell us some useful things about DF27, also.

alan
05-05-2014, 05:09 PM
Superficially it looks like U106 branches quite heavily in the first SNP generation after U106. That is interesting in itself as it makes it look like it had a good start with 7 branches. Then about 10 more branches in the next generation. Shows a pretty good start for a couple of SNP generations. Then 9 further branches in the next SNP generation too so still steady. The 3rd SNP generation after U106 seems to see many lines ceasing to branch but a small number start a long period of branching. This could represent a noticeable rise in the hierachical nature of society with elites hogging resources.

Now if it is about 100 years per SNP generation on average what we might be seeing is a quick burst in U106 in the same next SNP generation as the one where P312 split into U152, L21, DF27 and other smaller branches. So, I think it is fair to say they appear to be experiencing a similar expansion at the same sort of time as U152 and DF27 and perhaps a little before L21 given that the latter's branching in minor until a further SNP generation. So I would tend to conclude U106 started well along with the first generation of P312 branching. It seems to have branched steadily in non-spectacularly for the next couple of SNP generations until in the third SNP generation after U106 a lot of lines flatline with a minority starting a long history of steady strong branching. That does sound potentially like a more hierarchical socieity. Lets say the whole branching below L11 does coincide with the start of the beaker expansion period (note I say period and am not making a cultural assignation) which really exploded around 2600BC. I would tend to think that that big step up in beaker expansion probably is best matched by the SNP generation below p312 and U106. So I would tend when using a beaker model to think L11 dates to about 2800BC and P312 and U106 to 2700BC then the big burst commences in the generation where P312 and U106 throw up a number of subbranches around 2600BC. Note that these subbranches have strong geographical patterning so there must have been both demographic and geographical expansion involved which suggests to me that it is indeed a good match for the geographical expansion we see in beaker around this time.

For U106 it does seem to have a similar pattern of a few generations of strong branching that it shares with P312. While the lineage seems to have gained a different linguistic identity at a later time I feel it is way too close on the tree and similar in its early branching to see this as hugely distinct from P312. If it wasnt closely connected then why does it suddenly start branching for several generations at the same time as P312?

Also the flatlining of many branches and the domination of a limited number of the original branches seems to commence in the 3rd generation after U106. This sounds like the rise of elites and hierarchy to me. This actually does match the archaeological record where you see the rise off late/post-beaker more hierarchical societies in the centuries running up to 2000BC. The most famous would be Unetice but there were many across Europe like Wessex, Armorican dagger etc. I dont know where I would geographically place U106 in that period. I assume from its branching a few SNPs down that a few branches were elite in some way. Is there anything interesting in the geography of the subclades (other than in the isles - I dont think that is relevant).

Admittedly this is speculative and I hope to learn from others more up on the clades on this thread.

Preliminary U106 tree can be found online at: https://app.box.com/s/afqsrrnvv2d51msqcz2o

alan
05-05-2014, 05:15 PM
Point taken RR. I know it might change but there may already be some pattern that unites or divides the L11 derived branches. I will do some similar consideration of L21, DF27 and U152 once someone posts the most cutting edge trees for them. I dont want to start chewing over out of date trees by mistake.

alan
05-05-2014, 05:20 PM
Just moving RR's U152 tree over

http://r1b.org/imgs/U152_Tree_v003.png

alan
05-05-2014, 05:55 PM
U152, seems to have a healthy and sudden branching in the next SNP generation below it. The 2nd generation after U152 seems to show continuing healthy branching across much of the board. The 3rd generation however seems to see a lot of lines flat lining. I may be misunderstanding what is being presented on this but it looks like branching slowly flattens out around the 4-5 generation after U152 rather than frantic branching. The L2 branch seems to continue healthy branching for a few SNP steps longer than the rest of U152, for example Z193, which seem to flatline earlier. I wonder if that could imply different circumstances and/or social structure experienced within U152 with L2 having a few SNP step extra of branching.

Wing Genealogist
05-05-2014, 08:32 PM
Alan, I believe there may be a logical flaw in your argument about the upper branches of clades. For instance, while U106 has 7 branches, I would say only two (Z18 & Z381) are early branches. The other branches likely arose generations later (which would go to explain why they didn't proliferate like the larger two). As mutation can happen anytime as long as the parent clade (without any known mutations) is still around.

Wing Genealogist
05-05-2014, 08:52 PM
Alan, I think there may be a fundamental flaw in your thinking. While you state there are seven direct subclades of U106, I believe only two of them (Z18 & Z381) are early subclades, and the other subclades all appear to have occurred much later in time.

A clade can spawn new subclades at any age, as long as it does not die out.

Heber
05-05-2014, 11:38 PM
Here are the relevant trees under L11.

ISOGG 2014 Tree.

http://www.isogg.org/tree/index.html

Chris Morleys Tree.

http://ytree.morleydna.com/Experimen...ny20140207.pdf

Alex Williamson's Big Tree.

http://www.littlescottishcluster.com/RL21/NGS/Tree.html

YFull Tree

http://www.yfull.com/tree/

Dr. Jim Wilson Tree.

http://www.yourgeneticgenealogist.co...a-release.html

The Fudan University Tree.

http://arxiv.org/pdf/1310.6590.pdf

My own particular area of interest in L21 and it's defining mutations.

http://www.pinterest.com/gerardcorcoran/r1b-l21/
http://www.pinterest.com/gerardcorco...l21-df13-df21/
http://www.pinterest.com/gerardcorco...l21-df13-df49/
http://www.pinterest.com/gerardcorco...21-df13-c4466/
http://www.pinterest.com/gerardcorco...l21-df13-l513/
http://www.pinterest.com/gerardcorco...l21-df13-z253/
http://www.pinterest.com/gerardcorco...l21-df13-z255/
http://www.pinterest.com/gerardcorco...21-df13-l1335/
http://www.pinterest.com/gerardcorco...l21-df13-df41/

http://www.pinterest.com/gerardcorcoran/r1b-m269/
http://www.pinterest.com/gerardcorcoran/r1b-p312/
http://www.pinterest.com/gerardcorcoran/r1b-u152/
http://www.pinterest.com/gerardcorcoran/r1b-df27/
http://www.pinterest.com/gerardcorcoran/r1b-u106/

CTS and the DNA in Forensics community will have their annual conference in Brussels 14th - 15th May and will certainly present new details on the Phylogeny.
However I doubt they will publish to the level of granularity under L21 of Alex Williams tree (above).

Thursday 15 May

Y-chromosome
NGS related large-scale projects

9h00 O17 1,200 Y chromosome sequences from Phase 3 of the 1000 Genomes Project

C. Tyler-Smith, (Hinxton, United Kingdom)

9h30 O18 Growing the Y-chromosome tree with 13,000 high-confidence SNPs from next-generation sequencing

M. Jobling, C. Batini, D. Zadik, J. Wetton, E. Arroyo, W. Bodmer, G. Cavalleri, P. de Knijff, G. Destro
Bisol, B. Myhre Dupuy, L. B. Jorde, T. E. King, A. López de Munain, J. Milasin, A. Novelletto, H. Pamjav, M. Rocchi, A. Sajantila, W. Schem
pp, A. Tolun, C. Tyler-Smith, B. Winney, S. Watkins, P. Hallast (Leicester, United Kingdom)

9h50 O19 Y-chromosome haplogrouping and paternal ancestry inference via simultaneous analysis of 550 Y-SNPs with IonTorrent PGM

M. Kayser, A. Ralf, K. Zhong, M. van Oven (Rotterdam, Netherlands)
10h05 O20 In silico detection of phylogenetic informative Y-chromosomal SNPs from WGS data

A. Van Geystelen, T. Wenseleers, R. Decorte, M. J.L. Caspers, M. H.D. Larmuseau (Leuven, Belgium)

10h20 O21 Seeing the wood for the trees: a minimal reference phylogeny for the human Y chromosome
M. van Oven, A. Van Geystelen, M. Kayser, R. Decorte,
M. H.D. Larmuseau (Rotterdam, Netherlands)

http://www.forensischinstituut.be/upload/images/DiF2014%20Agenda.pdf

R.Rocca
05-08-2014, 05:02 PM
Point taken RR. I know it might change but there may already be some pattern that unites or divides the L11 derived branches. I will do some similar consideration of L21, DF27 and U152 once someone posts the most cutting edge trees for them. I dont want to start chewing over out of date trees by mistake.

Alan, the very tree we are looking for (an unbiased, computer-generated one with many world populations) has already been generated by Magoon et al. 2013. By taking the 1000 Genomes samples from around the world, they noted that...


The majority of nodes in the tree lead to bifurcations, however there are also many instances of 3-6 sub-branches occurring. We were particularly interested in investigating any star-like branching patterns, since these are usually associated with rapid population growth. The most extreme examples in the current phylogeny include between 8 and 18 parallel sub-branches, and are detailed in Table 1. We find evidence for star-like rapid population growth in two regions, Europe and Africa, and in two subhaplogroups, R1b-M269 and E-L576.

Star-like phylogeny structure, indicative of rapid population expansion, was found in European and African regions of the phylogeny. In Europe much has been made of the evidence for rapid increases in population, although there is not a clear consensus on the timing(s) and locations of the expansions. In Africa the best candidate for rapid population growth is the Bantu expansion.

So of all the world-wide populations sampled, three of the six branches of the human family that show a rapid population expansion are P312 derived (R1b-DF27, R1b-DF13 and R1b-L2). The expansion of the other three are in haplogroup E.

see page 6...

Gregory R Magoon, Raymond H Banks, Christian Rottensteiner, et al. (2013) Generation of high-resolution a priori Y-chromosome phylogenies using “next-generation” sequencing data

http://biorxiv.org/content/biorxiv/early/2013/11/22/000802.1.full.pdf

Heber
05-08-2014, 06:32 PM
Alan, the very tree we are looking for (an unbiased, computer-generated one with many world populations) has already been generated by Magoon et al. 2013. By taking the 1000 Genomes samples from around the world, they noted that...



So of all the world-wide populations sampled, three of the six branches of the human family that show a rapid population expansion are P312 derived (R1b-DF27, R1b-DF13 and R1b-L2). The expansion of the other three are in haplogroup E.

see page 6...

Gregory R Magoon, Raymond H Banks, Christian Rottensteiner, et al. (2013) Generation of high-resolution a priori Y-chromosome phylogenies using “next-generation” sequencing data

http://biorxiv.org/content/biorxiv/early/2013/11/22/000802.1.full.pdf


You can clearly see the expansion of R1b-DF27, R1b-DF13 and R1b-L2 here:
http://www.pinterest.com/gerardcorcoran/r1b-m269/
It is obvious that the extreme expansion is R1b-DF27 and this appears to have happened in Iberia.
This "extreme" expansion of R1b was also noted by CTS, Wei and Zhu.
http://eurogenes.blogspot.com/2012/09/next-generation-resequencing-data.html
This poses the question where was P312 and L11 during this time or immediately prior.
If you triangulate between DF27, L2 and DF13 it shifts the center of gravity west.
It will be interesting to see the long overdue detailed results of the Genographic Project in Asturias.