World Families Forums - R1b-L51 from the West

Welcome, Guest. Please login or register.
October 23, 2014, 01:26:16 AM
Home Help Search Login Register

+  World Families Forums
|-+  General Forums - Note: You must Be Logged In to post. Anyone can browse.
| |-+  R1b General (Moderator: rms2)
| | |-+  R1b-L51 from the West
« previous next »
Pages: 1 ... 7 8 [9] 10 11 12 Go Down Print
Author Topic: R1b-L51 from the West  (Read 31878 times)
alan trowel hands.
Guru
*****
Offline Offline

Posts: 2012


« Reply #200 on: June 14, 2012, 07:17:27 PM »

I tend to suspect U106 arose in NE Europe from an L11* line that moved there from somewhere to the south or west a couple of centuries earlier.  I would link this to the bell beaker culture.  Well certainly every bit as much as I would link P312 to it. 

Here is a very useful paper on beakers peripheral elements

 http://www.aegeobalkanprehistory.net/article.php?id_art=10
Logged
JeanL
Old Hand
****
Offline Offline

Posts: 425


« Reply #201 on: June 14, 2012, 07:27:35 PM »

So what? First, you list no error ranges in your example.

I list no sigmas because “my”(It was actually you who posted that table) example is from Marko.H calculations which show no sigmas.

I provided error ranges from Nordtvedt's output and the overlapping error ranges align nicely, as they should, with the phylogenetic tree. There is no reason to suspect any thing is off related to the closeness in ages between the U106&P312 TMRCA and the U106 TMRCA.
Secondly,  Is there any reason why the U106 TMRCA should not or can not be relatively close in age to the U106&P312 TMRCA?

I just find it weird that the intraclade age of R1b-U106 is what dictates the interclade age of R1b-U106&R1b-P312, and that the intraclade age of I2 is what dictates the interclade age of I1&I2. So yes there is a reason to suspect it, mainly because according to you intraclade sets lower bound whereas interclade sets upper bound, and how can the lower and upper bound be exactly the same? Moreover this shows that if there is any effects that are undermining the intraclade age of R1b-U106 they will be translated into the interclade age of R1b-P312 and R1b-U106.

Now why don’t you answer the question you seem to be ignoring for the past couple of posts:

Again how many of the R1b-P312+ 4000 haplotypes that were used to calculate R1b-P312’s TMRCA have their MDKA from the British Islands?

What’s more how many of the R1b-U106+ 1200 haplotypes that were used to calculate R1b-U106’s TMRCA have their MDKA from the British Islands?

Why not go download Nordtvedt's tool and use it?  You can tear it apart.  All of the formulas are there, sigmas and the whole bit.  If you make a list of errors, I'll present that to Ken for you if you are afraid to yourself.

Why?? I have my own tool, I programmed it using Matlab, I don’t need to tear his tool apart, and I have no bone to pick with him, his program doesn’t have any errors, the “errors” come into play from sampling bias, which is what is clearly going on when the data from FTDNA is used, at least for the R1b-P310+ subclades.
Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #202 on: June 14, 2012, 08:02:23 PM »

So what? First, you list no error ranges in your example.

I list no sigmas because “my”(It was actually you who posted that table) example is from Marko.H calculations which show no sigmas.

I provided error ranges from Nordtvedt's output and the overlapping error ranges align nicely, as they should, with the phylogenetic tree. There is no reason to suspect any thing is off related to the closeness in ages between the U106&P312 TMRCA and the U106 TMRCA.
Secondly,  Is there any reason why the U106 TMRCA should not or can not be relatively close in age to the U106&P312 TMRCA?

I just find it weird that the intraclade age of R1b-U106 is what dictates the interclade age of R1b-U106&R1b-P312, and that the intraclade age of I2 is what dictates the interclade age of I1&I2. So yes there is a reason to suspect it, mainly because according to you intraclade sets lower bound whereas interclade sets upper bound, and how can the lower and upper bound be exactly the same? Moreover this shows that if there is any effects that are undermining the intraclade age of R1b-U106 they will be translated into the interclade age of R1b-P312 and R1b-U106.

I'll recommend again that you download Ken Nordtvedt's documents and spreadsheet and figure it out.  I look at his powerpoints of the conceptual overview and have had several conversations with him.  At times, I see how it fits together, but only at a high level.  I have another things to worry about and I'm not that smart so I haven't tried to totally grapple with interclade parts of it.  Conceptually (the powerpoints) it makes sense and I trust that Ken is putting the concepts into the spreadsheet correctly.



Now why don’t you answer the question you seem to be ignoring for the past couple of posts:

Again how many of the R1b-P312+ 4000 haplotypes that were used to calculate R1b-P312’s TMRCA have their MDKA from the British Islands?
What’s more how many of the R1b-U106+ 1200 haplotypes that were used to calculate R1b-U106’s TMRCA have their MDKA from the British Islands?

I actually have other things to do and honestly when I've looked at that kind of thing, diversity by geography, usually the differences in these subclades (U152, L2, L21, DF27 (replacing Z196), U106, Z381, P312, etc.) are minor. You'd probably say the same thing that Busby says. The clines are not that steep.

What I am saying this line of counter-argument is not very compelling to me. I think I already know the answer so it's just matter of bantering with you.

Don't worry. I care about you.  I won't look at reloading Ken's Gen spreadsheet with this data though as I want to incorporate his new version for 113 STRs first.  I will do some variance comparison for you by geography.  I think we have to keep in mind that today's political boundaries are not necessarily critical to old Y DNA distribution patterns.  Oceans, mountain ranges, river valleys are also important as well as ancient political boundaries.

The intraclade variances will go up and down somewhat as you say because of the bias of the sample.  So far I haven't seen that move the needle much (EDIT: fixed typos) on interclade calculations though.

Why not go download Nordtvedt's tool and use it?  You can tear it apart.  All of the formulas are there, sigmas and the whole bit.  If you make a list of errors, I'll present that to Ken for you if you are afraid to yourself.


Why?? I have my own tool, I programmed it using Matlab, I don’t need to tear his tool apart, and I have no bone to pick with him, his program doesn’t have any errors, the “errors” come into play from sampling bias, which is what is clearly going on when the data from FTDNA is used, at least for the R1b-P310+ subclades.  

Super. Please produce your own estimates with your own tool, then.  Please consider publishing the formulas and providing charts that provide an overview.

Can you build a spreadsheet that creates random numbers and then pulls rows from a second spreadsheet full of haplotype data?  We need it to pull (copy/paste) rows based on selection criteria by column (i.e. country and/or subclade.)  I suspect we need to develop a process where this is done multiple times in batch (without manual intervention) with a regression analysis (of which I forgotten about since my collegiate days.)   I would do this, and may some day but it looks like a big task.
« Last Edit: June 15, 2012, 08:54:59 AM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
rms2
Board Moderator
Guru
*****
Offline Offline

Posts: 5023


« Reply #203 on: June 14, 2012, 08:02:46 PM »

I tend to suspect U106 arose in NE Europe from an L11* line that moved there from somewhere to the south or west a couple of centuries earlier.  I would link this to the bell beaker culture.  Well certainly every bit as much as I would link P312 to it. 

Here is a very useful paper on beakers peripheral elements

 http://www.aegeobalkanprehistory.net/article.php?id_art=10

The only exception I would take to what you posted is that we have two Bell Beaker y-dna results, and both were U106-. Of course, we don't know whether or not they were P312+. They could have simply been L11+ and no further up the tree.

I know two are not many, but they were found in what is today U106-rich country, and neither was U106+.

I tend to hold Bell Beaker accountable for Italo-Celtic, and I don't think U106 had any kind of a hand in that. So, yeah, I would give U106 a northeastern origin and have it caught up in the Corded Ware/Nordic Bronze Age thing fairly early.
Logged

JeanL
Old Hand
****
Offline Offline

Posts: 425


« Reply #204 on: June 14, 2012, 08:30:18 PM »

I'll recommend again that you download Ken Nordtvedt's documents and spreadsheet and figure it out.  I look at his powerpoints of the conceptual overview and have had several conversations with him.  At times, I see how it fits together, but only at a high level.  I have another things to worry about and I'm not that smart so I haven't tried to totally grapple with interclade parts of it.  Conceptually (the powerpoints) it makes sense and I trust that Ken is putting the concepts into the spreadsheet correctly.

Ok, let me put it in simpler words, I have no reason to download his spreadsheet, because I have no reason to doubt that the estimates are right, the “data” is what seems to be the problem.

I actually have other things to do and honestly when I've looked at that kind of thing, diversity by geography, usually the differences in these subclades (U152, L2, L21, DF27 (replacing Z196), U106, Z381, P312, etc.) are minor. You'd probably say the same thing that Busby says. The clines are not that steep.

Really, that’s not the impression I have gotten from observing the data from academic studies, certainly there is no east-west pattern for R1b-L11+ as per Busby.et.al.data, but there is definitely regional variance.

What I am saying this line of counter-argument is not very compelling to me. I think I already know the answer so it's just matter of bantering with you.

Ok, so the representativeness of the haplotypes used to calculate the MRCA of P312+ is not a compelling issue to you?

Don't worry. I care about you.  I won't look at reloading Ken's Gen spreadsheet with this data though as I want to incorporate his new version for 113 STRs first.  I will do some variance comparison for you by geography.  I think we have to keep in mind that today's political boundaries are not necessarily critical to old Y DNA distribution patterns.  Oceans, mountain ranges, river valleys are also important as well as ancient political boundaries.
Well yeah, I not mean countries per se, but regions, for example SW France ought to be investigated separately from NW France or NE France, etc.

The intraclade variances will go up and down somewhat as you say because of the bias of the sample.  So far I haven't seen that make much of a needle on interclade calculations though.

Well you haven’t seen much of a needle on inter clade calculations because you haven’t used the intraclade of say NW French R1b-L21 and NW French R1b-Z196, let see if that is different from the interclade age of all R1b-L21 and R1b-Z196 for example.

Super. Please produce your own estimates with your own tool, then.  Please consider publishing the formulas and providing charts that provide an overview.

The program I wrote doesn’t produce age estimates, it calculate variance and modal haplotype for a given dataset. However that variance(mean mutations per marker) needs to be corrected for back mutations in order to calculate the TMRCA, however since I have yet to find a reliable way to do such thing, I haven’t taken it to the next step. As for the formulas, well the program works in independent locus, and picks the modal value based on the allele value that minimizes the number of mutations in that given locus. So basically, it is a bunch of nested “for” and “do” loops, nonetheless I gotten similar results to the variance calculated using the mean haplotype methodology in Myres.et.al.2010.

Can you build a spreadsheet that creates random numbers and then pulls rows from a second spreadsheet full of haplotype data?  We need it to pull (copy/paste) rows based on selection criteria by column (i.e. country and/or subclade.)  I suspect we need to develop a process where this is done multiple times in batch (without manual intervention) with a regression analysis (of which I forgotten about since my collegiate days.)   I would do this, and may some day but it looks like a big task.

I don’t work directly with Excel, I only use it to store or obtain data from it. The Matlab codes I have written usually read off haplotypes from an Excel table and stored them on an array, then whenever the code is done, it will write off the results to a different spreadsheet. As for the regression analysis, well, you could certainly pick multiple combinations of n number of haplotypes and calculate the variance of them for x amount of simulations, then one finds the mean variance value and the standard deviation, if the stdev is fairly small relative to the mean, then we are good, and the randomly collected samples are actually representative of the whole data set. Of course the greater the value of x simulations, then less the likelyhood that some oddball outlier configuration will drive the stdev up, if 10000 simulations were to be done were 80 haplotypes were sample at random and their variance was calculated, and the mean and stdev are about the same order, then the data is no good, and using 80 haplotypes will very likely result in misrepresenting the variance of the whole set, so perhaps a greater number of haplotypes ought to be randomly sampled.
« Last Edit: June 14, 2012, 08:34:08 PM by JeanL » Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #205 on: June 15, 2012, 08:48:06 AM »

I'll recommend again that you download Ken Nordtvedt's documents and spreadsheet and figure it out.  I look at his powerpoints of the conceptual overview and have had several conversations with him.  At times, I see how it fits together, but only at a high level.  I have another things to worry about and I'm not that smart so I haven't tried to totally grapple with interclade parts of it.  Conceptually (the powerpoints) it makes sense and I trust that Ken is putting the concepts into the spreadsheet correctly.

Ok, let me put it in simpler words, I have no reason to download his spreadsheet, because I have no reason to doubt that the estimates are right, the “data” is what seems to be the problem.

If I understand you, you are assuming the data is not useful in it current state and therefore you will not consider it. I don't know how you know the data has a problem other than it doesn't fit your model.

... but whatever,   Fine.  Why don't you fix the data then?  You could help me do that if you could develop a spreadsheet that used the Excel "rand" function to randomly select rows of haplotypes out of a 2nd spreadsheet based on given sets of criteria (i.e. Geography and Subclade) and copy those rows into the target spreadsheet.  I'm not the re-sampling expert, but I assume you'd want to be able to create a number of these randomly selected sample spreadsheets in a big batch so you run TMRCAs on each.  If you can do that I'll figure out how to import these data sets into Ken Nordtvedt's Generation spreadsheet. I'm not making money on this stuff? Are you planning to?  There is no reason for you or me not to provide all of the data and formulas publicly. This has always been my practice.


I actually have other things to do and honestly when I've looked at that kind of thing, diversity by geography, usually the differences in these subclades (U152, L2, L21, DF27 (replacing Z196), U106, Z381, P312, etc.) are minor. You'd probably say the same thing that Busby says. The clines are not that steep.

Really, that’s not the impression I have gotten from observing the data from academic studies, certainly there is no east-west pattern for R1b-L11+ as per Busby.et.al.data, but there is definitely regional variance.

I think there are clines, as does RMS, but I think the clines are minor. Busby, and apparently yourself, think they are not present all (at least for L11.)

In this sense I'm agreeing with you. I don't think it is worth it to do the data gyrations you are asking for in separating various TMRCA runs by country because the differences between countries for P312, L21, U152, U106, Z381 will be relatively minor (probably 5-15%, maybe up to 20%.) Apparently you think they will be less, but either way it won't be conclusive.  So why should I attempt the gyrations of separating out the UK?  If we find 10% differences between France and England I'd expect you to say they were insignificant anyway.

I see that you say there is definitely regional variance. I'm not sure how there is both no geographic cline and regional variance at the same time, but since you see this definite regional variance, please show it to us, hopefully on long haplotypes and with a large sample.


What I am saying this line of counter-argument is not very compelling to me. I think I already know the answer so it's just matter of bantering with you.

Ok, so the representativeness of the haplotypes used to calculate the MRCA of P312+ is not a compelling issue to you?

No, I'm saying you've not made a compelling enough case for me to want to do extra work.

We've just discussed you think there are no geographic clines and I think they are minor, hence P312's diversity won't change drastically from one major European country to the next as long as the sample sizes are large.  You are asking to pull out data by country. I don't have time to waste time.  I can calculate variance by country very quickly but it takes a while import stuff into Ken Nordtvedt's spreadsheet.

I also understand, we just plain do NOT have enough data to do true randomly sampled cross-sectional surveys the way they should be done.   We don't have enough to data to be truly representative, period.  Sectioning the data by country is not enough. It should be done by ethnicity, by community/region (not necessarily country) and by subclade within those categories. I'm not a demographic expert, but what I listed is probably just the start.


Don't worry. I care about you.  I won't look at reloading Ken's Gen spreadsheet with this data though as I want to incorporate his new version for 113 STRs first.  I will do some variance comparison for you by geography.  I think we have to keep in mind that today's political boundaries are not necessarily critical to old Y DNA distribution patterns.  Oceans, mountain ranges, river valleys are also important as well as ancient political boundaries.

Well yeah, I not mean countries per se, but regions, for example SW France ought to be investigated separately from NW France or NE France, etc.

The intraclade variances will go up and down somewhat as you say because of the bias of the sample.  So far I haven't seen that move the needle much (EDIT: fixed typos) on interclade calculations though.

Well you haven’t seen much of a needle on inter clade calculations because you haven’t used the intraclade of say NW French R1b-L21 and NW French R1b-Z196, let see if that is different from the interclade age of all R1b-L21 and R1b-Z196 for example.

We don't have enough haplotypes, let alone long ones, to do all of this. If we did, you'd probably complain that since many of the people are actually American citizens with MDKA's in these places we should probably throw those out.

You should probably join the National Genographic Project and convince them to sample rural areas all over Europe for people who are known to have g-grandparents and as far back as they know from their respective rural regions.  Then National Geno folks do this kind of thing, but haven't attempted any Pan-European survey.
« Last Edit: June 15, 2012, 11:07:53 AM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #206 on: June 15, 2012, 08:59:50 AM »


I tend to suspect U106 arose in NE Europe from an L11* line that moved there from somewhere to the south or west a couple of centuries earlier.  I would link this to the bell beaker culture.  Well certainly every bit as much as I would link P312 to it. 

Here is a very useful paper on beakers peripheral elements
 http://www.aegeobalkanprehistory.net/article.php?id_art=10

The only exception I would take to what you posted is that we have two Bell Beaker y-dna results, and both were U106-. Of course, we don't know whether or not they were P312+. They could have simply been L11+ and no further up the tree.

I know two are not many, but they were found in what is today U106-rich country, and neither was U106+.

I tend to hold Bell Beaker accountable for Italo-Celtic, and I don't think U106 had any kind of a hand in that. So, yeah, I would give U106 a northeastern origin and have it caught up in the Corded Ware/Nordic Bronze Age thing fairly early.

Do you think U106 was in Corded Ware from its inception?  If so, what predecessor culture might it have come from?  I guess I'm just asking how and from what direction you think U106 got involved?
Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
JeanL
Old Hand
****
Offline Offline

Posts: 425


« Reply #207 on: June 15, 2012, 09:16:49 AM »

If I understand you, you are assuming the data is not useful in it current state and therefore you will not consider it. I don't know how you know the data has a problem other than it doesn't fit your model.

Ok, let’s keep this simple: Are the majority of the 4000 haplotypes used to calculate the MRCA of P312+ clades hailing of UK ancestry paternally, yes or no?

... but whatever,   Fine.  Why don't you fix the data then?  You could help me do that if you could develop a spreadsheet that used the Excel "rand" function to randomly select rows of haplotypes out of a 2nd spreadsheet based on given sets of criteria (i.e. Geography and Subclade) and copy those rows into the target spreadsheet.  I'm not the re-sampling expert, but I assume you'd want to be able to create a number of these randomly selected sample spreadsheets in a big batch so you run TMRCAs on each.  If you can do that I'll figure out how to import these data sets into Ken Nordtvedt's Generation spreadsheet. I'm not making money on this stuff? Are you planning to?  There is no reason for you or me not to provide all of the data and formulas publicly. This has always been my practice.

I can randomly sample x number of haplotypes, however, I don’t work with Excel, I work with Matlab, but yeah if you give me an Excel spreadsheet, I can load it into Matlab, and I can write a code that samples x number of haplotypes at random from the array. I’m still confused by what you meant when you said provide formulas publicly? It is a computer code, that’s about it, what formula do you need to see, I already explained how it works. It compares alleles values at any given locus and choose the modal value as the one that minimizes the number of mutations in that locus, at the end all the mutations of all loci are added and divided by the product of the total number of loci times the sample size.

I think there are clines, as does RMS, but I think the clines are minor. Busby, and apparently yourself, think they are not present all (at least for L11.)

There is no east-west cline for R1b-L11+, at least not  from the data presented by Busby.et.al.2011. There are regional clines, they just don’t follow an East-West distribution.

In this sense I'm agreeing with you. I don't think it is worth it to do the data gyrations you are asking for in separating various TMRCA runs by country because the differences between countries for P312, L21, U152, U106, Z381 will relatively minor (probably 5-15%, maybe up to 20%.) Apparently you think they will be less, but either way it won't be conclusive.  So why should I attempt the gyrations of separating out the UK?

How do you know that the differences between the TMRCA of those clades will be less than 20%, have you done the experiment? Whoao, talk about preconceived notions!! Here, take a look at table-S2 of Myres.et.al.2010 the variance for R1b-S116(all) that is R1b-P312+ goes from 0.2333 in England(n=48) to 0.307 in Vaucluse, France, that is 31.76% higher than in England. 

I see that you say there is definitely regional variance. I'm not sure who there is both no cline and regional variance at the same time, but since you see this definite regional variance, please show it to us, hopefully on long haplotypes and with a large sample.

Well Mike, sorry, but there isn’t any representative data out there that is “hopefully on long haplotypes”, nonetheless if you wish, you can look at Busby.et.al.2011 Figure-2, and you will see that the regional variance in Western Europe varies from 0.20 to 0.40. Of course, since you know beforehand that there aren’t any academic studies out there on R1b with long haplotypes, you are cooking the strawman, but sorry pal, I aint falling for it. The data is out there, now if you choose to ignore it, is up to you.


No, I'm saying you've not made a compelling enough case for me to want to do extra work.

I have made my case, and explained my hypothesis quite a few times already, now if you choose to ignore it, or if you don’t find it compelling, good for you, let’s not waste my time either then. 

We've just discussed you think there are no geographic clines and I think they are minor, hence P312's diversity won't change drastically from one major European country to the next as long as the sample sizes are large. You are asking to pull out data by country. I don't have time to waste time.  I can calculate variance by country very quickly but it takes a while import stuff into Ken Nordtvedt's spreadsheet.

That’s purely wishful thinking, you have no idea if they are going to change or not, I can tell you that at least in the Academic studies published P312’s diversity does vary, but if you don’t consider an increase from 0.19 to 0.307 as drastic, well then I guess not.

I also understand, we just plain do NOT have enough data to do true randomly sampled cross-sectional surveys the way they should be done.  We don't have enough to data to be truly representative, period.   Sectioning the data by country is not enough. It should be done by ethnicity, by community/region (not necessarily country) and by subclade within those categories. I'm not a demographic expert, but what I listed is probably just the start.

Well, sectioning the data by country is a start, however, sectioning the data by regions that have known historical background, and that show continuity for quite some time based upon other factors would be preferable.

We don't have enough long haplotypes to do all of this. If we did, you'd probably complain that since many of the people are actually American citizens with MDKA's in these places we should probably throw those out.

Really Mike!! Now you are psychic? You know what I am going to say then? Good for you!!

You should probably join the National Genographic Project and convince them to sample rural areas all over Europe for people who are known to have g-grandparents and as far back as they know from their respective rural regions.  Then National Geno folks do this kind of thing, but haven't attempted any Pan-European survey.

Why?? There are plenty of academics out there sampling regions right now, in the last 2 years there have been three major studies published on the R1b field, each one going more into more resolution, so why shouldn’t we expect more to come?
« Last Edit: June 15, 2012, 09:20:24 AM by JeanL » Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #208 on: June 15, 2012, 10:52:44 AM »



I'll recommend again that you download Ken Nordtvedt's documents and spreadsheet and figure it out.  I look at his powerpoints of the conceptual overview and have had several conversations with him.  At times, I see how it fits together, but only at a high level.  I have another things to worry about and I'm not that smart so I haven't tried to totally grapple with interclade parts of it.  Conceptually (the powerpoints) it makes sense and I trust that Ken is putting the concepts into the spreadsheet correctly.

Ok, let me put it in simpler words, I have no reason to download his spreadsheet, because I have no reason to doubt that the estimates are right, the “data” is what seems to be the problem.

If I understand you, you are assuming the data is not useful in it current state and therefore you will not consider it. I don't know how you know the data has a problem other than it doesn't fit your model.
... but whatever,   Fine.  Why don't you fix the data then?  

JeanL responded...
I can randomly sample x number of haplotypes, however, I don’t work with Excel, I work with Matlab, but yeah if you give me an Excel spreadsheet, I can load it into Matlab, and I can write a code that samples x number of haplotypes at random from the array. I’m still confused by what you meant when you said provide formulas publicly? It is a computer code, that’s about it, what formula do you need to see, I already explained how it works. It compares alleles values at any given locus and choose the modal value as the one that minimizes the number of mutations in that locus, at the end all the mutations of all loci are added and divided by the product of the total number of loci times the sample size.

All of the data I've described is posted. Download it yourself.  The benefits to the community as a whole are better if I keep the spreadsheets up to date, versus becoming a programmer.  There are no secrets other than, of course, as project admin I don't share kit contact information and the like for privacy concerns.
« Last Edit: June 15, 2012, 11:05:08 AM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #209 on: June 15, 2012, 11:15:50 AM »


In this sense I'm agreeing with you. I don't think it is worth it to do the data gyrations you are asking for in separating various TMRCA runs by country because the differences between countries for P312, L21, U152, U106, Z381 will relatively minor (probably 5-15%, maybe up to 20%.) Apparently you think they will be less, but either way it won't be conclusive.  So why should I attempt the gyrations of separating out the UK?

How do you know that the differences between the TMRCA of those clades will be less than 20%, have you done the experiment? Whoao, talk about preconceived notions!! Here, take a look at table-S2 of Myres.et.al.2010 the variance for R1b-S116(all) that is R1b-P312+ goes from 0.2333 in England(n=48) to 0.307 in Vaucluse, France, that is 31.76% higher than in England.  

I have no preconceived notions on variance of P312 by geography. If the intraclade variances are not much different, the intraclade TMRCAs won't be either. I've been tracking this stuff for over a year and done comparisons between geographies a hundred ways to Sunday. I'm not running an experiment just for you.  Run your own, but please use long haplotypes.  I think these 10 to 15 STR analyses are subject to uncertainty (and I've already explained why.)

Of course, since you know beforehand that there aren’t any academic studies out there on R1b with long haplotypes, you are cooking the strawman, but sorry pal, I aint falling for it. The data is out there, now if you choose to ignore it, is up to you.

If you think the data is adequate then you are smoking something.  

I just do the best I can with the data I've got.  That includes academic data, but through my own experimentation and observation I can tell you that running analyses on just a few STRs is just not enough.  

EDIT: The same goes with a sample of 20 or so from somewhere in France, like Vacluse, but ignoring the rest, or a sample of a handful in Switzerland. We don't have a good Pan-European survey.
« Last Edit: June 15, 2012, 11:25:58 AM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
JeanL
Old Hand
****
Offline Offline

Posts: 425


« Reply #210 on: June 15, 2012, 11:21:05 AM »

I have no preconceived notions on variance of P312 by geography. If the intraclade variances are not much different, the intraclade TMRCAs won't be either. I've been tracking this stuff for over a year and done comparisons between geographies a hundred ways to Sunday. I'm not running an experiment just for you.  Run your own, but please use long haplotypes.  I think these 10 to 15 STR analyses are subject to uncertainty (and I've already explained why.)

Well and I think the FTDNA projects are greatly overpopulated by folks with British Isles descent, and that yeah, you have the long haplotypes, but you lack the representative samples(and I’ve already explained why). So we are getting into an argument of what data is better, you claim FTDNA based on long haplotypes, I claim academics based on sampling standards, and recent findings that is not the amount of STRs but the choice of STRs.(Busby.et.al.2011)

If you think the data is adequate then you are smoking something. 

I just do the best I can with the data I've got.  That includes academic data, but through my own experimentation and observation I can tell you that running analyses on just a few STRs is just not enough.

I'm probably smoking something similar to what you are smoking that leads you to believe that the FTDNA data is representive of Western Europe. Moreover how can you tell that running analyses using 10 or 15 STRs gives erroneous results, you have yet to show that? So go ahead, and show how analyses performed on academic studies using 10 or 15 STRs give inconsistent results.
« Last Edit: June 15, 2012, 11:24:29 AM by JeanL » Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #211 on: June 15, 2012, 11:41:04 AM »

I see that you say there is definitely regional variance. I'm not sure who there is both no cline and regional variance at the same time, but since you see this definite regional variance, please show it to us, hopefully on long haplotypes and with a large sample.

Well Mike, sorry, but there isn’t any representative data out there that is “hopefully on long haplotypes”, nonetheless if you wish, you can look at Busby.et.al.2011 Figure-2, and you will see that the regional variance in Western Europe varies from 0.20 to 0.40. Of course, since you know beforehand that there aren’t any academic studies out there on R1b with long haplotypes, you are cooking the strawman, but sorry pal, I aint falling for it. The data is out there, now if you choose to ignore it, is up to you.  

If you think the data is adequate then you are smoking something.  

I just do the best I can with the data I've got.  That includes academic data, but through my own experimentation and observation I can tell you that running analyses on just a few STRs is just not enough.

Moreover how can you tell that running analyses using 10 or 15 STRs gives erroneous results, you have yet to show that? So go ahead, and show how analyses performed on academic studies using 10 or 15 STRs give inconsistent results.

I said that 10 to 15 STR analyses (or less) are subject to uncertainty, not that they are necessarily erroneous.

I don't think I have to show about the uncertainty of the academic's short haplotype data.  Just read the studies. We see folks like Balaresque, Myres and Busby disagree vehemently depending on their unique gyrations of the data.  There is disagreement among the academic types not just on the conclusions, but on what data to look at and how to slice it and dice it.

Even on short haplotypes, do you really think the data is represenative of Europe?  What is your academic conglomerated database sample size by region of France?

Well and I think the FTDNA projects are greatly overpopulated by folks with British Isles descent, and that yeah, you have the long haplotypes, but you lack the representative samples(and I’ve already explained why). So we are getting into an argument of what data is better, you claim FTDNA based on long haplotypes, I claim academics based on sampling standards, and recent findings that is not the amount of STRs but the choice of STRs.(Busby.et.al.2011) 

You apparently are arguing with yourself.  I agree with you that FTDNA projects are heavily biased by folks with Isles MDKA's.   I agree!  I just think the long haplotypes and deep SNP testing are the level of data needed, but I agree that there is an Isles bias.

« Last Edit: June 15, 2012, 11:46:26 AM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
JeanL
Old Hand
****
Offline Offline

Posts: 425


« Reply #212 on: June 15, 2012, 11:45:42 AM »

EDIT: The same goes with a sample of 20 or so from somewhere in France, like Vacluse, but ignoring the rest, or a sample of a handful in Switzerland. We don't have a good Pan-European survey.

Well not having a good Pan-European survey hasn’t stop you from postulating hypothesis based on the variance that you have calculated from FTDNA projects, yet when someone tries to argue for a different hypothesis based on trends observed in some Academic studies, then it turns out we don’t have a “good” Pan-European survey, seems like a double standard to me. So what’s it going to be, dismiss data from Academic studies because it comes in 10-15 STR format, and for some reason(unknown to me) that would definitely give inconsistent results. While we are at it, what is the minimum threshold the Academics should aim for in terms of number of STRs, please do tell us how you arrive to that minimum number, it would be interesting to see the formulation. In the mean time, we can all keep calculating the TMRCA of P312+ based on an astonishing 4000 67 STR haplotypes, never mind that more than 60% of those are R1b-L21+, and that there is a disproportionate amount of people with the MDKA from the British Isles in those 4000 haplotypes.
Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #213 on: June 15, 2012, 11:47:51 AM »

EDIT: The same goes with a sample of 20 or so from somewhere in France, like Vacluse, but ignoring the rest, or a sample of a handful in Switzerland. We don't have a good Pan-European survey.

Well not having a good Pan-European survey hasn’t stop you from postulating hypothesis based on the variance that you have calculated from FTDNA projects, yet when someone tries to argue for a different hypothesis based on trends observed in some Academic studies, then it turns out we don’t have a “good” Pan-European survey, seems like a double standard to me. ....

Okay, I apologize for being too hard on you.
Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
JeanL
Old Hand
****
Offline Offline

Posts: 425


« Reply #214 on: June 15, 2012, 11:50:32 AM »

I said that 10 to 15 STR analyses (or less) are subject to uncertainty, not that they are necessarily erroneous.

Well and are we to believe that 67 STR analyses are not subject to uncertainty?

I don't think I have to show about the uncertainty of the academic's short haplotype data.  Just read the studies. We see folks like Balaresque, Myres and Busby disagree vehemently depending on their unique gyrations of the data.  There is disagreement among the academic types not just on the conclusions, but on what data to look at and how to slice it and dice it.

Yes from the way you wrote that it seems that all studies were published parallel and simply provided opposite views. In reality Busby.et.al.2011 analyzed the works of Myres.et.al.2010 and Balaresque.et.al.2010 and found inconsistencies in the data presented by Balaresque.et.al, but moreover they found the effect of a choice of STRs on age estimates. I have yet to see Balaresque or anyone of her team publish something saying they disagree with that.
 
Okay, I apologize for being too hard on you.

No worries, I admire you work, and I like the fact that you defend your beliefs so fiercely.
« Last Edit: June 15, 2012, 11:54:47 AM by JeanL » Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #215 on: June 15, 2012, 12:05:32 PM »

I said that 10 to 15 STR analyses (or less) are subject to uncertainty, not that they are necessarily erroneous.

Well and are we to believe that 67 STR analyses are not subject to uncertainty?

Of course long haplotypes also allow uncertainty, but tripling or quadrupling the number of STR experiments with long haplotypes will reduce the uncertainty and with Ken Nordtvedt's tools we actually have the uncertainty measured and reported.
« Last Edit: June 15, 2012, 01:44:22 PM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #216 on: June 15, 2012, 12:08:23 PM »


Okay, I apologize for being too hard on you.

No worries, I admire you work, and I like the fact that you defend your beliefs so fiercely.

Believe it or not, I enjoy a good argument. I just like to keep the different points separate on their own merits before considering in context. When the sub-points get all tangled together it is confusing.

I've argued with my siblings at the dinner table growing up sp much that I guess I miss it.
Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #217 on: June 15, 2012, 12:39:54 PM »

In the interest of moving on, I'll just present this as added information and not as evidence supporting any hypothesis.

....  We have a lot of data on these clades like U152, L2, DF27, L21, U106, Z381, etc.  You can just look at the modals, and go down another layer on the tree to look again.  The probable ancestral values kind of pop out at you. ...

From the R1b modals thread, just posted...
....
I haven't updated these Ysearch records yet, but they don't hold 111 markers anyway.  Based on some prodding from JeanL, I went through and rechecked them in the spreadsheets.

My opinion is that the Western Atlantic Modal Haplotype (WAMH) or the Super WAMH version of it are obsolete.  We now have much greater granularity in subclades.  WAMH is effectively superseded by the R1b-P312 (S116) modal.

If I compare the modal for P312 at 111 STRs with its three largest subclades I get the following differences from P312 for the major subclades of P312:

DF27 (CDYa=37 534=16 710=34 714=25 712=22)

U152 (456=15 549=12)

L21 (449=30 CDYa=37 714=25 522=11 712=20)

Except 522, these are all pretty fast STRs.  I some cases, such as 456, the modal teeters. U152's 15 at 456 is only slightly in the lead over the value of 16. In the case of L21, its 16 modal is only slightly in the lead over 15.

There is just not a lot of difference, even at 111 STRs.  
....
I'll go check U106 so we can compare it as well.
« Last Edit: June 15, 2012, 12:42:17 PM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
JeanL
Old Hand
****
Offline Offline

Posts: 425


« Reply #218 on: June 15, 2012, 12:56:54 PM »

Of course they are, but tripling or quadrupling the number of STR experiments with long haplotypes will reduce the uncertainty and with Ken Nordtvedt's tools we actually have the uncertainty measured and reported.

Yes, but that would be valid under the assumption that it is quantity of STR what matters and not choice of STR.  Of course if we are to assume that the uncertainty arises only from the data and not the mean mutation rate used per haplotype, then more is better, but if we take into account the choice of STR, then more isn’t always better.
Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #219 on: June 15, 2012, 01:49:52 PM »

Of course they are, but tripling or quadrupling the number of STR experiments with long haplotypes will reduce the uncertainty and with Ken Nordtvedt's tools we actually have the uncertainty measured and reported.

Yes, but that would be valid under the assumption that it is quantity of STR what matters and not choice of STR.  Of course if we are to assume that the uncertainty arises only from the data and not the mean mutation rate used per haplotype, then more is better, but if we take into account the choice of STR, then more isn’t always better.

This is on the STR Wars thread as well, I compare variance both with a set of mixed speed markers and then with a subset that meet a 7000 year linear duration according to Marko Heinila. In all cases multi-copy (i.e. CDY, 464, 459, etc.) and null potential STRs (i.e.439, 425) are thrown out. In other words, in variance comparisons I have a rational, quality driven approach to using STRs. As far as Ken's TMRCA tool, he is essentially using everything but multi-copy STRs and requires the user to adjust nulls to an incremental GD.
« Last Edit: June 15, 2012, 01:51:23 PM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
Richard Rocca
Old Hand
****
Offline Offline

Posts: 523


« Reply #220 on: June 15, 2012, 02:02:30 PM »

Jean, any chance you can take Mike up on his offer and move this elsewhere? I really do want to read about L51 when I click on this thread.
Logged

Paternal: R1b-U152+L2*
Maternal: H
alan trowel hands.
Guru
*****
Offline Offline

Posts: 2012


« Reply #221 on: June 15, 2012, 06:55:52 PM »


I tend to suspect U106 arose in NE Europe from an L11* line that moved there from somewhere to the south or west a couple of centuries earlier.  I would link this to the bell beaker culture.  Well certainly every bit as much as I would link P312 to it.  

Here is a very useful paper on beakers peripheral elements
 http://www.aegeobalkanprehistory.net/article.php?id_art=10

The only exception I would take to what you posted is that we have two Bell Beaker y-dna results, and both were U106-. Of course, we don't know whether or not they were P312+. They could have simply been L11+ and no further up the tree.

I know two are not many, but they were found in what is today U106-rich country, and neither was U106+.

I tend to hold Bell Beaker accountable for Italo-Celtic, and I don't think U106 had any kind of a hand in that. So, yeah, I would give U106 a northeastern origin and have it caught up in the Corded Ware/Nordic Bronze Age thing fairly early.

Do you think U106 was in Corded Ware from its inception?  If so, what predecessor culture might it have come from?  I guess I'm just asking how and from what direction you think U106 got involved?

Well the current mainstream theories for beaker and corded ware give them very different origins.  IF these models are correct then there is no way U106 can be corded ware and P312 beaker. It just doesnt makes sense given L11 SNP just above both.  Also, you tend to get dates around 2500BC for P312 but somewhat younger for U106.

My favourite option is that around 2500BC some beaker groups headed north-east from perhaps central Europe and as a consequence an L11* lineage arrived in east Germany/Poland.  That lineage remained there and few hundred years later U106 was born among this L11* group who had moved to Poland.

The only way L11 can be both beaker and corded ware is if we turn the clock back to the Dutch model which actually derived beaker from corded ware/single grave people.  Personally I would rule nothing out.  I am have been long enough in this game to feel very cautious about any conclusions about the beaker culture.  There have been so many theories, changes in ideas, U-turns etc that I am not the sort of person who thinks the newer the model the more likely it is correct.  You would think that might be true with the increase in data over time but I have seen so much changes in ideas on beakers, even turning full circle.  Botrtom line is they moved too fast for RC dating to sort out the detail if you take into account all the caveats involved.  I think however that ancient DNA will solve this soon.  Once we have the same sort of numbers of beaker burials tested as we have for the earlier Neolithic then things may become apparent.  However I will reiterate that the majority of archaeologist do not see corded ware and beaker as comming from a common root in a way that would parallel P312 and U106 coming from a common root.  However, they could be wrong.  While Corded Ware origins seem to be nailed down, Beaker to me remains an enduring archaeological mystery where things just do not feel right to me and I think we are mssing a crucial part of the picture, perhaps in eastern Europe.    
« Last Edit: June 15, 2012, 06:57:10 PM by alan trowel hands. » Logged
rms2
Board Moderator
Guru
*****
Offline Offline

Posts: 5023


« Reply #222 on: June 15, 2012, 09:51:17 PM »

What I think is that we must be flexible about age estimates, both for the archaeological cultures and for the y haplogroups.
Logged

Jean M
Guru
*****
Offline Offline

Posts: 1253


« Reply #223 on: June 16, 2012, 07:22:37 AM »

Well the current mainstream theories for beaker and corded ware give them very different origins. .... The only way L11 can be both beaker and corded ware is if we turn the clock back to the Dutch model which actually derived beaker from corded ware/single grave people.

Not so Alan. Your  "current mainstream theories" are out of date. Frankly what the majority of archaeologists who are not keeping up with developments think is irrelevant.  It is not evidence. Harrison and Heyd established not only the sequence from Yamnaya to Bell Beaker, but the influence of Yamnaya on a whole slew of other cultures including Corded Ware. There were separate routes from Yamnaya to the various daughter cultures.

The old idea of Corded Ware as a local descendant of previous cultures seemed logical at the time to archaeologists looking at a sequence of pottery shapes in their own back yards from Ertebolle (and related) through TRB to Corded Ware. We now know from ancient DNA that the TRB people were not descended from local hunter-gatherers. From a study of skull shape, they most likely came from the Balkans. It follows that the Corded Ware people were not direct descendants of the TRB people either, since they look more like hunter-gatherer types, as would be the case  if they had moved north from the steppe. Culturally of course they carried influences from Yamnaya.

On the specific problems of dating Corded Ware see Wlodarczak 2009, which I just added to the Mini-library. It is completely impossible for Bell Beaker to be derived from Corded Ware since both are of the same age. In fact Wlodarczak 2009 thinks the odd outliers among the early dates for Corded Ware should be ignored, and dates from dendrochronology in Switzerland are more reliable and precise, which would place the start of Corded Ware later than the start of BB.  

I know descent of BB from CW was a favourite idea of yours, but it does not wash and is not about to be resurrected.  
« Last Edit: June 16, 2012, 09:59:12 AM by Jean M » Logged
rms2
Board Moderator
Guru
*****
Offline Offline

Posts: 5023


« Reply #224 on: June 16, 2012, 10:29:13 AM »

What I was thinking, and did not really have time to post last night, was that U106 was in the northeast and became part of the Corded Ware culture, whether as part of its origin or under its influence. If Bell Beaker came from the east (specifically the Hungarian Plain) via Yamnaya influences, then P312 provided the bulk of it because it was farther south than U106 at the time. If BB came from Iberia, then P312 was the bulk of it because P312 had also gotten farther west than U106 by that time and is just in general a bigger clade than U106.
Logged

Pages: 1 ... 7 8 [9] 10 11 12 Go Up Print 
« previous next »
Jump to:  


SEO light theme by © Mustang forums. Powered by SMF 1.1.13 | SMF © 2006-2011, Simple Machines LLC

Page created in 0.173 seconds with 19 queries.