World Families Forums - TMRCA calculations

Welcome, Guest. Please login or register.
September 16, 2014, 12:38:50 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)
| | |-+  TMRCA calculations
« previous next »
Pages: 1 ... 4 5 [6] 7 Go Down Print
Author Topic: TMRCA calculations  (Read 8494 times)
Jdean
Old Hand
****
Offline Offline

Posts: 678


« Reply #125 on: May 31, 2012, 06:47:32 PM »


One can imagine many reasons that might affect mutation rates, like autosomal genes and their activity (epigenetics),  different mtdna haplogroups producing different levels of oxidative stress, and so on.  The question would be if it is somehow possible to predict mutation rates better is some specific group than by using average over everything: this is a rather complex issue.

But of course non of these would stay exclusively in a Y line.
Logged

Y-DNA R-DF49*
MtDNA J1c2e
Kit No. 117897
Ysearch 3BMC9

MarkoH
Member
**
Offline Offline

Posts: 20


« Reply #126 on: May 31, 2012, 07:36:39 PM »

But of course non of these would stay exclusively in a Y line.

but would still make rate estimation more difficult and result in excess "variance of variance" over constant mutation rate model.
Logged
Jdean
Old Hand
****
Offline Offline

Posts: 678


« Reply #127 on: May 31, 2012, 07:46:36 PM »


but would still make rate estimation more difficult and result in excess "variance of variance" over constant mutation rate model.

Presumably the easiest way to incorporate this uncertainty would be to apply a fudge factor to the confidence range ?
« Last Edit: May 31, 2012, 07:49:42 PM by Jdean » Logged

Y-DNA R-DF49*
MtDNA J1c2e
Kit No. 117897
Ysearch 3BMC9

MarkoH
Member
**
Offline Offline

Posts: 20


« Reply #128 on: June 01, 2012, 01:43:40 AM »

Presumably the easiest way to incorporate this uncertainty would be to apply a fudge factor to the confidence range ?

yes, I have toyed with the idea that there is a fast and large variation in mutation rates (rather than slow and relatively small like haplogroup dependence); the paper mentioned before seems more or less consistent with the idea.  However,  I haven't figured out how to estimate the fudge factor in an appealing way from the STR datasets.
Logged
Jdean
Old Hand
****
Offline Offline

Posts: 678


« Reply #129 on: June 01, 2012, 04:57:55 AM »


yes, I have toyed with the idea that there is a fast and large variation in mutation rates (rather than slow and relatively small like haplogroup dependence); the paper mentioned before seems more or less consistent with the idea.  However,  I haven't figured out how to estimate the fudge factor in an appealing way from the STR datasets.

The one thing I'm a little dubious about with Ken's spreadsheet is the closeness of the confidence intervals that it can produce, something that increases that sounds right to me.

We were discussing differences of opinion in the no. of yrs in a generation not long ago. I decided to incorporate this into Ken's calculations by using a high estimate for the upper range and a low one for lower but was told that this may be unacceptable to mathematicians as the output becomes non liner, any thoughts ?
Logged

Y-DNA R-DF49*
MtDNA J1c2e
Kit No. 117897
Ysearch 3BMC9

MarkoH
Member
**
Offline Offline

Posts: 20


« Reply #130 on: June 01, 2012, 05:33:56 PM »

The one thing I'm a little dubious about with Ken's spreadsheet is the closeness of the confidence intervals that it can produce, something that increases that sounds right to me.

Even with significant amount of data, two-sample variance error limits look most reasonable to me.

Quote
We were discussing differences of opinion in the no. of yrs in a generation not long ago. I decided to incorporate this into Ken's calculations by using a high estimate for the upper range and a low one for lower but was told that this may be unacceptable to mathematicians as the output becomes non liner, any thoughts ?

There is very little data about this. Mutation rate estimates themselves apply to modern populations with the idea that haplotype based estimation methods assume relative rates that did not change historically (nothing is assumed about historical absolute rates).  The natural length of generation then comes from modern populations. Studies find age dependece Gusmao (2005) and Goedbloed (2009) which in general would make generations to years conversion more complex.  Recently some discussion on rootsweb list claimed that such dependence has been looked for but never been observed (!) and that generations is more fundamental than years. My view is the opposite of that: If the number of cell divisions would matter, the logical assumption is that mutation rate increases with length of generation so that the mutations/year also increases. This is more or less the exact opposite of the the common view.  It is then not clear how the historical length of generation should be used (even if known) to increase accuracy.

« Last Edit: June 01, 2012, 05:36:24 PM by MarkoH » Logged
Jdean
Old Hand
****
Offline Offline

Posts: 678


« Reply #131 on: June 02, 2012, 05:51:17 AM »

The one thing I'm a little dubious about with Ken's spreadsheet is the closeness of the confidence intervals that it can produce, something that increases that sounds right to me.

Even with significant amount of data, two-sample variance error limits look most reasonable to me.

Quote
We were discussing differences of opinion in the no. of yrs in a generation not long ago. I decided to incorporate this into Ken's calculations by using a high estimate for the upper range and a low one for lower but was told that this may be unacceptable to mathematicians as the output becomes non liner, any thoughts ?

There is very little data about this. Mutation rate estimates themselves apply to modern populations with the idea that haplotype based estimation methods assume relative rates that did not change historically (nothing is assumed about historical absolute rates).  The natural length of generation then comes from modern populations. Studies find age dependece Gusmao (2005) and Goedbloed (2009) which in general would make generations to years conversion more complex.  Recently some discussion on rootsweb list claimed that such dependence has been looked for but never been observed (!) and that generations is more fundamental than years. My view is the opposite of that: If the number of cell divisions would matter, the logical assumption is that mutation rate increases with length of generation so that the mutations/year also increases. This is more or less the exact opposite of the the common view.  It is then not clear how the historical length of generation should be used (even if known) to increase accuracy.



Presumably the discussions on rootsweb (could you provide a link) disagree with the findings in these studies rather than ignore them ?

I can remember Chandler claiming no connection between father age and increased mutations had been found but don't think he quoted studies that backed this up.
Logged

Y-DNA R-DF49*
MtDNA J1c2e
Kit No. 117897
Ysearch 3BMC9

spanjool
Member
**
Offline Offline

Posts: 38


« Reply #132 on: June 03, 2012, 04:03:36 AM »

Charles Kerchner found higher mutation rates within surname projects then outside.
The same can be expected in geographic cluster due to sampling effects and less gene flow.
Downstream subclades will also see a nonlinear change in mutation rates.

Another factor is population growth. Also that has been irregular as well in time: since 1400 year is very steep increase in population as in geographic areas.

(Pre)-Neolithic demographic studies have shown different speed of population growth  in different areas within the same timeframe. F.e. the Netherlands saw increase in their poplution while a the same time the Isles had a decline.

Boucquet-Appel has done studies on these aspects of effective population sizes.

Including this information in the TRMCA calculation will stratify them in time and geographic frames.

Hans


Logged

R1b-Z220*
MarkoH
Member
**
Offline Offline

Posts: 20


« Reply #133 on: June 03, 2012, 02:31:19 PM »

I can remember Chandler claiming no connection between father age and increased mutations had been found but don't think he quoted studies that backed this up.

I was referring to that....   Perhaps bigger studies seem to find some sings of a link, smaller ones don't.

Logged
MarkoH
Member
**
Offline Offline

Posts: 20


« Reply #134 on: June 03, 2012, 03:14:41 PM »

Charles Kerchner found higher mutation rates within surname projects then outside.
The same can be expected in geographic cluster due to sampling effects and less gene flow.
Downstream subclades will also see a nonlinear change in mutation rates.

Another factor is population growth. Also that has been irregular as well in time: since 1400 year is very steep increase in population as in geographic areas.

(Pre)-Neolithic demographic studies have shown different speed of population growth  in different areas within the same timeframe. F.e. the Netherlands saw increase in their poplution while a the same time the Isles had a decline.

Boucquet-Appel has done studies on these aspects of effective population sizes.

Including this information in the TRMCA calculation will stratify them in time and geographic frames.

You refer to the effects similar to the observation that most SNP's are relatively recent?  Though some population level diversity measures are affected, many hobbyist methods are not distorted by this.





« Last Edit: June 03, 2012, 03:50:35 PM by MarkoH » Logged
spanjool
Member
**
Offline Offline

Posts: 38


« Reply #135 on: July 26, 2012, 04:28:58 AM »

In a recent published article in Plos One by Rocca et al returned the observation that age estimates to high up in the clades is hampered by a too diverse mix of haplotypes.
A conflict that does not arise (the article states) in estimates done with haplotypes within more downstream SNP defined haplogroups.
Exactly the statements that in the beginning started this topic.
Hans
Logged

R1b-Z220*
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #136 on: July 26, 2012, 07:35:24 AM »

In a recent published article in Plos One by Rocca et al returned the observation that age estimates to high up in the clades is hampered by a too diverse mix of haplotypes.
A conflict that does not arise (the article states) in estimates done with haplotypes within more downstream SNP defined haplogroups.
Exactly the statements that in the beginning started this topic.
Hans

I have been trying to estimate the TMRCA for R - Z253.  There appear to be at least 3 groups in the current set of data:  1. One set with a 13 at 388 2. One set with a 13 at 426and 3. A set with more usual allele values (13,24,14,10 ...).

One difficulty, as always, is the limited number of dys loci measured.  Most are only 37. Then, as I mentioned at the beginning of the thread, there are some downstream, younger, haplotypes.(In my current analysis I bypass these issues by using only R - Z253 entries, no downstreamers)

I can create a modal for each set of the above entries (3, 7 and about 11 respectively) but what kind of arithmetic do I use to converge these modals?  More importantly, is modal convergence a correct way to make a TMRCA estimate?
Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #137 on: July 26, 2012, 11:13:22 AM »

In a recent published article in Plos One by Rocca et al returned the observation that age estimates to high up in the clades is hampered by a too diverse mix of haplotypes.
A conflict that does not arise (the article states) in estimates done with haplotypes within more downstream SNP defined haplogroups.
Exactly the statements that in the beginning started this topic.
Hans

Could you explain this a little further? The problem occurs when and when is it not a problem?
« Last Edit: July 26, 2012, 11:13:40 AM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #138 on: July 26, 2012, 12:31:15 PM »

This was the point I was trying to make when I started this thread.  If you include lower SNP entries in your calculation of TMRCA, then you bias the estimated TMRCA to the lower SNP.

A good example is Z253 which is antecedent to L226 I believe.  You'll bias your estimate to the younger side when you include L226, even though it also is Z253.

I may have backed off later on in the discussion but that was my initial idea and I still think its right.  

This may be the problem with a large group  such as R -L21, where this logic would argue only use entries that are L21, not any lower SNP's present or you will bias your estimate.  You have all the data you need to explore this issue?

I recall my logic as all the L226 are descended from one man.  At that point that man is the age limit for this group.  The next group in the hierarchy would have the same situation.  Just because they have the higher SNP, doesn't mean you include them in that calculation.  To take an example, you can't find P - 312's age by using all the entries from lower(younger) SNP's, you simply will average down.  JMHO.
« Last Edit: July 26, 2012, 04:32:31 PM by ironroad41 » Logged
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #139 on: July 26, 2012, 12:49:38 PM »

I've been mulling this over the weekend and I have convinced myself, that is the only way to make sense out of this.  This is why, I believe, all TMRCA's have appeared to be too young, they were a "gemischt" of subsequent, sometimes unfound/unknown SNP's which confounded the calculation.  JMHO!

Why do you think all the TMRCA calcs 'appear' too young ?

If you would like an upper boundary for L21 calculate the interclade age for P312, you don't even have to use L21 people in the calculation.



I have been looking at this question for quite a while and I think I can provide a better answer.  On other threads I have shown, based on work done by Leo Little, that gene diversity is not a good measure of age and further there appears to be a relationship between modal value and mutation rate.  In reality, I'm not sure how you distinguish between higher diversity or faster mutation rate for a dys loci?

re: Interclade calculations, there may be a problem if the data is sparse for some SNP's relative to others.  If you look at Markos intraclade and interclade estimates for many of the Hgs, you will see large gaps in data where there was not sufficient data to make an interclade.  Its like the people who lived from about 9000 BC to 6000 BC were wiped off the map (p.s. I think they were).  Its just not R1b its some of the other Hgs also.  The few people who survived and propagated are swamped out by the younger entries.

I'll say it again, between 9K and 6K BC or so, Doggerland was the best place to live in Western Europe based on climate and food supply.  I believe there was a large contingent of the extant Hgs. present there. The Storeggae Tsunami changed all that and set the populations back for several thousand years and redistributed the Hgs. to higher elevations for safety.  This was where L21, U106 and U152 along with some I's, G's and J's and possibly E's lived at that time.
« Last Edit: July 26, 2012, 12:51:02 PM by ironroad41 » Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #140 on: July 26, 2012, 06:04:19 PM »

This was the point I was trying to make when I started this thread.  If you include lower SNP entries in your calculation of TMRCA, then you bias the estimated TMRCA to the lower SNP....

Whether there is a lower level SNP or not, any intraclade TMRCA estimate can be biased by an unbalance representation by any one of the downstream related groups.

This only underlines the value of interclade estimates. If you do interclades for pairs of known (SNP marked) subclades below and above the target group you are trying to estimate... you can fence in the age estimate for that target group.

In other if we were looking at DF21, it would be great to have interclade estimates for L21&U152, L21&DF27, then DF21&L513, DF21&DF49, DF21&Z253, then P314.2&Z246, Z246&L720, Z246&S190, P314.2&L720, etc.

I think if you can look at all of these on one piece of paper/graphic, you'll have a pretty good feel for the true age of DF21.

The only real outstanding issue is mutation rates, evolutionary versus germ-line, etc.  Some would say STR selection also, but you can deselect the ones you think are badly behaved.  ... try it both ways, with and without the concerning STRs although for my money if you have enough STRs you don't need to worry about that.
« Last Edit: July 26, 2012, 06:04:47 PM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #141 on: July 27, 2012, 07:10:23 AM »

This was the point I was trying to make when I started this thread.  If you include lower SNP entries in your calculation of TMRCA, then you bias the estimated TMRCA to the lower SNP....

Whether there is a lower level SNP or not, any intraclade TMRCA estimate can be biased by an unbalance representation by any one of the downstream related groups.

This only underlines the value of interclade estimates. If you do interclades for pairs of known (SNP marked) subclades below and above the target group you are trying to estimate... you can fence in the age estimate for that target group.

In other if we were looking at DF21, it would be great to have interclade estimates for L21&U152, L21&DF27, then DF21&L513, DF21&DF49, DF21&Z253, then P314.2&Z246, Z246&L720, Z246&S190, P314.2&L720, etc.

I think if you can look at all of these on one piece of paper/graphic, you'll have a pretty good feel for the true age of DF21.

The only real outstanding issue is mutation rates, evolutionary versus germ-line, etc.  Some would say STR selection also, but you can deselect the ones you think are badly behaved.  ... try it both ways, with and without the concerning STRs although for my money if you have enough STRs you don't need to worry about that.

I believe your first comment agrees with my premise?

You say you can hem in with interclades.  What I observe in Markos R1b data is that L11 and older are at least 11K BP.  All the nearest sublclades are 4+ and younger.  You can't apparently hem them in?  You reach an intraclade maximum thats always lower than interclade and the interclade is almost a binary number 11k or 4k.  That data makes no sense?  I admit Marko is not using variance, but I don't think this is part of the issue here.  
« Last Edit: July 27, 2012, 07:11:17 AM by ironroad41 » Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #142 on: July 27, 2012, 11:41:29 AM »

This was the point I was trying to make when I started this thread.  If you include lower SNP entries in your calculation of TMRCA, then you bias the estimated TMRCA to the lower SNP....

Whether there is a lower level SNP or not, any intraclade TMRCA estimate can be biased by an unbalance representation by any one of the downstream related groups.

This only underlines the value of interclade estimates. If you do interclades for pairs of known (SNP marked) subclades below and above the target group you are trying to estimate... you can fence in the age estimate for that target group.

In other if we were looking at DF21, it would be great to have interclade estimates for L21&U152, L21&DF27, then DF21&L513, DF21&DF49, DF21&Z253, then P314.2&Z246, Z246&L720, Z246&S190, P314.2&L720, etc.

I think if you can look at all of these on one piece of paper/graphic, you'll have a pretty good feel for the true age of DF21.

The only real outstanding issue is mutation rates, evolutionary versus germ-line, etc.  Some would say STR selection also, but you can deselect the ones you think are badly behaved.  ... try it both ways, with and without the concerning STRs although for my money if you have enough STRs you don't need to worry about that.

I believe your first comment agrees with my premise?

You say you can hem in with interclades.  What I observe in Markos R1b data is that L11 and older are at least 11K BP.  All the nearest sublclades are 4+ and younger.  You can't apparently hem them in?  You reach an intraclade maximum thats always lower than interclade and the interclade is almost a binary number 11k or 4k.  That data makes no sense?  I admit Marko is not using variance, but I don't think this is part of the issue here.  

I've never said that there are no biases in intraclade TMRCA calculations.  No data set is perfect unless it is complete and representative. All of that does not mean that statistical calculations are worthless. That is much of the purpose of probability statistics, to objectively measure and analyze while reporting and controlling error.

I disagree with your observation that according to Markos Heinila's data that L11 is at least 11K ybp in age. What in the Markos data indicates that to you?  Please be specific.

Are you confusing (again) his interclade pairing where since L11 has no direct peer (brother) under L51, Markos show an uncle (R1b-M73) for an interclade pairing with L11 (P310 in his chart)?  We've been through all of this before.  That L11-M73 interclade age of 11K ybp is not directly useful. Notice it is the same as the M269-M173 interclade age, also 11K ybp.  All we can say from Markos' chart on this is L11 can be no older than the age of L51, which can be no older than the age of L23, which can be no older than M269 and M269 can be no older than its interclade MRCA with M73.... which is 11K ybp.

Markos also shows that the U106-P312 interclade MRCA is 4.5K ybp.

Given the limited data for upstream branches of L11, all we can say from Marko's chart is that L11 must be younger than 11K and older than 4.5K ybp.

You've got it backwards when you say "L11 and older are at least 11K BP." Instead L11 has to be younger not older, according to Markos' report.

If I have it wrong, please be specific and I'll retract what I have said and apologize.  If not, please quit presenting your confused view on this. Others reading a long may believe it.
« Last Edit: July 27, 2012, 11:53:16 AM by Mikewww » Logged

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

Posts: 425


« Reply #143 on: July 27, 2012, 01:32:50 PM »

I've never said that there are no biases in intraclade TMRCA calculations.  No data set is perfect unless it is complete and representative. All of that does not mean that statistical calculations are worthless. That is much of the purpose of probability statistics, to objectively measure and analyze while reporting and controlling error.

Mike I know your reply was directed at Robert’s comment, but I’m gonna go ahead if you don’t mind and give you my two cents on the subject matter. The statistical calculations might not be worthless but a lot of intrinsic assumptions made to do such calculations are highly misleading. For example, all calibration performed thus far has been on a relatively recent time frame, so yes mixed sets of STRs appear to work and predict TMRCA for pedigrees with recent known paper trail, although even for recent pedigrees the TMRCA often times fall within the margin of errors, so there is still an error. The positive argument for a possible confirmation of the age of R1b-M269 in the time frame of 4-5 kya is its presence in the German Beaker site dated to ~4.2-4.8 kya(I’m not sure if I got that date right, so please do not judge me on it), and its subsequent absence in farming sites in Europe(Treilles-France 5kya, Avellanar-Catalonia-Spain 7 kya, and LBK-Germany). However the aDNA data thus far is so scarce that it is borderline dangerous to make any meaningful conclusions based upon it, nonetheless, yes the aDNA data lends some credibility to germ-line dates based upon R1b findings in Germany, however its findings in Germany sets a lower bound for the date of the SNP, meaning that R1b-M269(xU106) is at least 4000-5000 ybp, and it was present in Europe at that time already.  However, it is also aDNA which goes to show the inability of variance germ-line dating, as it can be readily seen that the germ-line dating of E-V13 far undermines its true age, so once again, this makes it very clear that everything is far from proven, and that aDNA can still change our hypothesis very significantly.

Are you confusing (again) his interclade pairing where since L11 has no direct peer (brother) under L51, Markos show an uncle (R1b-M73) for an interclade pairing with L11 (P310 in his chart)?  We've been through all of this before.  That L11-M73 interclade age of 11K ybp is not directly useful. Notice it is the same as the M269-M173 interclade age, also 11K ybp.  All we can say from Markos' chart on this is L11 can be no older than the age of L51, which can be no older than the age of L23, which can be no older than M269 and M269 can be no older than its interclade MRCA with M73.... which is 11K ybp.

Markos also shows that the U106-P312 interclade MRCA is 4.5K ybp.

Given the limited data for upstream branches of L11, all we can say from Marko's chart is that L11 must be younger than 11K and older than 4.5K ybp.

You've got it backwards when you say "L11 and older are at least 11K BP." Instead L11 has to be younger not older, according to Markos' report.
 

I feel kind of uncomfortable when words such as “it has to”, or “must be younger”, no in fact in order to make such statements one ought to be sure that one's calculations must be or have to be correct, which is something that is far from certain as I mentioned before. Now, I’m gonna do a public service here and provide the formula that Dr.Nordtvedt uses for his interclade calculations and try to explain what assumptions it intrinsically makes.



http://2.bp.blogspot.com/_ro2ijOk8JWc/SJc8K3atqVI/AAAAAAAAAF0/wJ6XDxfuvbM/s320/interclade.jpg

The terminology is as follows:

NA and NB represent the sample size of each one of the haplogroups being analyzed.(i.e. P312,U106/J1,J2/etc) x and y represent the allele value(that is the repeat number) of each locus from each one of the data sets. Finally mu represents the mutation rate. So by dividing the left side by 2mu we get the interclade age in generations. Now here are some of the intrinsic assumptions made by the formula:

The mutational process behaves in such a way that back mutations should “cancel each other out”, however assuming that both haplogroups coalesce to a single repeat number, means that since the time of coalescence the back mutations in each one of the haplogroups have followed the same process such that the same number or similar number of back mutations have occurred in each haplogroup. That is if we are comparing a given STR and one haplogroup has value of 12 and the other has value of 14, the number of actually mutations that have occurred can vary widely from simply 2 single-step mutations, especially if we are looking at TMRCA in a scale of 5000+ ybp and using STRs with mutation rates of 10-3 mutations per generation. The other problem is that mutation rate depends on the repeat value, locus with higher repeat values mutate faster than those with slower repeat values, so how can one use a constant mutation rate, when in fact the mutation rate is actually a function of the repeat value?
« Last Edit: July 27, 2012, 01:46:56 PM by JeanL » Logged
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #144 on: July 27, 2012, 01:43:44 PM »

I'll let marko reply for me.  This is cc'd from a discussion about E Hg and some calculations Marko make for that Hg.
 As with any analysis that is true to form, there is always some wiggle room.  In Marko’s words:
 
 “My estimate for the appearance of E1a1 (is) between 3,800 and 24,000 years,
 which is not very precise.....This could be improved from both ends. The
 upper limit could perhaps be reduced with comparison with E1a*.
 My program does not do that automatically due to (the) complicated nature
 of the paragroups (negative SNP test data is often unavailable).  My guess would then be that if N89142, 84685, N31281 would prove to be E1a1s as well, the limits for the origin of E1a1 are changed to 4,800 --- 10,000…..provided that 149788, 110410, 132476, 73259, 132641, 72713, 155287, 120468 are at most E1a*'s. The guess for the E1a* closest to E1a1 is 120468.”

This gives Markos estimation of the accuracy of his estimates and how dependent on available entries his numbers are.  What I am concerned about is the "continuity" of his estimates.  Marko has admitted there is a bottleneck or whatever you want to call it in the data between L11 and lower sub clades.

Maybe I am confused about what Marko is really saying, old men get funny that way.  But, what is clear is that there is a loss of data, maybe significant which may make tree construction difficult.

I have mentioned my personal problem with my ancestor from scotland and no one near him to go further back in time.  There are three of us now in Z 253 like me, but how do I know that any convergence I might make is just to a common ancestor, not a founder?  It is not apparent to me that I can rewire up Z 253 with the entries I now have?

Since we're being testy here, I am still eagerly awaiting to read your thoughts re: my comments on diversity and modal value/mutation rate dependency.
« Last Edit: July 27, 2012, 02:58:10 PM by ironroad41 » Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #145 on: July 27, 2012, 10:12:19 PM »

.

... That L11-M73 interclade age of 11K ybp is not directly useful. Notice it is the same as the M269-M173 interclade age, also 11K ybp.  All we can say from Markos' chart on this is L11 can be no older than the age of L51, which can be no older than the age of L23, which can be no older than M269 and M269 can be no older than its interclade MRCA with M73.... which is 11K ybp.

Markos also shows that the U106-P312 interclade MRCA is 4.5K ybp.

Given the limited data for upstream branches of L11, all we can say from Marko's chart is that L11 must be younger than 11K and older than 4.5K ybp.

You've got it backwards when you say "L11 and older are at least 11K BP." Instead L11 has to be younger not older, according to Markos' report.  

I feel kind of uncomfortable when words such as “it has to”, or “must be younger”, no in fact in order to make such statements one ought to be sure that one's calculations must be or have to be correct, which is something that is far from certain as I mentioned before.

I understand your discomfort and did not mean to intend that anything was absolute. That's why I used the qualifier "according to Markos' report."  In other words, if Marko's report/analysis is correct the age of L11 would be between 11K and 4.5K ybp.

The whole context was my disagreement with Ironroads' position saying L11 was "at least" 11K ybp based on Marko's data, when it shows exactly the opposite....  again, though, I agree there are assumptions like mutation rates, etc. so nothing is absolute.
« Last Edit: July 27, 2012, 10:13:54 PM by Mikewww » Logged

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

Posts: 2964


WWW
« Reply #146 on: July 27, 2012, 10:22:27 PM »

....Now here are some of the intrinsic assumptions made by the formula:

The mutational process behaves in such a way that back mutations should “cancel each other out”, however assuming that both haplogroups coalesce to a single repeat number, means that since the time of coalescence the back mutations in each one of the haplogroups have followed the same process such that the same number or similar number of back mutations have occurred in each haplogroup. That is if we are comparing a given STR and one haplogroup has value of 12 and the other has value of 14, the number of actually mutations that have occurred can vary widely from simply 2 single-step mutations, especially if we are looking at TMRCA in a scale of 5000+ ybp and using STRs with mutation rates of 10-3 mutations per generation. The other problem is...

JeanL, I admit you are beyond me mathematically but I do want to fully understand the two problems you are highlighting.

I think you are saying that one STR's diversity in one haplogroup might be different over the same time period (# of generations) than to a second haplogroup. Is that a correct summary of the first problem you cite?
Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #147 on: July 28, 2012, 08:04:11 AM »

.


Markos also shows that the U106-P312 interclade MRCA is 4.5K ybp.

L11 has to be younger not older, according to Markos' report.  

I feel kind of uncomfortable when words such as “it has to”, or “must be younger”, no in fact in order to make such statements one ought to be sure that one's calculations must be or have to be correct, which is something that is far from certain as I mentioned before.


The whole context was my disagreement with Ironroads' position saying L11 was "at least" 11K ybp based on Marko's data, when it shows exactly the opposite....  again, though, I agree there are assumptions like mutation rates, etc. so nothing is absolute.
[/quote]

Well, MIke you are the "guru".  I'm tired of your needling and traps.  For me this is a pastime in which I can learn.  Obviously as a guru, you already know everything, so there is little you can or want to learn.  You are on "quicksand" as far as I can tell, you really don't understand the math and you are trusting someone else to prop you up.  You're in good hands now, Jean L might help you understand what is really involved here.  For your sake, I hope so.

p.s.  You say that L11 has to be younger that 11k.  You didn't read markos comments I published.  It all depends on the nodes (entries) you have available to work with.  It can really go either way.  If you really understood what these estimates mean, you'd understand that.
« Last Edit: July 28, 2012, 09:15:42 AM by ironroad41 » Logged
stoneman
Old Hand
****
Offline Offline

Posts: 141


« Reply #148 on: July 28, 2012, 09:45:03 AM »

Ironroad41
I have a few SNPs that no-one else in Western Europe has and Ireland is the origin. I am U106 by the way.
Logged
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #149 on: July 28, 2012, 09:53:23 AM »

They may be personal SNP's.  I am not expert in SNP nomenclature, D. Reynolds, et.al. may be able to better comment.  My comments re: STR's is that I have some very unusual allele values for several of my dys loci.  Who knows has a thread on U106 right now that may be of interest to you?
« Last Edit: July 28, 2012, 09:54:00 AM by ironroad41 » Logged
Pages: 1 ... 4 5 [6] 7 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 0.113 seconds with 18 queries.