Page 1 of 2 12 LastLast
Results 1 to 10 of 13

Thread: Analysis of R-L21>DF13>FGC5494>FGC5549>FGC5511>A1487

  1. #1
    Registered Users
    Posts
    725
    Sex
    Location
    Texas
    Ethnicity
    English, Irish, German
    Nationality
    US
    Y-DNA (P)
    L21>L226>FGC5639

    England Germany Netherlands France Ireland Switzerland

    Analysis of R-L21>DF13>FGC5494>FGC5549>FGC5511>A1487

    It was found that the BigTree TMRCA estimates may be too recent for this part of the haplotree OR it could be statistical variation is more challenging for this part of the haplotree. The BigTree TMRCA estimates appear to be way too recent as predictable haplogroups almost always fall into the 1500 to 2500 YBP range. The pattern of large numbers of branch equivalents was a red flag as well. I collected data for the following branches. The BigTree has the following TMRCA estimates and branch equivalents:

    FGC5549 - 1550 YBP - no branch equivalents
    FGC5511 - 1460 YBP - 3 branch equivalents (excluding complex area calls)
    A1487 - Not predicted by BigTree - 15 branch equivalents (excluding complex area calls)

    The YSTR signature of A1487 is a very strong nine markers. The statistical model had zero false positives and zero false negatives across 56,000 testers under haplogroup R. Two markers of the signature had a three step variation from the L21 modal values which really isolated the testers. This predictable haplogroup has 28 confirmed testers and 45 predicted testers - total of 73 testers. The signature sizes for FGC5511 and FGC5549 were only three or four markers which falls below the minimum of six markers (which will have lower but acceptable accuracy). There are 35 branches under A1487 yet only 28 confirmed testers were found - so there must be a lot of private testers involved.

    Here is the source file for the analysis (requires EXCEL):

    http://www.rcasey.net/DNA/Temp/A1487...20201116B.xlsx

    Here is the SAPP input file and SAPP output file:

    http://www.rcasey.net/DNA/Temp/A1487_SAPP_Input_TEXTOUT_20201116B.txt

    http://www.rcasey.net/DNA/Temp/A1487_SAPP_Output_20201116B.html


    If you want the png image output instead of the html chart, just remove the TEXTOUT line in the SAPP input file and re-run SAPP.
    Last edited by RobertCasey; 11-16-2020 at 04:07 PM.

  2. #2
    Registered Users
    Posts
    112
    Sex
    Ethnicity
    Irish/Scot/English
    Y-DNA (P)
    ZZ44/BY12234
    mtDNA (M)
    H4a1a1

    Ireland Scotland England
    In the Fitzpatrick DNA Project we have much more data than you have to look at in the A1487block including paper trails that give us benchmarks to gauge the years/SNP's

    All of the Fitzpatricks under A1488 are descendants of the Father or Grandfather of the 1st Baron of Ossory (there are paper trails) so the TRMCA is about the year 1400 to 1480, the TRMCA numbers on the SAPP chart are off by quite a bit. We have 35 BigY 700's under A1488 and there are no Brennan's under A1488 and no Fitzpatricks above A1488. We have a total of 65 Fitzpatrick men under A1488 with STR results and none of them appear to go above A1488. We also have very strong signature markers for the lines under A1488.

    The chart shows a a distance in time of A1496 @ 0 AD to FT12563 @ 1550 AD and there is only 4 SNPs between the two which makes it 388 years/SNP which makes no sense

    How can A1488 and FT12563, who are genetic brothers, be so far apart with one at TRMCA of 0 AD its brother at 1550 AD.

    This is an older not up to date SAPP file that is based on only 111 markers + and it matches the FTDNA block tree perfectly

    https://www.dropbox.com/s/e8ssgdld9k...0Tree.png?dl=0
    Last edited by IanFitzpatrick; 11-17-2020 at 11:56 AM.

  3. #3
    Registered Users
    Posts
    112
    Sex
    Ethnicity
    Irish/Scot/English
    Y-DNA (P)
    ZZ44/BY12234
    mtDNA (M)
    H4a1a1

    Ireland Scotland England
    407990 Edgcombe
    61497 zUnkName
    185279 McMeans
    244269 McMeans


    These kits are not even FGC5494 so not A1487

    They are under ZZ44
    Last edited by IanFitzpatrick; 11-16-2020 at 08:52 PM.

  4. #4
    Registered Users
    Posts
    725
    Sex
    Location
    Texas
    Ethnicity
    English, Irish, German
    Nationality
    US
    Y-DNA (P)
    L21>L226>FGC5639

    England Germany Netherlands France Ireland Switzerland
    Quote Originally Posted by IanFitzpatrick View Post
    407990 Edgcombe
    61497 zUnkName
    185279 McMeans
    244269 McMeans


    These kits are not even FGC5494 so not A1487

    They are under ZZ44
    According to FTDNA reports, these testers are currently predicted to be R-M269

    407990 Edgcombe - L21 project
    61497 McMeins - Minnis project
    185279 McMeans - McMahon project
    244269 McMeans - Scottish project

    However, I agree with you that these are speculative since they are the lowest matches that the model predicted and they are missing key mutations for the multi-step mutations that are part of the A1487 signature. They are valid boundary condition testers that could be A1487. What is your source that these are confirmed ZZ44_1 ?

  5. The Following User Says Thank You to RobertCasey For This Useful Post:

     IanFitzpatrick (11-17-2020)

  6. #5
    Registered Users
    Posts
    112
    Sex
    Ethnicity
    Irish/Scot/English
    Y-DNA (P)
    ZZ44/BY12234
    mtDNA (M)
    H4a1a1

    Ireland Scotland England
    My mistake, I was looking at the wrong marker column 447 vs 449 - DYS449=26 is a ZZ44 signature marker and there is a Edgecombe surname cluster under ZZ44 which confirmed my misread.

    I would be shocked if they are A1487, I have 175 kits under A1487 that are matched with TMRCA calculations to BigY results and there are (7) with DYS390=22 & (4) DYS390=20 none at 23 or 24

    This A1487 chart is based on 111 markers only and again, many of the matches have paper trail connections, I have also derived an A1487 111 marker modal, these TMRCA calculations match up very well with the SAPP calculations on the tree I posted earlier (mutation rates on specific STR markers have been adjusted to the sampling studied and known TMRCA values)

    A1487 Haplotype TMRCA Chart.png
    Last edited by IanFitzpatrick; 11-17-2020 at 03:33 PM.

  7. #6
    Registered Users
    Posts
    725
    Sex
    Location
    Texas
    Ethnicity
    English, Irish, German
    Nationality
    US
    Y-DNA (P)
    L21>L226>FGC5639

    England Germany Netherlands France Ireland Switzerland
    I did indeed miss several valid testers (doing these manually takes too much time as well):

    Added - 10 testers from the Fitzpatrick project
    Added - 2 testers from other projects
    Updated - 5 terminal YSNPs that have been upgraded

    Also, I made other significant change for Robinson/198893. Before, I had him as A1487 negative as I had assumed to be Big Y tested. After looking
    at the YSNP report, he appears to be SNP pack tested. So, I assumed that FGC5511 was ancestral and allowed him to be predicted.

    With all these changes, the signature (over 75 % for positive testers) went from nine markers to ten markers. The constants on this model are
    positive for the genetic distance (this has only happened once after over 100 runs of AcaStat. It remains a perfect 100 % prediction model
    but for the first time, I had inconsistent prediction that where lower genetic distance predicted negative yet higher genetic distance at the
    same signature match was predicted positive. This new version, clipped out several of the "boundary condition" testers. Prediction is only as
    good as the boundary condition testers are well tested. Once more "questionable calls" are tested, the model learns with better information.

    SAPP TMRCA estimates are not very accurate - but I leave these in as people want this information. There is a lot of pressure to provide
    TMRCA estimates, so like you, I have created a unique approach to derive TMRCA dates for YSNP branches. I is by far the most accurate
    model for branches between L226 (500 AD) and the 30 surname clusters under L226. This approach only works with at least 500 testers
    for a predictable haplogroup. I just the average for those where there are no surname clusters which results in too old of estimates
    under surname clusters.

    https://www.youtube.com/watch?v=sKaxanrxBgs

    SAPP really struggles with the earlier branches but excellent for determining larger signatures. Also, I only who has Big Y tested
    via the BigTree - adding negative downstream YSNP branches makes SAPP a lot happier and better.

    I really like your TMRCA summary, it looks a look more consistent than the currently used methods: McDonald/BigTree/YFULL
    and David Vance's YTR based TMRCA. Both of our methods look a lot more rational. Unfortunately, my approach requires
    500 to 1,000 testers but over time and ever increasing sample size, it will improve. Many of our L226 testers are now getting
    zero, one or two private YSNPs. Y700 YSTRs are now revealing more than YSNPs (but you need around 200 Big Y testers just
    to determine the modal values for your haplogroup).

    www.rcasey.net/DNA/Temp/A1487_HG_R_20201117E.xlsx

    www.rcasey.net/DNA/Temp/A1487_SAPP_Input_TEXTOUT_20201117E.txt

    www.rcasey.net/DNA/Temp/A1487_SAPP_Output_20201117E.html

    If you want to exchange TMRCA source files, let me know. I am curious of what kind of an YSNP counting approach that you use.

    BTW - between L226 and our 30 surname clusters, the years per YSNP branch vary from 48 to 342 years per YSNP branch.
    So there is huge statistical variation found across 900 testers and 300 Big Y testers. The average just dropped to 73 years
    per YSNP branch (between L226 and 30 surname clusters). For now, private and branch equivalent YSNP counting works the
    best with smaller sample sizes and YSNP branches below surname clusters. Once I reach the size of R-M222, my TMRCA
    approach for L226 will obtain TMRCA estimates that genealogist want.

  8. The Following User Says Thank You to RobertCasey For This Useful Post:

     IanFitzpatrick (11-18-2020)

  9. #7
    Registered Users
    Posts
    112
    Sex
    Ethnicity
    Irish/Scot/English
    Y-DNA (P)
    ZZ44/BY12234
    mtDNA (M)
    H4a1a1

    Ireland Scotland England
    This group is very unique in the amount of testing done is well above average, the project has been testing for over 10 years and many were attracted to this line since it is descended from the 1st Baron of Ossory with two confirmed paper trails from branches of two of his sons.

    The methods and logic that are applied here cannot easily be applied to other groups,

    One of the notable differences not found in other blocks, there are two brothers and their sons that have tested and this accounts for the wringing out of so many SNP's that would not usually be found, the SNP's run right to the basement of the block.

    With so many testers under A1488, we have a solid paper trail timelines to use as a measuring scale. The years/SNP are pretty much down to 40 years/SNP. When using this value, the timing of the branching above seems to fit the classic emergence of surnames around 1000 AD.

    What was surprising, and not very well accepted by some, there are no Fitzpatrick lines that are rooted above A1488, only Norman ones. The search for a Fitzpatrick with an SNP above A1488 is partly the reason there are so many testers on a line that has an SNP/STR tree that obviously converges at about 1400-1480 AD.

    There is a well populated Dalton line which appears to go back at least 3-4 SNP's above 1488, unfortunately the Dalton's are being discouraged from taking part in our research by the Admin of their project. There is are Fitzgerald lines, one that branches off A1488 and one with very deep roots back to BY89649 ca 1160 AD.

    The Brannan/Brennan/Branam line is actually 4 distinct lines and with one Irish line and 3 distantly related lines that appear in the Americas under the surname Bramham or variants of which most of them morphed into Brannen/Brennan. These lines appear to be Welsh or there is evidence suggesting Yorkshire origins for one line



    The lack of any other deep Irish surname line under A1487 suggests the 1st Barons line is most likely an NPE from either a Dalton or Fitzgerald and even a Butler line could be considered. Anyone who knows the history of this region in the 1400-1500's would not find this the least bit surprising, these four family's were intertwined in history and marriage.

    Also, this is one of the few Irish dominated lines under FGC5494 but when you take into account the number of tests done on Fitzpatricks under A1488 it is skewing the picture, If you accept this is a DNA line that originates from a Norman invader line, then even this line is not a deeply rooted Irish line . The FGC5494 group as a whole seems to have left its place of origin much later than the other sons of DF13 of which there are Ancient DNA finds showing they were in Ireland circa 2400 BC.

    There is a huge issue with counting SNP's which I know you understand, within a specific line there are a large numbers of missing SNP's which are either out of range or very poor quality to be named. This can bee seen on any block tree or on the Big Tree. The floor of every line should be, within a reasonable number, equal to the bottom. However, we see many lines falling well short of this. The unequal number of SNP's in each individual line is due to a number of factors, some of which will never be resolved in my opinion.

    This is why a strictly statistical approach to resolve the aging of lines using SNP's will create such huge conflicts as you expand from a single line into the surrounding blocks.

    At this time, until we have better resolution and quality of testing, aging lines is a more art than science and without some benchmark such as a paper trail, the accuracy of the calculations is not really one that is very useful.

    I would really be careful about putting trees such as the SAPP tree you have posted out in the public, the range of accuracy on the TMRCA and the way the branching has been placed can lead people down the wrong path very easily. SAPP is a great tool but using STR's to date TRMCA has fooled me so many times that I use it as a guideline only, never the only method to date SNP's. Also, I never use STR's under 111 markers, it is just a waste of time. STR's are just way too unreliable for dating unless the sample size is large enough and at least 111 markers and even then it has too many outliers. Low Marker tests have their place in predicting within reason the haplotype of the tester which is very useful to a research project.

    I use FTDNA STR mutation rates but I have adjusted some markers under this specific group which are obviously creating inaccurate TMRCA calculations. Creating the rates for a block takes a great number of samples and it is constantly being refined. Using this method I can predict most 67 marker results position within 1 or 2 SNP's on this block, it has be shown to be repeatable. It has taken 3 years to refine this model for A1487.

    Speaking of L226, we are expanding our sponsorship into 3 deep rooted Fitzpatrick lines under L226.
    Last edited by IanFitzpatrick; 11-18-2020 at 08:59 PM.

  10. #8
    Registered Users
    Posts
    725
    Sex
    Location
    Texas
    Ethnicity
    English, Irish, German
    Nationality
    US
    Y-DNA (P)
    L21>L226>FGC5639

    England Germany Netherlands France Ireland Switzerland
    Quote Originally Posted by IanFitzpatrick View Post
    The Brannan/Brennan/Branam line is actually 4 distinct lines and with one Irish line and 3 distantly related lines that appear in the Americas under the surname Bramham or variants of which most of them morphed into Brannen/Brennan. These lines appear to be Welsh or there is evidence suggesting Yorkshire origins for one line

    Attachment 41251
    I am very curious about logic behind this spreadsheet summary. If you could post a link to the source EXCEL file or you could send me a copy via email:

    http://www.rcasey.net/DNA/Casey/Contact.html

    I am trying to duplicate your analysis for the Caron Branham cluster. I have two pieces of missing data:

    Brannon/59191 is the only tester that is Y67 - so there will be some adjustments for the missing markers ?

    Brannon/811351 is not found in several of the common projects associated with A1487. Is this tester private ?
    If you found him in sub obscure project, let me know the project name so that I have as much data as possible.

    This genetic cluster appears to have the following signature from the 10 marker A1487 signature (ignoring CDY markers):

    385a (11>9)
    464c (16>17) - backwards mutation of the A1487 signature - pretty weak marker for the A1487 signature.
    460 (11>10)
    714 [25>24]
    452 [30>29]
    525 [20>21]
    650 [19>18]
    513 [11>12] - backwards mutation of the A1487 signature - pretty strong marker for the A1487 signature.

    With respect to SAPP TMRCAs - I believe that David Vance added a switch to turn off reporting of the TMRCA calculations. They do get better with larger haplogroups but are obviously problematic for A1487. A1487 seems to be on the low end of accuracy for these TMRCA estimates. Speaking of other problems with SAPP. I find that mixing Y67 and Y111 is very problematic - however, Y67 only and Y111 only are much better. Y67 vs. Y111 is less markers vs. smaller sample size. Also, SAPP misses the upper parts of the YSTR but this is automation. It does an excellent job of finding larger signatures. Another way to help with increasing SAPP accuracy is adding negative downstream YSNPs. This really restricts SAPP from moving testers down haplotree. I can only use the Big Tree to identify Big Y testers to add these downstream negative values. I wish FTDNA add Big Y flags in their YSNP or YSTR reports (my only source for Big Y testers under A1487).

  11. #9
    Registered Users
    Posts
    112
    Sex
    Ethnicity
    Irish/Scot/English
    Y-DNA (P)
    ZZ44/BY12234
    mtDNA (M)
    H4a1a1

    Ireland Scotland England
    I am sorry but I did not realize that matrix had kit numbers on it.

    I do not publish kit numbers and sorry but I do not share data due to privacy issues

    Edit to add chart back with kit numbers removed

    3F9AD062-BC2A-4046-A0BE-C9CC9424DEBF.jpeg
    Last edited by IanFitzpatrick; 11-19-2020 at 12:53 AM.

  12. #10
    Registered Users
    Posts
    112
    Sex
    Ethnicity
    Irish/Scot/English
    Y-DNA (P)
    ZZ44/BY12234
    mtDNA (M)
    H4a1a1

    Ireland Scotland England
    Regarding the BY140757 block which we refer to as the "Bran Names". This block circa 1200 AD contains the Bran type names and it has two sons and a few generations of branching between ca. 1200 to 1300 and then large bottlenecks. This block is under tested.

    I have 38 Kits placed under this SNP with almost all of them having a "Bran" name

    Yes, DYS513= 12>11 is a good A1487 modal signature but the back mutation happens under FT70038 which has 5 SNP's under it and the signature happens within that block so its not really a block signature yet, where it happens is TBD

    The FT70038 DYS385=11-14>9-14 is a signature under that block

    DYS464 markers are a throwaway, not much better than CDY in reliability IMO, I typically do not use these markers for dating or tree building.

    DYS714 (25>24) is not a signature for this block, I have it at 50/50 under BY140757 but it is scattered and not a reliable marker in general, this also applies to DYS452, DYS525

    460 (11>10) is not a signature of this block, it happens above this block since it runs into A1496 as well and then (10>11) back mutation is a signature for a SNP under FT11900

    DYS650 (19>18) happens above BY140757

    Here are the only signatures I have for the SNP levels under A1487 that actually have a signature, most do not.

    These are solid signatures

    FT111213 DYS425=12>10
    FT149180 DYS439=12>11
    A1506 DYS459= 9-10>9-9
    FT70038 DYS385=11-14>9-14
    FT12563 DYS481=22>23
    A1488 DYS712= 20>19 & DYS549= 13>14
    BY116564 DYS439= 12>13
    FT206758 DYS439=12>11 & DYS447=25>24
    A1497 DYS460=10>11
    A1489 DYS385=11-14>12-14
    Last edited by IanFitzpatrick; 11-19-2020 at 04:05 PM.

Page 1 of 2 12 LastLast

Similar Threads

  1. Replies: 1
    Last Post: 06-30-2020, 09:02 PM
  2. Replies: 7
    Last Post: 10-19-2019, 09:30 PM
  3. DF13* (L21>DF13*)
    By vineviz in forum DF13
    Replies: 57
    Last Post: 03-22-2018, 04:03 PM
  4. FGC5494 Distribution
    By MBW1986 in forum DF49
    Replies: 0
    Last Post: 01-28-2016, 11:17 PM
  5. Replies: 44
    Last Post: 09-19-2015, 01:11 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •