Go Back   Family Tree DNA Forums > Universal Lineage Testing (Autosomal DNA) > Family Finder Advanced Topics

Family Finder Advanced Topics Advanced discussion about Family Tree DNA's Family Finder Product.

Reply
 
Thread Tools Display Modes
  #31  
Old 31st August 2013, 10:41 AM
Táltos Táltos is offline
FTDNA Customer
 
Join Date: Jan 2012
Posts: 1,166
Quote:
Originally Posted by ursus View Post
To the best of my knowledge, I am 100% Jewish, mostly Ashkenazic, but definitely partially Sephardic. FWIW, PF gives me:

Middle East Bedouin, Bedouin South, Druze, Iranian, Jewish, Palestinian 53.79% ±23.33%
Europe Sardinian, Spanish, Tuscan, Basque 46.21% ±23.33%
Thanks for sharing ursus. Now here is another thing that bugs me about PF, why such a high margin of error for some of you guys? That would make me feel that my results are really way off. But then again perhaps that is their way of being truthful that their Population Finder is not that good.
Reply With Quote
  #32  
Old 31st August 2013, 11:13 AM
NYMark NYMark is offline
FTDNA Customer
 
Join Date: Jan 2012
Posts: 458
I've posted this on another thread, but it's seems worth reiterating. Population Finder will not necessarily show Ashkenazi Jews as 100% Middle Eastern. My half-siblings (different dads) are both 100% Ashkenazi. My half sister's PF is 100% Middle Eastern. My half-brother's is around 45% ME (incorrectly identified as Druze or Palestinian), with an approximately 7% margin of error. I'm half, and mine is around 35% ME (don't recall the exact number off the top of my head) with an over 19% margin of error.

Regardless of your views on Ashkenazi origins, 100% Middle Eastern makes no sense.

@Javelin, I think you're referring to a Swiss match of mine, German Swiss on one side and Italian Swiss on the other; she got something like an 8.9% on the JTest, not on Population Finder.

Quote:
Originally Posted by Táltos View Post
Thanks for sharing ursus. Now here is another thing that bugs me about PF, why such a high margin of error for some of you guys? That would make me feel that my results are really way off. But then again perhaps that is their way of being truthful that their Population Finder is not that good.
Reply With Quote
  #33  
Old 31st August 2013, 11:27 AM
Táltos Táltos is offline
FTDNA Customer
 
Join Date: Jan 2012
Posts: 1,166
Quote:
Originally Posted by NYMark View Post
I've posted this on another thread, but it's seems worth reiterating. Population Finder will not necessarily show Ashkenazi Jews as 100% Middle Eastern. My half-siblings (different dads) are both 100% Ashkenazi. My half sister's PF is 100% Middle Eastern. My half-brother's is around 45% ME (incorrectly identified as Druze or Palestinian), with an approximately 7% margin of error. I'm half, and mine is around 35% ME (don't recall the exact number off the top of my head) with an over 19% margin of error.

Regardless of your views on Ashkenazi origins, 100% Middle Eastern makes no sense.

@Javelin, I think you're referring to a Swiss match of mine, German Swiss on one side and Italian Swiss on the other; she got something like an 8.9% on the JTest, not on Population Finder.
Your right NYMark 100% Middle Eastern makes no sense for Ashkenazi. Though I do believe they have their origins in the Middle East, so they should show some Middle Eastern. (Believe me I don't even want this thread to get into any kind of political nonsense. I have had to read enough of that with my father's haplogroup since receiving my brother's results.)

Anyway, yes they should also show some European as they mixed over time with others in Europe. So if your Ashkenazi, and get the half and half Middle Eastern and European results, why so high a margin of error? This high margin of error also occurs for the Sicilians and Southern Italians when they get higher percentages of Middle Eastern as well.

So I wonder if this would improve if they actually had reference samples called Ashkenazi, and also Sicilian/Southern Italian?
Reply With Quote
  #34  
Old 31st August 2013, 12:10 PM
josh w. josh w. is offline
FTDNA Customer
 
Join Date: Mar 2005
Posts: 2,536
Quote:
Originally Posted by josh w. View Post
I agree that there is a definite southern Italian component among Ashkenazi autosomes. However, my non Jewish signs are from northern Europe. My Mtdna J1c subclade J1c4 has only been found in eastern and northern Europe (including Sweden) although present data is rather skimpy. Equally puzzling is that an eastern European component is only sometimes found in Jewish autosomal studies. One of the answers is that the Ashkenazi population is far from monolithic. Feder found significant regional variation in Ashkenazi Mtdna rates with K1a1b lower among Russian Jews. 'Jewish diseases' are also regional----Tay Sachs is mainly found among Litvaks, Lithuanian Jews. I await my Gedmatch results.
Got my Gedmatch results. I am of Jewish origin but not particularly Near Eastern. For Oracle, the most likely origin is Ashkenazi (or 50% Ashkenazi and 50% Ashkenazi) However for regional sources, the largest component is Mediterranean (Eurogenes, Dodecad, ) or Caucasian ( Harrapa World) The next largest component is Near Eastern or Southwest Asian with non-Mediterranean Europe as the fourth largest component
Thus my PF results are a bit confusing. The results are accurate with respect to ethnicity. However, in terms of regions of origin, lumping Jews with other Near Eastern groups may be an oversimplification. I realize my point has political implications. However, there needs to be more explanation of PF findings. How many who are listed as 100% Jewish have their largest component from the Near East.

Last edited by josh w.; 31st August 2013 at 12:19 PM.
Reply With Quote
  #35  
Old 31st August 2013, 12:22 PM
josh w. josh w. is offline
FTDNA Customer
 
Join Date: Mar 2005
Posts: 2,536
Quote:
Originally Posted by josh w. View Post
Got my Gedmatch results. I am of Jewish origin but not particularly Near Eastern. For Oracle, the most likely origin is Ashkenazi (or 50% Ashkenazi and 50% Ashkenazi) However for regional sources, the largest component is Mediterranean (Eurogenes, Dodecad, ) or Caucasian ( Harrapa World) The next largest component is Near Eastern or Southwest Asian with non-Mediterranean Europe as the fourth largest component
Thus my PF results are a bit confusing. The results are accurate with respect to ethnicity. However, in terms of regions of origin, lumping Jews with other Near Eastern groups may be an oversimplification. I realize my point has political implications. However, there needs to be more explanation of PF findings. How many who are listed as 100% Jewish have their largest component from the Near East.
I realize that there was a Near Eastern influence on the Mediterranean and on the Caucasus. My point is that this influence was already present before the Diaspora.
Reply With Quote
  #36  
Old 31st August 2013, 12:24 PM
Javelin Javelin is offline
FTDNA Customer
 
Join Date: Jan 2009
Posts: 990
Quote:
Originally Posted by josh w. View Post
Got my Gedmatch results. I am of Jewish origin but not particularly Near Eastern. For Oracle, the most likely origin is Ashkenazi (or 50% Ashkenazi and 50% Ashkenazi) However for regional sources, the largest component is Mediterranean (Eurogenes, Dodecad, ) or Caucasian ( Harrapa World) The next largest component is Near Eastern or Southwest Asian with non-Mediterranean Europe as the fourth largest component
Thus my PF results are a bit confusing. The results are accurate with respect to ethnicity. However, in terms of regions of origin, lumping Jews with other Near Eastern groups may be an oversimplification. I realize my point has political implications. However, there needs to be more explanation of PF findings. How many who are listed as 100% Jewish have their largest component from the Near East.
Nobody who is Levantine Near Eastern shows up as 100% Near Eastern -- all have a significant Mediterranean component (Lebanese, Syrians, Palestinians, etc.). It's a little misleading to look at imaginary ancestral components in isolation.
Reply With Quote
  #37  
Old 31st August 2013, 12:28 PM
NYMark NYMark is offline
FTDNA Customer
 
Join Date: Jan 2012
Posts: 458
There are huge problems with Population Finder, no doubt about it. Here's hoping that improvements are coming down the pike.

That said, this is a bit of a damned if you do, damned if you don't situation. 23andMe has been criticized for identifying Ashkenazi as a European population, even though that's clearly accurate for the time frame they're using. Ancestry has similarly been criticized for "European Jewish", which seems to include Sephardic as well but is woefully unreliable whatever it may be measuring.

Personally, I find 23andMe's approach to be the most useful and realistic for a genealogical site. Overall, it seems to be by far the most accurate for Ashkenazim too, although there have been some complaints and some clearly erroneous results.

The fact that full Ashkenazim can range from around 40% (the low end of my bro's margin of error) to 100% Middle Eastern is troubling for two reasons. First, it's potentially misleading; I have corresponded with an Ashkenazi Jew who was convinced their PF results meant they were Mizrahi (and believe me it's hard to dissuade someone who thinks that's what the scientists are telling them.) It's also not at all helpful for people who are of mixed ancestry and are trying to find out their percentages. PF tells such people next to nothing of value. The "Orcadian" category has many of the same issues.


Quote:
Originally Posted by josh w. View Post
Got my Gedmatch results. Just as I expected, I am of Jewish origin but not particularly Near Eastern. For Oracle, the most likely origin is Ashkenazi (or 50% Ashkenazi and 50% Ashkenazi) However for regional sources, the largest component is Mediterranean (Eurogenes, Dodecad, ) or Caucasian ( Harrapa World) The next largest component is Near Eastern or Southwest Asian with non-Mediterranean Europe as the fourth largest component
Thus my PF results are a bit confusing. The results are accurate with respect to ethnicity. However, in terms of regions of origin, lumping Jews with other Near Eastern groups may be an oversimplification. I realize my point has political implications. However, there needs to be more explanation of PF findings. How many who are listed as 100% Jewish have their largest component from the Near East.
Reply With Quote
  #38  
Old 31st August 2013, 01:05 PM
josh w. josh w. is offline
FTDNA Customer
 
Join Date: Mar 2005
Posts: 2,536
Quote:
Originally Posted by Javelin View Post
Nobody who is Levantine Near Eastern shows up as 100% Near Eastern -- all have a significant Mediterranean component (Lebanese, Syrians, Palestinians, etc.). It's a little misleading to look at imaginary ancestral components in isolation.
Agree but my point was that Levantine was the third largest component
Reply With Quote
  #39  
Old 31st August 2013, 01:15 PM
Javelin Javelin is offline
FTDNA Customer
 
Join Date: Jan 2009
Posts: 990
Quote:
Originally Posted by josh w. View Post
Agree but my point was that Levantine was the third largest component
If I read your posts right, actually it's your second largest component on two tests, since you are conflating three different tests.

Also, when you say "I realize that there was a Near Eastern influence on the Mediterranean and on the Caucasus," it's a bit wrong to speak of these components as "influences."

In essence this is algebra with constructed imaginary components based on peak frequencies in populations. No matter where you come from in Eurasia you will have some amount of x + y + z where x = Caucasus y = Near Eastern z = Mediterranean. The question is, how does your distribution of x + y + z differ from those within your population, and how does it compare to the averages of other populations?

This is where we shift from algebra to statistics, and look for individuals' standard deviations away from their group. We can use this approach to prune outliers, for example. Next, we can create an average cluster for different ethnicities, then compare those.

This idea of breaking down genomes into components is quite interesting and potentially useful, but much more on a population level than an individual one. It's also not as if, when you show as "5% something," that means 5% of your ancestry is from that region. Otherwise, pretty much nobody in the world would be from northern Europe.
Reply With Quote
  #40  
Old 31st August 2013, 01:20 PM
josh w. josh w. is offline
FTDNA Customer
 
Join Date: Mar 2005
Posts: 2,536
Quote:
Originally Posted by NYMark View Post
There are huge problems with Population Finder, no doubt about it. Here's hoping that improvements are coming down the pike.

That said, this is a bit of a damned if you do, damned if you don't situation. 23andMe has been criticized for identifying Ashkenazi as a European population, even though that's clearly accurate for the time frame they're using. Ancestry has similarly been criticized for "European Jewish", which seems to include Sephardic as well but is woefully unreliable whatever it may be measuring.

Personally, I find 23andMe's approach to be the most useful and realistic for a genealogical site. Overall, it seems to be by far the most accurate for Ashkenazim too, although there have been some complaints and some clearly erroneous results.

The fact that full Ashkenazim can range from around 40% (the low end of my bro's margin of error) to 100% Middle Eastern is troubling for two reasons. First, it's potentially misleading; I have corresponded with an Ashkenazi Jew who was convinced their PF results meant they were Mizrahi (and believe me it's hard to dissuade someone who thinks that's what the scientists are telling them.) It's also not at all helpful for people who are of mixed ancestry and are trying to find out their percentages. PF tells such people next to nothing of value. The "Orcadian" category has many of the same issues.
My southwest Asian component at Gedmatch was never greater than 17% on the various strategies. I am not suggesting that I am typical. However, my suspicion is that once Jewish ancestry informative SNPs are located, the algorithm assigns them high priority and ignores other information at PF (Higher percentage estimates may require more such SNPs). Yes, there is no right answer. As you note any solution will have religious and political implications.

Last edited by josh w.; 31st August 2013 at 01:23 PM.
Reply With Quote
Reply

Bookmarks


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -5. The time now is 10:40 AM.


Family Tree DNA - World Headquarters

1445 North Loop West, Suite 820
Houston, Texas 77008, USA

Phone: (713) 868-1438 | Fax: (832) 201-7147
Copyright 2001-2010 Genealogy by Genetics, Ltd.
Powered by vBulletin® Version 3.8.4
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.