Will Ftdna ever develop an autosomal test

[josh w.]

Tom, couldn't find anything googling except the type of study I mentioned above. One reason I asked the question is that sometimes on the DnaTribes forum there are discrepancies. Do they describe the correlations anyplace. Thanks for your information.

[tomcat]

Not explicitly. But they have recently posted a rather interesting study on their blog. It'd FREE, as are their many sample.pdf's.

Tom

[tomcat]

... One reason I asked the question is that sometimes on the DnaTribes forum there are discrepancies ...

As a user of the DNATribes' test, I would say there are no "discepancies" in any DNATribes report but only approximations that may appear discrepant to known genealogy. For example, in my recent Upgrade/Update, I got several matches to Turkish populations although I am not Turkish. However, as my mother was Native American, a macro-population derived from Central Asian stock and the Turks were also a steppe people, and as my father's family were Ukrainian Jews and Jewish marker profiles are similar to Turk's, my Turkish matches make sense.

Of course I would prefer a match to Jews of the Cherkassy Oblast and a specific North American tribe but the growing database of reference populations does not yet offer those options. And, given the peculiarity of my mixed ancestry, I may never get the strong scores that some receive in their DNATribes reports. Nevertheless, I am an intrigued user of DNATribes' offering and an avid researcher of the autosomal STR's on which their test is based.

I have no financial interest in DNATribes other than the money I spent with them, an interest approximate to what I spent on DNAPrint's Ancestry by DNA 2.5. I think Tribes is the better value. Due to my ABDNA scores I am foreclosed from Euro 1.0 but Tribes' Continental analysis already offers far more definition than Euro 1.0 and they are continually upgrading their database and statistics. And, for anyone of primarily European ancestry, the STR results from Tribes can also be researched in a European database - www.str-base.org. - (ENFSI will accept 8 of the 13 CODIS markers). You don't get an authoritatively gold-stamped certificate from Tribes as you do from DNAPrint, but you do get information you can work with in the same way so many work with their yDNA results. It is a process. It is interesting.

Tom

[josh w.]

Tom, you have almost convinced me, but let me give some examples where I have doubts about DnaTribes. The first problem involves an extrapolation from my Ydna results. I am a J2 of Jewish ancestry. At the 12 marker level, the majority of my exact matches appear to be of Jewish background. On the other hand, the majority of my distant matches are from Central Asia. If Ftdna subsequently removed the majority of it's Jewish samples, it would create the incorrect impression that my ancestors hailed from Central Asia when that would only be true of my "distant cousins". This is similar to the state of affairs with DnaTribes where comparitively speaking, there are limited samples from northwest European and Ashkenazim populations. By default higher scores will go to other regions or groups.
The second problem arises from DnaTribes overreliance on the modal or most common haplotype in a given region. Let me use a Ydna example since autosomal haplotype nomenclature is not familiar to me. Suppose my ancestors hailed from a region where they did not have the most common haplogroup in that region, e.g. J2 in France. I might receive high scores for regions (e.g. west India or the Caucasus) where J2 is more common even though my own genetic line never passed through these areas. Conversely, I would not have a high score for France. The history of J2s in France will not be well understood by focusing on the mode.
Of course these problems can be avoided by just getting the loci values and ignoring DnaTribes results, but a better database and algorithm would help compared to doing it on your own.
Josh

[tomcat]

Josh,

I cannot disagree with you - if you are both paternally and maternally Jewish with NW European origins you won't find much company in Tribes' database, at present, and, as a regional minority, you will not be helped by any version of the maximum likelihood method of statistics.

On the other hand, if we knew beforehand what we would find we would not take these tests.

Do have another statistical method applicable to this problem?

BTW, and apropos your earlier inquiry, I have revised my posting to www.dnatribeslog.org going into more detail on my Italian/Iberian cluster and note that it, and certain other features of my profile, seem most like the distribution of yHapJ2 as evidenced in this atlas - http://www.relativegenetics.com/geno...ges/haploMaps/

Waiting to hear if I am J2.

Tom

[NormanGalway]

Tom, I am encouraged by your positive experience with Tribes. Some of the other feedback seems to be negative, and reading some of the logs, occasionally downright kooky (cf, the Kentuckian who is a "Tamil.")

My chief concern is that they do not have an Irish, English, or French "Tribe," and are also weak in Northern Europe generally.

I am sensitive to this because I had an experience with Oxford Ancestors in which they issued me a "Viking certificate" based on very weak and limited information. Not to knock them, but frankly that was overreaching in my view. Especially since it was only based on 12 markers.

Since I am in a very rare Y-DNA haplogroup (old I1b2), I wonder what sort of results the Tribes algorithm will cough up, and how meaningful they could be since the key constituents of my ancestry are missing from their database?

All of the above to say that I am intensely interested in getting an autosomal test, but unsure if the time is right. I know you're a disinterested party - just curious what your thoughts are on the above.

[tomcat]

Norman,

Have you looked into the ENFSI database at - www.str-base.org/? They have population samples for Austrian, Belgium, Switzerland, Czech Rep., Germany, Denmark, Estonia, Spain, Finland, France/Lille and France/Toulouse, Croatia, Ireland and North Ireland, Italy, Netherlands, Norway, Poland, Portugal, Sweden, Slovenia, England/Wales, Scotland/Dundee and Scotland/Glasgow.

As per my earlier in this thread they accept 8 of the 13 CODIS markers employed by tribes. It's a free, online search, a way to get a bit more mileage out of a DNATribes test and augment Tribes' results. Am sure you would get something out of it.

Is this your yHap?

http://www.relativegenetics.com/genomics/images/haploMaps/I1b2_large_RG.jpg

Tom

[NormanGalway]

Tom,

That is it. for some reason, the link is not coming up, but I have seen that map before.

I guess I am ready to take the plunge then if 8 markers in that database are enough to "show" anything... because of the wacky experience I had with only 12 markers (or 10, I can't remember), I am a bit wary, but curiosity is getting the better of me...

Also, is it worth getting BGA Plus or is just the basic package the good option?

[tomcat]

So yHap I1b2 is really thick on Sardinia and Corsica? Check-out DNATribes to see if they have any samples there and for corresponding area of North Africa.

I ran the 8 on ENFSI and got Finland followed by Italy, Switz., France/Lille, Netherlands, Spain. Then got the 2 missing markers sequenced and ran the 10 to get Switz., Spain, Denmark, Netherlands, France/Lille, Italy. Pretty much the same in different order but the Sandinavian changed!

Extra markers available through DNA-Fingerprint.

Tom

[tomcat]

Norman,

I went for the Basic the first time around and just Upgraded but go for the whole enchilada or you will only have 6 markers for the ENFSI run.

Tom

[NormanGalway]

ok, thanks Tom -- will do.

[josh w.]

Tom, thanks again for your info. What is needed is an algorithm that incorporates more information than the mode. Perhaps expanding the number of loci is a solution, i.e. more differentiated haplotypes might be more geographically specific in the way that subclades sometimes are. I checked with Sorenson regarding the kind of output information they would provide. They responded that their autosomal tests were years away.
I didn"t mean to single out DnaTribes. ABDNA has its own limitations. Earlier they had promised a more geographically specific test, but I can no longer find any reference to it on their webpage.

[josh w.]

P.S. My concerns and the problems that I mentioned were not limited to J2s or those of Jewish ancestry. My son is hoping to trace his maternal Christian line which we suspect has northwestern European origins. His maternal grandfather was adopted and is now deceased so only an autosomal test might help ( or perhaps an X test).

[tomcat]

Josh,

The issues with DNATribes is deeper than the algorithm, it is the markers themselves. CODIS and all atSTR typing systems are based on markers chosen for their utility in distinguishing individuals - the odds of any two persons from a stable and ethnically homogenous population matching are many, many millions-to-one. The COOL thing about DNATribes is their insight that the worldwide databases of forensic results could be tapped to yield useful biogeographical information. I very much doubt that DNAPrint has a comparable database but their marker set does include a group of highly informative SNP's. That said I think Tribes is doing really good and useful work.

As you wrote, we are years (and many millions of dollars) away from a gold-standard BGA, but most of us want answers today. In my case, I want to know whatever I can about my mother's biological parents, who by name and particulars are missing from ALL the usual records - census, draft, SSDI, etc. Any little hook will help in directing a search of the less accessible files.

An X-Chr. test for your son seems the best bet - much more direct than anything available on the autosomal front. I have ordered two x-panels from DNA-F. Howsabout if we leave-off STRing this pot and move on to promoting X?

Tom

[josh w.]

Agreed,
Josh