New Pop Finder?

[gtc]

According to that new DNA test company, the next population finder will be named "my origins".

Interesting find, assuming it's the same product as PF.

[Mihajlo]

Interesting find, assuming it's the same product as PF.

It has to be that because this name has been proposed in a poll by the ftdna facebook page.

[McNinja]

I wonder if it isn't going to be similar to 23andMe's Ancestry Composition which avoids controversy by lumping people into relatively modern categories to avoid ancestral predictions. One thing they did which confuses people was group Caucasian under Middle Eastern (so if you're getting Middle Eastern and are European, it likely means Caucasus). Those two are hard to distinguish. It'd be better if they tried to face that issue head on rather than use any workarounds.

For me the current PF gave an accurate prediction (so did Dr. McDonald). More accurate than admixture calculators. That means the new one will probably be less accurate for me, oh well.

I like the idea of a PCA-based approach rather than admixture calculator. The allele frequency categorization in admixture is inherently less accurate since people and populations can easily become shifted around when you're not using an absurdly high number of components (like 30+).

[josh w.]

I wonder if it isn't going to be similar to 23andMe's Ancestry Composition which avoids controversy by lumping people into relatively modern categories to avoid ancestral predictions. One thing they did which confuses people was group Caucasian under Middle Eastern (so if you're getting Middle Eastern and are European, it likely means Caucasus). Those two are hard to distinguish. It'd be better if they tried to face that issue head on rather than use any workarounds.

For me the current PF gave an accurate prediction (so did Dr. McDonald). More accurate than admixture calculators. That means the new one will probably be less accurate for me, oh well.

I like the idea of a PCA-based approach rather than admixture calculator. The allele frequency categorization in admixture is inherently less accurate since people and populations can easily become shifted around when you're not using an absurdly high number of components (like 30+).

A problem that I had with PF is that it only listed the best guess and not other possibilities. Gedmatch which is PCA (or MDS- Dodecad) based lists other possibilities with Oracle. There is a confidence interval for locations in PCA space--computed by bootstrap approaches. Why didn't PF list other possibilities or are their estimates so precise that other possibilities can be ruled out--they do list the confidence interval for each component.

[Zaru]

Consistency and percentages.

A problem that I had with PF is that it only listed the best guess and not other possibilities. Gedmatch which is PCA (or MDS- Dodecad) based lists other possibilities with Oracle. There is a confidence interval for locations in PCA space--computed by bootstrap approaches. Why didn't PF list other possibilities or are their estimates so precise that other possibilities can be ruled out--they do list the confidence interval for each component.

PF left a lot of mystery around its analysis. For example: I'm 65% "Finnish/Orcadian"- well, how much of each?? Even split? Will Orcadian become an obsolete term?

The Finnish result perplexed me until I returned a match two weeks ago with a gd=0@67 markers. But the calculators across Gedmatch are quite inconsistent in placing my ethnicity- some say "Germanic", some say "Swedish", "Norwegian", "Hungarian","Russia North", "Lithuanian", only once is "Finnish" mentioned, and a very low (8%) of Fennoscandian and some "Siberian" and "Volga Uralic" thrown in(but the Finns in the database there seem to have either Fennoscandian or Finnish ONLY). Having adoptions in both maternal and paternal sides makes this all the more interesting for me.

So, in short, I am hoping for some clarity and consistency that seems to correlate with my Finnish Y match.

[MoberlyDrake]

I hope they can distinguish between Middle Eastern and Italian like 23andMe the various tools at Gedmatch can. If not, I'll opt out with 4 of my 5 kits. While it's true that Sicily was conquered and inhabited by peoples from the Middle East, it was too long ago to matter for genealogists. I've researched my Sicilian line back to 1700 and before in some cases and haven't found anyone who wasn't Italian.

If someone is searching for Italian ancestors and your profile says you're Middle Eastern, they may think there can't be any connection.

[S9 H9]

According to that new DNA test company, the next population finder will be named "my origins".

http://prosapiagenetics.com/faq.php

just click on : Why ordering your kit? What other tests would I get?

"myOrigins - this tool is offered by Family Tree DNA. With myOrigins you can discover your genetic makeup"

Their product name, "Geographic Population Structure (GPS)", is the same phrase used in yesterday's communication published in Nature Communications:

http://www.nature.com/ncomms/2014/140429/ncomms4513/full/ncomms4513.html

It's not a coincidence. GPS contributor Tatiana Tatarinova is from the Keck School of Medicine, which hosts the academic version of GPS:

http://chcb.saban-chla.usc.edu/gps/

Reading the Nature Communications letter, it comes across as a bit to enthusiastic about their own results, though they have many disclaimers, which may for some reason not found their way onto the commercial website.

[Petra]

Prosapia Genetics

Reading the Nature Communications letter, it comes across as a bit to enthusiastic about their own results, though they have many disclaimers, which may for some reason not found their way onto the commercial website.

As a typical German with neurosis on data misuse, I do not trust their website Prosapiagenetics.

There is no impressum, whre I can read who is the company behind the GPS testing, there are no "privacy policies/terms of use" written.

Why should I upload my sensitive DNA-data even for the temporal "free for all" test to get "better results" for an population finder with "gps" technique? They might collect data to get a benchmark to FTDNA?

Or do they need data to research medical purposes? ->http://www.newswise.com/articles/gen...lized-medicine

Yes, I am very curious to see new population finder results. But I won´t upload my data to other companies without security.

So I will wait what the release of FTDNA has to offer...

[josh w.]

Their product name, "Geographic Population Structure (GPS)", is the same phrase used in yesterday's communication published in Nature Communications:

http://www.nature.com/ncomms/2014/140429/ncomms4513/full/ncomms4513.html

It's not a coincidence. GPS contributor Tatiana Tatarinova is from the Keck School of Medicine, which hosts the academic version of GPS:

http://chcb.saban-chla.usc.edu/gps/

Reading the Nature Communications letter, it comes across as a bit to enthusiastic about their own results, though they have many disclaimers, which may for some reason not found their way onto the commercial website.

I am not exactly impressed with the first author's (Elhaik) track record. Contrary to all other autosomal research on the topic, he concluded that Khazars had the primary influence on the Ashkenazi population---this is not the place to go into the many errors in research design.

[Ann Turner]

I am not exactly impressed with the first author's (Elhaik) track record. Contrary to all other autosomal research on the topic, he concluded that Khazars had the primary influence on the Ashkenazi population---this is not the place to go into the many errors in research design.

Dienekes has posted a critique of the method:

http://dienekes.blogspot.com/2014/05/the-geographic-position-structure-gps.html

[Táltos]

Hallelujah! Is this really happening tomorrow?

If so, will it be 10 am Houston time?

Why is there no count down clock? JK!

[gtc]

Hallelujah! Is this really happening tomorrow?

I've not heard anything different ... so far.

If so, will it be 10 am Houston time?

Who knows?

Why is there no count down clock? JK!

Probably because there's no sale associated with it.

[vinnie]

At about 8:45 p.m. EST, this was posted in ISOGG's FB forum:

"...We are rolling out in beta mode but it will be a little longer before everyone sees the release."

Apparently, they're trying to make sure there are as few bugs as possible - obviously a good move.

[gtc]

At about 8:45 p.m. EST, this was posted in ISOGG's FB forum:

"...We are rolling out in beta mode but it will be a little longer before everyone sees the release."

The original PF is still tagged as "BETA".

Apparently, they're trying to make sure there are as few bugs as possible - obviously a good move.

That would make a welcome change.

[Taz85]

So no new PF tomorrow. Not surprising