23andMe's latest proves what I forecasted

[Zoro]

A while back I said that with each update everyone's result will get more boring because their goal is to make everyone 100% their own ethnicity. Their latest update proves what I said. It's no secret anymore that they do this by mathematically getting rid of non-major admixture.

For example according to what they say they break up you genome into windows of 100 SNPs or so. Let's say you are W. Asian. If most of the SNPs in the window are W. Asian, they make the whole window W. Asian. Even more blatently (according to them) let's say they assign the windows AAEAA where A is Asian window and E is European window. Next they will get rid of European window and make that also Asian window because it's surrounded by Asian windows. Their logic is the European must be old because of small segment size.

So I got a call from my only Kurdish relatives who test with V5 chip and got the update to tell me that what I told him a while back about 23andMe wanting to make everyone 100% is true because that's exactly what they did with their latest V5 update. He said he already knew that because all the ancestors are from Kurdistan. He said he had taken the test originally to see from what parts of Asia and Europe Kurd ancestry is from but 23andMe is useless for that.

Here are his latest results. They deleted all non Iranian/Caucasian/Mesopotamian ancestry

[IMG][/IMG]
[IMG][/IMG]

[Zoro]

So now that they made him 99% Iranian/Caucasian/Mesopotamian he has to figure out on his own where do Iranians/Caucasians/Mesopotamians have ancestry from lol . Great help they are.

[Peterski]

What does he score with other companies?

[Peterski]

For example according to what they say they break up you genome into windows of 100 SNPs or so.

I've also heard a version that they break your genome into 100 SNP windows, each representing 1% of the total.

Not into windows 100 SNPs long each.

Edit:

But no I guess it has to be 1000 windows, because they report percentages as small as 0.1% not 1% like FTDNA.

[Zoro]

What does he score with other companies?

I have these for him. Not sure which company they're from

[IMG][/IMG]
[IMG][/IMG]

[IMG][/IMG]
[IMG][/IMG]

[Peterski]

I have these for him. Not sure which company they're from

[IMG][/IMG]
[IMG][/IMG]

^^^
These results make sense but why do they count "West Asian and North African" as part of "European" ???

Also it should be very easy to separate West Asian from North African, actually. Maybe except for Egypt.

[Zoro]

I've also heard a version that they break your genome into 100 SNP windows, each representing 1% of the total.

Not into windows 100 SNPs long each.

Edit:

But no I guess it has to be 1000 windows, because they report percentages as small as 0.1% not 1% like FTDNA.

It used to be 100 SNP windows. Now they're 300 SNP windows. I think this makes it easier to knock out minor ancestry because for European if 60/100 SNP window are Asian then the window will be made Asian but now 60/300 the window will be made European.

First, we just separate the customer’s genome data into little segments, like boxcars on a train. Each boxcar only contains the data for about 300 DNA markers — out of a total of over 650,000 that we test for on our genotyping chip. And dividing 650,000 by 300 DNA markers, you get about 2,000 boxcars.

[Peterski]

It used to be 100 SNP windows. Now they're 300 SNP windows. I think this makes it easier to knock out minor ancestry because for European if 60/100 SNP window are Asian then the window will be made Asian but now 60/300 the window will be made European.

"First, we just separate the customer’s genome data into little segments, like boxcars on a train. Each boxcar only contains the data for about 300 DNA markers — out of a total of over 650,000 that we test for on our genotyping chip. And dividing 650,000 by 300 DNA markers, you get about 2,000 boxcars."

OK so they divide into 2000 not 1000 windows. They could report 0,05%.

BTW, 23andMe should use the same categories as American colleges use, where Somalis are counted as white.

I can't find the link now but there was a thread about this recently on TA.

[Zoro]

He also did this one

[IMG][/IMG]

[IMG][/IMG]

[Zoro]

Personally I don't take any ancestry results too seriously because humans on average have about 3,500,000 SNPs and all these companies use of the Illumina chipsets which genotype only about 700,000 SNPs so there's always a bias as to which SNPs are used.

Even if you have done whole genome sequencing (WGS) the Gedmatch calculators or testing companies if you upload will only use a small fraction of what you upload. Then there's the problem of which references to use. For example if an Armenian takes the test and they happen to use Armenian references he will turn out 80 or 90% Armenian so he will never know what other parts of the world his ancestry is from

Actually there are about 60,000,000 SNPs but on average people only have 3,500,000 that are not reference SNPs. So if you as European have 3,500,000 alternate allele SNPs then an Asian will also have 3,500,000 alternate allele SNPs but they are not necessarily the same ones you have since there are a total of 60,000,000. The question becomes how to decide which ones to use in a calculator.

Here is another forecast I have. People's results will drastically change 10 years from now. People from one country are not as homogenous as what the companies lead us to believe