23andMe Updates Paternal Haplogroup Assignments

[sirdan]

Kind of old news, However important.

(...)
We have identified and corrected several “variant metadata” errors for this update, improving haplogroup assignments. While we were at it, we also excluded variants performing poorly on the latest genotyping chip, V5, further improving haplogroup calls. As a result, some customers will see improved paternal haplogroup assignments, whereas most will not.


Why Some Haplogroup Assignments Have Changed
As described above, Yhaplo draws variant metadata from ISOGG. This includes the ancestral and derived alleles of each variant. For most people, sporadic metadata errors did not impact Yhaplo’s ability to call haplogroups. However, in some cases, they led to haplogroup assignments that were a bit off. For example, the ancestral and derived alleles of SNP L1335 were reversed. This led to Yhaplo incorrectly assigning R-L1335 to some individuals (red branch in Figure 2). With such errors corrected, Yhaplo is more accurate, and these changes are reflected in the most recent Paternal Haplogroup Report.
Additional Resources
For additional details on the Yhaplo algorithm, please see our bioRxiv manuscript. We have open-sourced Yhaplo on GitHub for non-commercial use, pursuant to the terms of the non-exclusive license agreement included with the software distribution.


Unfortunately, I am still left with wrong I1 subclade assignment. Just hope this will be continuous update, as this is "Coming Soon" labeled.

[East Anglian]

They still continue to give me a SNP with a 17,000 year old TMRCA. And wrongly provenance it to Southern India. So as usual with that business, no improvement.

Still stuck in Last Glacial Maximum.

[Capsian20]

Nothing update for me its still E-V257

[Moeca]

23andme is floundering in regards to DNA, it has always been rubbish or myself..............it will fall into the ancestry sphere of limited or no DNA

but MyHeritage, just announced on roots tech that they can get DNA from stamps on letters that your ancestor wrote with 99% accuracy ................seems like MH will pursue a more DNA path

[Vinitharya]

I think 23andMe does a good job with the autosomal. I mean, it isn't great, but it is more within the margin for error-at least for me-than Ancestry is.

[sirdan]

23andme has good country origin estimates.

And it was first to reveal haplogroups.

Also auto generate tree.

Also genome viewer.

I think it was left for too long with no maintaince and focused too much on medical news. I hope it will not be completely abandoned.

[AimSmall]

After six months, they need to bring back the ability to see matches in the chromosome browser and to download the matching SNPs. Without this, I'm really not interested in them anymore.

[Rober_tce]

For me there isn’t changes in my haplogroup, the last update was some years ago (2020)

[ArmandoR1b]

Kind of old news, However important.

(...)
We have identified and corrected several “variant metadata” errors for this update, improving haplogroup assignments. While we were at it, we also excluded variants performing poorly on the latest genotyping chip, V5, further improving haplogroup calls. As a result, some customers will see improved paternal haplogroup assignments, whereas most will not.


Why Some Haplogroup Assignments Have Changed
As described above, Yhaplo draws variant metadata from ISOGG. This includes the ancestral and derived alleles of each variant. For most people, sporadic metadata errors did not impact Yhaplo’s ability to call haplogroups. However, in some cases, they led to haplogroup assignments that were a bit off. For example, the ancestral and derived alleles of SNP L1335 were reversed. This led to Yhaplo incorrectly assigning R-L1335 to some individuals (red branch in Figure 2). With such errors corrected, Yhaplo is more accurate, and these changes are reflected in the most recent Paternal Haplogroup Report.
Additional Resources
For additional details on the Yhaplo algorithm, please see our bioRxiv manuscript. We have open-sourced Yhaplo on GitHub for non-commercial use, pursuant to the terms of the non-exclusive license agreement included with the software distribution.


Unfortunately, I am still left with wrong I1 subclade assignment. Just hope this will be continuous update, as this is "Coming Soon" labeled.

I had posted at Anthrogenica many times about the errors of false positives and false negatives of 23andme v5. The errors only affect a small percentage of customers but it did affect some people. It has taken them a long time to finally acknowledge the problem. I had also informed the Hunter Provyn, the creator of CladeFinder, of some of the errors which he made sure the program would correct by not using that data.

Here is a list of SNPs that were false positives
C-Z12480
D-Z43868
I1-S239.1/L205.1/L939.1
I1-S244/Z58
I1-BY352
I2-CTS3529/PF3720/Z2649
O-F1531
Q-S239.2/L205.2/L939.2
Q-FGC1897/Y2225
R-S47

Here is a list of SNPs that were false negatives
R-Z272
R-Z295/S1217
R-CTS241/DF13/S521

[ArmandoR1b]

For me there isn’t changes in my haplogroup, the last update was some years ago (2020)

Only a very small percent of people will see a change. What is your haplogroup at 23andme v5? What is it with FTDNA?

[ArmandoR1b]

23andme is floundering in regards to DNA, it has always been rubbish or myself..............it will fall into the ancestry sphere of limited or no DNA

but MyHeritage, just announced on roots tech that they can get DNA from stamps on letters that your ancestor wrote with 99% accuracy ................seems like MH will pursue a more DNA path

Doesn't MyHeritage use the FTDNA lab? That's my understanding. MyHeritage won't do Y111 or Big Y testing though even though they use the FTDNA lab.

[ArmandoR1b]

They still continue to give me a SNP with a 17,000 year old TMRCA. And wrongly provenance it to Southern India. So as usual with that business, no improvement.

Still stuck in Last Glacial Maximum.

You have a very rare haplogroup. They would have to get into Big Y type testing or WGS in order to report a younger haplogroup for you.

[ArmandoR1b]

Nothing update for me its still E-V257

That's interesting. E-V257 is HG38 position 12363865 see http://ybrowse.org/gb2/gbrowse/chrY/?nam.....12363865
So it should show up with a search at 23andme of chromosome Y at https://you.23andme.com/tools/data/?quer...2012363865 but it doesn't so you were tested for something else equivalent and not E-V257 itself. If you search for E-Z287 position 7422413 at https://you.23andme.com/tools/data/?quer...%207422413 it shows up. C is derived so you should be positive for that.

Everyone tested by 23andme v5 is tested for all of the same positions. There is a limited number of positions that are tested. You will never get a better result at 23andme v5 because there was nothing else tested that you were derived for that had a false reading. It's the same with the FTDNA Family Finder result. Once it's done there will be no updates without Big Y-700 testing or SNP pack testing at FTDNA.

Here are the SNPs relevant to you that I could find that you should have been tested for:
CTS1243/Z827
M310.1/PF2402.1
M81/PF2554
M183/Page33
M5008/PF2498
M165

Put those in to Ybrowse then get their hg38 position then copy and past that position into the 23andme Raw Genotyping Data browser and get your result. You should get results similar to https://cladefinder.yseq.net/ as far was what was tested and what the result is.

By the way, What do you get from https://cladefinder.yseq.net/ ?

[Capsian20]

Nothing update for me its still E-V257

That's interesting. E-V257 is HG38 position 12363865 see http://ybrowse.org/gb2/gbrowse/chrY/?nam.....12363865
So it should show up with a search at 23andme of chromosome Y at https://you.23andme.com/tools/data/?quer...2012363865 but it doesn't so you were tested for something else equivalent and not E-V257 itself. If you search for E-Z287 position 7422413 at https://you.23andme.com/tools/data/?quer...%207422413 it shows up. C is derived so you should be positive for that.

Everyone tested by 23andme v5 is tested for all of the same positions. There is a limited number of positions that are tested. You will never get a better result at 23andme v5 because there was nothing else tested that you were derived for that had a false reading. It's the same with the FTDNA Family Finder result. Once it's done there will be no updates without Big Y-700 testing or SNP pack testing at FTDNA.

Here are the SNPs relevant to you that I could find that you should have been tested for:
CTS1243/Z827
M310.1/PF2402.1
M81/PF2554
M183/Page33
M5008/PF2498
M165

Put those in to Ybrowse then get their hg38 position then copy and past that position into the 23andme Raw Genotyping Data browser and get your result. You should get results similar to https://cladefinder.yseq.net/ as far was what was tested and what the result is.

By the way, What do you get from https://cladefinder.yseq.net/ ?

Thank you , i get in YSEQ E-L19 but negative a subclade M81, for SNPs in family Finder its seems to its better than SNPs in 23andME
       

[Moeca]

23andme is floundering in regards to DNA, it has always been rubbish or myself..............it will fall into the ancestry sphere of limited or no DNA

but MyHeritage, just announced on roots tech that they can get DNA from stamps on letters that your ancestor wrote with 99% accuracy ................seems like MH will pursue a more DNA path

Doesn't MyHeritage use the FTDNA lab? That's my understanding. MyHeritage won't do Y111 or Big Y testing though even though they use the FTDNA lab.

correct