PDA

View Full Version : Anyone with erroneous DNA.land results?



Little bit
06-15-2016, 01:08 PM
I'd like to compare with others who got erroneous DNA.land results. My DNA.land results with my 23andme raw data seem to be in line with expectations but my DNA.land results with my Ancestry raw data seems very unexpected and erroneous. I uploaded my Ancestry raw data twice and got the exact same erroneous results each time. I suspect that my higher no-call rate on my Ancestry raw data might be causing the distortion and I'm curious if others got erroneous results and whether they have higher no-call rates, too. I sent a help ticket to DNA.land about it and this was their response:


Thank you so much for letting us know about these results. Ive sent them over to our scientific team to take a look at. Weve had issues with exaggerated African percentages in our Ancestry Report for a small percentage of our users, and are seeking to resolve this for our next iteration of the Report.


My DNA.land 23andme results:
9802
9801

My DNA.land Ancestry results:
9803
9804

On paper, I am 87.5% UK/Irish American Colonial/recent immigrants and 12.5% German/Swiss PA Dutch. This is MyOrigins results:
9805

My no-call rates are as follows:

23andme raw data: 0.257 % ( 2469 of 959251 SNPs) are NoCalls
Ancestry raw data: 1.483 % ( 10123 of 682549 SNPs)

Does anyone think that the no-calls are the culprit or having the same problem? Thanks!

Ann Turner
06-16-2016, 01:24 PM
Virtually everybody shows some differences with the Ancestry Report, although mine were more subtle.

How many segments do you see when you compare your two files? I see 27, so phasing and/or imputation is introducing some errors.

crossover
06-16-2016, 02:45 PM
well every one whose dna i uploaded to dna.land got inflated balkan %

Little bit
06-16-2016, 06:07 PM
How many segments do you see when you compare your two files? I see 27, so phasing and/or imputation is introducing some errors.

@ Ann Turner:

Are you comparing the raw data downloads or via relative finding within DNA.land? Previously, I didn't check the boxes for finding matches but I just did for both submissions and I'll see how they compare me to me.

Little bit
06-16-2016, 07:28 PM
Ok, here are my matches to my Ancestry raw data upload, both are me (23andme raw data upload and my Opensnp 23andme file):

9820

Not sure why sharing should be different between my v-3 upload, which I did, and the Opensnp me, which someone else did. That Opensnp raw data was my v2/v3 raw data as is the one I uploaded??????? :confused:

anglesqueville
06-16-2016, 09:58 PM
LittleBit: First, it happened the same to my norman compatriot Helgenes50, just after the update of DNALand, with his mother's kit. Her results were absolutely nonsensical, with an impossible african part. He erased her kit and uploaded her datas again. The results of the new kit were usual, at least reasonable, without any african component. Perhaps you should try this experiment. That said, I have now in my archives the DNALand, 23&me, ftdna results of a dozen of individuals of my familial clan, on three generations. Superficially seen ( with one only eye, and after drinking a lot of stout), they all seem acceptable. Seen close up, well, the lanscape changes. All those ancestral-compositions/ethnic-make-up/ancestry-results are a darts game, and there is no champion among the gamers.

Ann Turner
06-17-2016, 01:25 PM
Ok, here are my matches to my Ancestry raw data upload, both are me (23andme raw data upload and my Opensnp 23andme file):

9820

Not sure why sharing should be different between my v-3 upload, which I did, and the Opensnp me, which someone else did. That Opensnp raw data was my v2/v3 raw data as is the one I uploaded??????? :confused:
Dunno. Maybe the phasing/imputation algorithm changed between the dates of the two uploads?

wombatofthenorth
06-21-2016, 08:29 PM
well every one whose dna i uploaded to dna.land got inflated balkan %

Funny thing is my dad actually gets a little Balkan on 23 but none at all on DNA.LAND.

crossover
06-22-2016, 03:19 AM
Funny thing is my dad actually gets a little Balkan on 23 but none at all on DNA.LAND.

wow so you got the opposite problem?

CrazyDaisy
07-05-2016, 12:55 AM
I get fake Northeast European and Finnish results on DNAland. I wonder what it's a proxy for.

Morges
07-18-2016, 07:41 AM
People tested with AncestryDNA have a lot of problems with DNA Land.

geomattica
11-13-2016, 04:17 PM
Both myself and a relative got an inflated eastern European/North Slavic. I don't know how accurate this is. This relative, who I am descended from, has German, Irish or Scottish, English, and maybe some Scando ancestry. On FTDNA she scores in that Finnish category but I don't. Both her and I for some reason match with pure Finnish people over at FTDNA. I don't know if this is because of a 'generic effect' or because of a somewhat distant but relatively recent Fenno-Scandinavian ancestor. Also am not sure if this has to do with why our East Euro may be elevated over at DNA Lands. Still weird though because none of us have overly recent Slavic ancestry particularly my relative (she should have none at all actually).

JohnHowellsTyrfro
11-13-2016, 06:53 PM
I submitted my FTDNA data. I'm all "old" UK ancestry as far as I know. (Welsh with some English).
I got 83% Northwest European (fine) and 13% Balkan? and 3.2% Kalash ? I also seem to have some unexpected matches around Libya? I've e mailed and asked for their thoughts. :)
I get things like Finnish, Ashkenazi even Native American elsewhere - some early Central Asia influence maybe?
John

12568

geomattica
12-01-2016, 01:58 AM
Update:

Actually it looks like on that one side of the family of my relative I had tested there was indeed a Finnish woman in the maternal line. I got my hint last week I think on ancestry dot com revealing the link and who she married. Never in a million years did I think I would have someone in the family who was Finnish. But I wonder if that's why me and my relative also have more than expected Slavic. Its funny too because I get <1% Slavic over at 23andME and none on FTDNA. I'm not sure if its reference population setups or algorithms sometimes as to why these results can vary so much among calculators.