Friday, October 9, 2015

DNA.Land Launches


DNA.Land was launched today by scientists from the New York Genome Center and Columbia University. I was asked to advise on this project several months ago and have been excitedly anticipating its launch. 

Posted by Dr. Yaniv Erlich on Facebook today:


After long months of work, DNA.Land (https://dna.land) is finally on air.

We developed a platform to massively crowd source genetic, genealogy, and health information from millions of people. Our goal is to help our participants know more about their genome and help science. By uploading their genome data to our website, participants will be able to find genetic relatives, learn about their ancestry, and get a more complete version of their genome data. During this process, they will also be able to contribute to cutting-edge genetic research...

To quantify their level of contribution to science, the website will also present to each participant a badge with a score that corresponds to the amount of data he or she contributed and the percentile of data contribution...

If you were tested with 23andMe, Family Tree DNA, or Ancestry - take a look at https://dna.land/

The website is not-for-profit and was developed by scientists from Columbia University and the NY Genome Center.

We would love to hear your feedback.

Anyone who has tested at 23andMe, Family Tree DNA or AncestryDNA may participate by uploading the raw data to the site. There are very good instructions on DNA.Land for completing the raw data upload process. Please make sure to use a valid email address or you will not be able to access your results. Due to consent and privacy issues, DNA.Land currently does not support multiple uploads to a single account, so you if you have consent from family members to perform the upload to DNA.Land, it will require a different email address/account to do so. 


The raw data upload and the imputation step will take about two hours and the ancestry reports will be ready in about 24 hours. (Imputation is the act of inferring the base -- ATCG -- at a location that has not been tested. More info here and here.)

I have received my reports and am sharing screen shots from my account below. Right now there are only two, but there will be more features added in the future. 


Relative Matching:



I really like that DNA.Land classifies the shared segments as either "ancient" (green) or "recent" (orange), which we have not seen previously on other sites. Not surprisingly, shorter segments are more likely to be designated as "ancient," but you cannot always tell just by looking at the chromosome graph. This is because in some regions, the rate of recombination is slower than others, resulting in a lower centiMorgan value versus nucleotide count (megabase pairs). You can see this in my screenshots above where DNA.Land has classified the segment on Chromosome 9 in my matches 1, 4 and 5 as "ancient" even though it appears to be relatively large in comparison to some of the other "recent" segments. This is because it is crossing the centromere and the recombination rate is expected to be slower there. At this point, you cannot see the start and stop points of the shared segments on the site, but I am advised you can do so by viewing the HTML source. (Hopefully, this will be surfaced later.)

Since I was one of the first ones to upload to the database, most of my matches are from OpenSNP. For these, no email address is provided (column one), but for most of your matches, there should be an email address for contact. I have some concerns with the closeness of the relationship estimates shown in column two because I would not expect to see so many close relatives in such a small database, but I am confident the team will reassess this during this beta phase. I think the "Relationship Likelihood" graph in column five is a good addition to demonstrate the uncertainty of the relationship predictions. 


Ancestral Origins Report:


As you can see these reports are pretty basic right now, but they will continue to develop and improve. Knowing how much the genetic genealogy community likes to see the predictions of the small percentage ancestral origins, I have recommended that this be expanded. 

Please remember that this is an early beta version and DNA.Land will be refined as they work through issues and add additional features. Feedback and questions should be sent to info@dna.land.

This is a wonderful way to contribute to genetic research, so please check it out! 

FAQ are here.
Facebook page where you can ask questions here.
More information from the Nature news article here

30 comments:

  1. This is very exciting Cece! I'm so happy you're a part of it.

    ReplyDelete
  2. just loaded mine and my new found biological sister (or half sister) excited to find out which!

    ReplyDelete
    Replies
    1. You need to upload to Gedmatch to determine if she is full or half. DNA.Land won't tell you that.

      Delete
  3. Wonderful! I'll be uploading my own raw data presently!

    ReplyDelete
  4. Are there any matching segment details, or is that information represented only graphically?

    ReplyDelete
  5. Glad to see another DNA website, but concerned about the accuracy of the Ancestry Report. I am fully Armenian, but show up as 37.86% Ashkenazi/Levantine, 55.54%Turkish-Iranian-Caucausian and 6.6% Other. Will there be modifications?

    ReplyDelete
  6. Thanks for sharing. Will add mine too.

    ReplyDelete
  7. This comment has been removed by the author.

    ReplyDelete
  8. Thanks for sharing, CeCe.

    To the average user, what are the imputed BCF and VCF files useful for / where can they be used?

    ReplyDelete
  9. CeCe— Thanks for explaining some of the issues with your own example. Got my results... very interesting! Fascinating, in fact and not at all what I expected!

    ReplyDelete
  10. Your presentations this morning prompted me to upload all three tests. Anxious to see the results. (And also excited to use some other of your DNA-research "tips" from both of this morning's sessions.

    ReplyDelete
  11. Seriously flawed! I got so far in the UL process and it gave me no option but to cancel out!

    ReplyDelete
  12. This comment has been removed by the author.

    ReplyDelete
  13. Do you know if they will be accepting raw data zip files that are not of the big three soon? We would be very interested....

    ReplyDelete
  14. Your link for dna.land goes to facebook. Just FYI.

    ReplyDelete
  15. Two questions:

    Will it be necessary to upload data again to take advantage of the subsequent improvements in DNA.land's information to individuals?

    I downloaded my imputed VCF file and am submitting to Promethease. Will future improvements in DNA.land's algorighms, etc. produce a better VCF file in the future?

    ReplyDelete
  16. Will it be necessary to resubmit or reanalyze data to take advantage of downstream improvements?

    ReplyDelete
  17. Q for anyone : known family is uploaded to site and not showing as 'high certainity' nor do I have any matches. A glitch?

    ReplyDelete
    Replies
    1. How closely are the known family related to you? It does sound like a glitch, so please send feedback to DNA.Land.

      Delete
  18. Really enjoying DNA land!Will I eventually get an update on my 5% other?

    ReplyDelete
    Replies
    1. I would like to know as well. Dna land told me I'm 66 percent southern European, 20 percent American Indian and 14 percent other. It matches up with AncestryDNA pretty well. Ancestry told me I'm 60 percent European, 22 percent native American, 5 percent from Mali, 2 percent North African, and 9 percent west Asian. I'm assuming the database doesn't have the enough data to see the African and middle eastern. Hopefully as more are uploaded we get a clearer picture.

      Delete
  19. I built some tools to show you how your ancestry compares to your facebook friends. Check it out:
    https://www.infino.me/ancestry/

    ReplyDelete
  20. I have uploaded my mother's, father's, and my own tests from Ancestry to DNA.land. Their results were rather different from Ancestry, especially for me. They have me 19% Italian (my father at 8% Italian) even though the only Italian in the family came over to New Amsterdam in the 1630s and married someone Dutch. DNA.land says I'm 8% African, yet there's no African in either parent's test. They also have me as 1.7% Taiwanese and 1.4% Native Oceanian (Australian Aboriginal) and 6.1% Dravidian. Both of my parents are listed as 100% Western Eurasian, which they are except for the little bit of Native American in my mom. DNA.Land did not pick that up in her, but they did say I'm 1.4% Native American. I'm curious as to how they came up with my heritage being far different from both my parents and far different from other DNA tests I and my parents have taken. No, I'm not adopted, and the tests show that yes, those are my parents. I wrote to DNA.land to ask, but got no reply. Any ideas? Oh, I also put in my mom's test through 23andMe and DNA.land gave somewhat different results from what they gave for her Ancestry test results.

    ReplyDelete
  21. Dear AncientPlanter,

    Thank you for alerting us to this issue. I recall writing back to you about this, but apologize if your e-mail happened to slip through the cracks on our end. It won't happen again.

    After the launch of our recently updated Ancestry Report, a small number of our users reported receiving exaggerated African percentages. Perhaps this is what occurred with your account? If so, please send us an e-mail at info@dna.land with the e-mail address associated with your DNA.Land account. We are constantly looking to improve our service, and are currently seeking to expand our reference panel for a future Ancestry Report.

    That being said, it is also very common for results from different services to be somewhat different - especially with smaller percentages of Ancestry. This is why, until very recently, DNA.Land listed all ancestral percentages lower than 5% as "other."

    Think of your genome as a beam of light. The reference panel of each service’s algorithm acts as a prism that decomposes this beam of light to its potential components. Since each prism has different physical properties, the rainbows they produce will all be at least a little bit different.

    As each reference panel has its own strengths and caveats, it is best to observe your genomes through as many prisms as you can. Only in this way can you better illuminate your past.

    Thank you,
    Richard A.
    User Engagement Coordinator, DNA.Land

    ReplyDelete
  22. I uploaded my data last night and received my results today. So far no matches but I'm hopeful. Great site, easy to use and understand.

    ReplyDelete
  23. I did my ancestry DNA I was 97% European 3% western Asia . upload to DNA.LAND I get 5% African 5% Mende/AKAN I am from Louisiana so they say we may have African ancestry my problem is how can it be higher than my western European when my mom is British and German .my western European is just 3.2% .

    ReplyDelete
  24. I got very different results between ANCESTRY and DNA.LAND.
    Ancestry: 35% Iberian, 20% Italian/Greek, 12% Native American, 8% European Jewish, 5% North African, 5% Europe West, 4% Middle East.


    DNA.LAND: 53% Southwestern Europe, 23% Southern Europe (Italy/Balkans) , 11% Native American, 6.2% North African.


    As you can see, DNA.LAND calculated a lot more Iberian genes and no Jewish. Why such a big difference?

    Thanks

    ReplyDelete