 |
| Tim Janzen and I discussing the AncestryDNA features at RootsTech with AncestryDNA staff |
Raw Data Downloads
On Thursday, AncestryDNA fulfilled their promise to allow customers to download their raw data. As Dr. Ken Chahine had assured me back in November, the file is not encrypted and is compatible with third party tools.
I sent my file to a number of third party providers:
- After working with it a bit, John Olson announced on the site that he expects that Gedmatch will be accepting AncestryDNA uploads in about two weeks.
- David Pike told me that he has updated his tools to work with the AncestryDNA files.
- Leon Kull has reportedly updated his HIR search site to work with them as well.
- Dr. Ann Turner has created an Excel macro to convert the AncestryDNA files to 23andMe format.
- Bennett Greenspan told the audience that Family Tree DNA will be accepting AncestryDNA transfers into Family Finder starting on May 1st.
- Dr. Catherine Ball confirmed that the raw data file is not phased and that they are delivering it as they receive it from the chip manufacturer Illumina. She also confirmed what Dr. Ann Turner had already discovered - the data labeled as "Chromosome 25" is from the PAR region. Further, the "Chromosome 23" label refers to the X chromosome data and "Chromosome 24" refers to the Y chromosome.
- Unlike Family Tree DNA, AncestryDNA is not removing any SNPs from the data - medically relevant or not.
- The overlap between AncestryDNA's raw data file and 23andMe's should be around 690,000 SNPs due to the fact that they are both using the same Illumina OmniExpress Plus base chip. The ~10,000 SNP difference can be accounted for due to a different set of poorly preforming probes and test SNPs. Family Tree DNA's should have a similar overlap for the same reasons.
- There is no mitochondrial DNA included in the raw data file because it is not included on the Illumina chip that they are using. (23andMe adds the mtDNA SNPs).
Search Function
As I expected from earlier conversations with AncestryDNA, a search function is next on the list. Kenny Freestone, Product Manager for AncestryDNA, discussed it in his presentation under the heading "What's Next". Although it is already in the works, Kenny could not provide a firm timeline for its availability when I asked.
We will be able to filter our list of matches by surname, location and username. As anyone who has worked with their AncestryDNA matches knows, this is sorely needed. There is no doubt that the many requests from customers pushed this up their list of priorities.
Genetic Ethnicity Update
Later this year, AncestryDNA will be updating their Genetic Ethnicity feature. They will provide more granularity in Europe and West Africa. We can also expect more accurate breakdowns. A number of AncestryDNA personnel acknowledged to me over the weekend that certain "ethnicities" (i.e. - Scandinavian) are overestimated for many customers. However, they also emphasized that much of the perceived problem with their admixture analysis stems from the question of "where and when". What they mean by this is that it is very difficult (and sometimes impossible) to pinpoint where specific DNA signatures were at an exact time in history.
As I always remind my readers, this portion of the science has a long way to go and will improve with more data and time. On the "more data" front, during her speech at the AncestryDNA luncheon on Friday, Dr. Ball was reportedly requesting that genealogists who know that all eight of their great grandparents were born in the same place share this information with AncestryDNA. This seems to imply that, like 23andMe has successfully done, AncestryDNA plans to use customer data to improve their predictions. They are also starting to work on incorporating the coveted SMGF collection into their admixture analysis, which should improve it greatly.
The good news is that AncestryDNA customers don't have to wait for this update to gain more insight into their ancestral origins. Now that AncestryDNA has made the raw data available, customers will be able to upload their raw data file to the various third party sites to try out the admixture calculators and/or send it to Dr. McDonald for his very highly regarded analysis.
Matching
AncestryDNA are currently working on an algorithm to improve matching for endogamous populations, specifically Ashkenazi Jews.
As I reported in November, the minimum threshold for matching is 5 megabase pairs. This was reconfirmed in a conversation I had with Dr. Ball on Friday. I also learned that there is no minimum SNP requirement. We discussed the possibility of AncestryDNA switching to centiMorgan measurements in the future.
Price
The test is now $99 for everyone - subscribers and non-subscribers. This was likely in response to 23andMe's recent price drop. Having attracted well over 120,000 customers in less than a year in business, AncestryDNA is proving to be an important player in this field. This new policy to attract subscribers and non-subscribers alike will only improve their market share.
International Customers
It does not appear that AncestryDNA has plans to offer their test to international customers in the near future, instead choosing to focus on the U.S. market for now.
Matching Segment Data and Chromosome Browser
On Friday at RootsTech, Dr. Tim Janzen and I sat down for a meeting with AncestryDNA management. Among others, we were joined by Dr. Ken Chahine, Senior Vice President and General Manager of DNA, and Dr. Catherine Ball, Vice President of Genomics and BioInformatics. (Dave Dowell also attended a portion of the meeting.) I found them to be very receptive to hearing our requests and the reasons behind them. At no time did they state that they had decided not to build a chromosome browser or release matching segment data to their customers in the future. Dr. Ball did express some privacy concerns, but was open to hearing ideas of how this could be addressed.
 |
| Tim Janzen explains his feelings while Ken Chahine looks on |
During the meeting, Tim very emphatically explained his feelings on the need for matching segment data (above) and I resorted to begging (below)... {hehe}
 |
| Catherine Ball, Ken Chahine, Tim Janzen, me, Dave Dowell and Steve Baloglu |
On Saturday, after attending Kenny Freestone's presentation, four advanced genetic genealogists approached him to discuss the chromosome browser issue. In addition to myself, Tim Janzen, Angie Bush, and Nathan Machula were present for the conversation. Kenny didn't have much to say and mostly listened to the arguments that we presented covering why we feel that it is essential that AncestryDNA offer the matching segment data behind their relative predictions. At no time did he state that AncestryDNA would not offer a chromosome browser or that the delay in doing so was because AncestryDNA didn't think that their customers could understand it. He did, however, confirm that it was not a top priority at this time. He also said that he personally reads all of the requests sent through the feedback button, so if you want them to reassess their priorities, then be sure and let them know.
Tim emphasized that both 23andMe and Family Tree DNA included a chromosome browser feature at the launch of their autosomal DNA product and wondered aloud why AncestryDNA had not done so as well. I explained to Kenny (as well as in my meeting with management) that, as genealogists, we expect conclusions to be evidence based. It is not in line with this principle to simply be told that a certain common ancestor is responsible for a DNA match and be expected to take AncestryDNA's word for it. Where is the proof? Since Kenny had shown a chart during his presentation of his ancestral lines that he claimed were genetically confirmed by AncestryDNA matches, I also pointed out the fact that those lines that he had shaded in weren't really confirmed without the actual genetic data to support that claim. To illustrate, I laid out my experience as follows:
On my AncestryDNA account, I was happy to find a shaky leaf hint a few weeks ago.
The only problem is that this match doesn't appear anywhere on my mother's 47 pages of matches. Do you know what this means? It means that I must have inherited the DNA responsible for the match through my father's side. Since all DNA inherited through my mother's line must come through her, AncestryDNA has identified the wrong common ancestor as the source of the DNA shared between LGB and me. A fluke of the algorithms...? Perhaps. Let's look at some more of my matches.
Once again, as you can see, the common ancestor identified by AncestryDNA is on my mother's side. A thorough search of my mother's matches shows that, once again, this person is not reported as a match to my mother. From this, we can only reach the conclusion that the DNA responsible for this match comes through my father's side - not my mother's. The common ancestor that I share with "Baerion" must be beyond a brick wall in her family tree or on my paternal side. In general, I have had more success filling in the branches of the maternal side of my family tree than the paternal side, so this is certainly possible.
Just to demonstrate that this isn't an isolated occurence, here is another one:
This match doesn't appear on my mother's match list either! So, out of my ten matches that have shaky leaves attached, three of them apparently have common ancestors wrongly identified as the source of our matching DNA. Do you see the problem here? Does AncestryDNA? If this match were, instead, at 23andMe or Family Tree DNA, I could check the DNA segment that we share and compare it to my other matches and/or my chromosome map. This would provide additional information and/or evidence to help me determine through which of my ancestral lines this segment of DNA was inherited. Might there be other explanations for these discrepancies? It is certainly possible, but without the underlying genetic data, it is impossible to say.
I am in the fortunate position to have tested my mother at AncestryDNA in addition to myself, so I can clearly see there is an issue. What about all of those people who have not tested a parent and are blindly accepting AncestryDNA's shared ancestor hints because they don't know otherwise? Isn't that kind of like copying someone's tree and just taking their word for it that it is correct with no sources or evidence attached? For now, those of us who do understand the finer points of autosomal DNA matching will have to do our best to convince our matches to upload to Gedmatch so they can see for themselves what they are missing.
As much as I, too, am disappointed that AncestryDNA has not yet provided the matching segment data, it is clear to me that the reasons behind this decision are far more complex than what others may claim is an attempt to dumb down the product because Ancestry.com thinks its customers are stupid. From my many conversations with Ken Chahine and others from AncestryDNA over the past year, I have come to appreciate that working within the framework of this 1.6 billion dollar corporation comes with its own set of challenges.
The Future
Tim Sullivan, CEO, has made it clear that Ancestry.com is committed to the DNA business and Ken Chahine has always been upfront with me and come through with his promises. So, I am going to give them the benefit of the doubt. From our very first conversation, I have advocated for the genetic genealogy community and looked out for our best interests and I won't stop doing so. I believe that they will do the right thing for their customers and the genetic genealogy community eventually. It may not happen as quickly as we would all like (yesterday!), but they are not the big bad wolf and I think it does us all a disservice to continually paint their intentions in a negative light. We are in early days yet. Let's give them a break.
