Although some of the images below may look familiar from my earlier posts, I am recycling them to illustrate and identify key components of the product that we discussed on the call. Any of the images can be enlarged for closer examination by clicking on them.
Keeping It Simple
As I have mentioned before, the folks at Ancestry.com have attempted to keep the science involved with interpreting the genetic data simple for the layman and have done a very good job of presenting the ancestry information in a straightforward, easy-to-understand fashion. There are many informational pop-ups that explain what the customer is seeing on their screen as well as easy-to-use tools that assist the customer in identifying the common ancestor(s) with their matches.
Ken Chahine, Senior VP in charge of DNA, perhaps in response to my earlier characterization of their new product, specifically emphasized that this is cutting edge science (especially their BGA feature) and that they definitely want to serve both the beginner and the advanced genetic genealogist with their AncestryDNA product, adding, "I hope going forward that people realize that while we made the user interface very, very straight-forward that it does not in any way represent 'dumbed down science' and to the contrary, to be able to give you...[the specific genetic ethnicity labels]... is taking what I would consider probably the best population genetics research that is out there right now." Some of their statements regarding progress in this area (detailed later) sound promising, so I will reserve further judgment until I get a chance to thoroughly investigate the final product.
Streamlining the work of autosomal DNA matching
This is probably the most significant advancement for genetic genealogists who are using their autosomal DNA to identify common ancestors with other genealogists. At AncestryDNA your matches' trees are automatically compared and shared surnames and birth locations are clearly listed. When possible, shared common ancestors are even identified with those familiar little "shaky leaf" symbols. You can view a seven generation pedigree of your match's tree right on the screen with a link to the actual tree for exploring more generations. According to Chahine, by having the trees readily available on the site with their commonality instantly identified, thus cutting down on the legwork traditionally associated with tracking down the common ancestor between matches, they have produced quick and easy confirmation of common ancestors between many of the Beta testers. I'm not surprised to hear this and look forward to exploring my own results. (The only problem is that I will have to ask these confirmed relatives to test at 23andMe or FTDNA, so I can get the specific segment information to add to my chromosome mapping charts.)
Admixture Analysis Tool - "Genetic Ethnicity"
As I've said before, this is a very strong feature with impressive and attractive graphics. They are presently using 22 different labels to identify the unique populations to which they are matching our DNA with the "Genetic Ethnicity" feature. These are drawn from reference samples from many different sources, both public and proprietary. The number of specific categories will likely increase. (Population reference samples are compiled from "anchors" which are individuals whose families have been in one place for generations.) These categories include Jewish and both Native North American and Native South American. There is also an "Uncertain" label that is used when your DNA matches one or more ethnic regions with too little information to confidently predict. In some cases, they stayed away from geography for the "ethnicity" labels since that isn't always representative of ancestry. I learned that they are not yet showing Beta testers the African detail that we saw on the Blair Underwood WDYTYA episode, but plan on doing so in the future. While stating that this has "never been done to the level we have done it", they repeatedly emphasized that this feature is still in Beta and will undergo changes and improvements based on the Beta test feedback.
Genetic ethnicity versus family tree
AncestryDNA explores the relationship between your Ancestry.com family tree and your genetic results. They do this by comparing your predicted "ethnicity" to the actual birth locations of your ancestors in your tree. The birth locations from your tree are also shown in total percentages in order to make it easy to compare to your "genetic ethnicity" percentages. If none of the places in your tree match up to some of the hints your DNA is giving, perhaps there are new avenues to explore that might result in overcoming genealogical brickwalls, by giving clues to the "ethnicity" of your ancestors with unknown origins.
Common Ancestral Origins Between Matches
AncestryDNA will compare your "genetic ethnicity" to that of each of your matches and attempt to isolate the "genetic ethnicity" of your common ancestor based on the overlap. In the second image below you can see that there is only a small commonality between these two participants' admixture results, so the idea is that the common ancestor is very likely to be of the identified shared "ethnicity". This allows you to focus in on a particular branch of your family tree or provides a hint of what may lie just beyond one of your genealogical brickwalls.
Plotted Birth Locations
One screen plots birth locations on a map, pulled from both your match's tree and from your own, highlighting common locations (seen in the first image above). Ken Chahine said that this feature has been "a remarkably valuable tool" so far in the Beta phase, explaining "that when you can't find the common ancestor, frequently what you find is common locations of individuals who have been in tiny towns in the East Coast of the United States in the 1800s", thus giving the ability to hone in on locations that may have only had several thousand people a few generations ago, offering an "opportunity to look at the research" in a unique way.
Relative Matching with Confidence Levels
The confidence levels shown on the right hand side of the match list is based on tests run using 200 real individuals with 30,000 or more possible relationships between them. This confidence number tells you what percent of the time the AncestryDNA's relationship range predictions corresponded with the actual real data in their experiments. For example, in the graphic below, you can see a 96% confidence level for Match #2. Chahine explains, "We ended up...calibrating our algorithm, so that we could tell the customer, if we find a 4th cousin how confident we are that that person really is a 4th cousin, and...about 96% of the ones we identified in our experiments were, in fact, 4th cousins." (It was later clarified that he meant that they will fall within the predicted range, which for 4th cousins as seen in the image below is 4th - 6th.) He went on to say, "I would argue for the 4th cousins you should be trying pretty hard [to find a common ancestor] because our data suggests that this person has a very, very strong likelihood to actually be a 4th cousin."
They also just added a pretty useful little sliding bar for easy filtering of matches of different degrees of predicted relationships, allowing you to focus in on closer relatives or, instead, explore all of those more speculative matches who may share an ancestor with you in the last ten generations or so by including the "Distant Cousin" category (5th cousins or further) in your display.
Chahine noted that even at the 50% confidence level, some Beta testers have been able to confirm a common ancestor with their matches. He recommended (in line with my usual advice) that testers at least look over these more distant matches to determine if there is any low hanging fruit, such as already identified common ancestors, surnames or locations.
It was explained that the matching system is "dynamic" with data "pulled live" when you confirm a common ancestor with a match, thus "increasing the power of the product".
One other convenient feature is that the person who administers the kit is listed in addition to the actual match's "name". As genetic genealogists well know, they are not always the same person. It is also noted how long the administrator has been a member of Ancestry.com and their last sign-in date.
The DNA interface observes the global privacy settings that each customer has on their account, meaning that if your Ancestry.com tree is private, it will not be linked to your results. Your matches will be required to contact you for permission to view your tree, as is always the case on the site. No privacy settings have been or will be compromised.
On another note, I have heard from Beta testers that recently more and more of their matches are attaching their family trees to their DNA results. This is quite essential for success.
Timeline and Beta
The test is currently in a private Beta phase. They advised us that the results that we have seen so far are in "no way the final product". They explained that AncestryDNA wanted to put something "lean" out there for Beta in order to receive feedback and get a better sense of what the needs of their customers are before finalizing the product.
We were told that more results should be out in the "next few days". Whether that meant for our group specifically or the next wave of Beta testers, I am not sure. Either way, they said that they "will be getting it out to more and more customers throughout this year". They also commented that their team has "a lot to add" to what we have already seen in Beta and said these additions are "right around the corner".
Feedback from Beta
They have been very happy with the constructive feedback they have received. John Pereira, VP of Business Development, said they are "absolutely inundated with feedback so far and they have to sift through that" before the general release. They have also been "incredibly surprised" with the quality and quantity of confirmed matches already. Chahine enthused, "It has been amazing how many have already found their common ancestors!"
Plans for the future
The team said that their goal is to provide "the very best test that is out there" and "they really want to be a leader in the space". Pereira emphasized that their focus is to combine the genetic information with already existing Ancestry content for ease of use, but assured that they do intend to "get more tools out there to help the customers" streamline their advanced research. He said to expect the product to change rapidly and advised us to "keep coming back to the site because there will be lots of new changes". We were advised that providing the raw genetic data and specific segment matching information are "in the queue" of potential future additions to the product, as is allowing raw data uploads from 23andMe and FTDNA. They made it clear that they are not opposed to these ideas, but they cannot guarantee what will be included with this product in the future because "it is a moving target" with "priorities changing every day" in response to feedback from the testers. (So, if the actual genetic data is important to you, please let them know!) Pereira shared that they will likely add points throughout the site where advanced customers can "dive down into the science", while keeping the basic user interface relatively simple (this may include more detailed genetic info) and that they definitely have plans to answer the questions in regard to the deeper science once they have a good sense of what those questions really are. The team said that they are very happy with the product they have put forward so far, but are "even more excited" about the improvements that are expected in the future.
Responding to my query, Chahine acknowledged that they are looking at solutions for the challenges of relative matching for endogamous populations like Ashkenazi Jews and that it is "really high on the list" to come up with special algorithms that deal with various populations that are not predicted well with the current matching algorithm.
They plan to eventually send email updates to customers, advising them of new matches and possible common ancestors like they currently do with family tree hints.
They have no information on future pricing at this time.
Other interesting tidbits
Ken Chahine confirmed my suspicion that Y-DNA was used in conjunction with autosomal DNA in the WDYTYA Blair Underwood episode that caused some controversy over the seemingly exaggerated exactness of the 10th cousin prediction. Chahine explained, "We were able to take Blair's African component from an autosomal test and actually assess that...his DNA signature was most closely associated with a tribe that was identified in Northwest Cameroon and then we ended up using a Y-DNA marker to help us identify a potentially distant cousin, which we did, and the episode introduced Blair to that distant cousin. What I was really amazed about was how well the science integrated, so on the ethnicity side, the match and the ethnicity were very, very close to each other in terms of geography."
Summary
We've all been waiting for what seems like an eternity to see what Ancestry.com will do with this new test. I appreciated the opportunity to get a good look at it and to be able to share it with my readers. As I have stated before, this test has a lot of potential for encouraging the masses to try out DNA testing, but it needs to go further with the details of what's happening "under the hood" to satisfy the serious genetic genealogy researcher. I think they realize this and will take steps in this direction. Will it be enough to win over the genetic genealogy community? Time will tell and I will keep you updated, as always.
**Update - Leave it to Blaine Bettinger to pull it all together for us! He blogged about his thoughts from this same call over at The Genetic Genealogist and his take is well worth the read: Ancestry.com's AncestryDNA Product. **
[10/26/12: This test is now out of Beta, so you can order it here.]
