Looking for DNA Advice (3 Generations)

+11 votes

I have finally made some headway with my paternal family, and my father (72) and grandfather (94) are both ready to take DNA tests. I know that AncestryDNA has a larger customer base, but some people prefer Family Tree DNA for tools like Family Finder.

With all of the Holiday sales going on, I am ready to dive in, but would love some input on a single testing service for 3 generations of male descendants. We are looking for the "best bang for the buck" scenario.

From what I understand, autosomal is the way to go. Is this this the best bet? Should I also look for a service that includes X-chromosome mapping as well? Are there any Pros and Cons to services like Family Tree DNA and AncestryDNA? Is one truly better than the other from a genealogical standpoint?

Note: My mother has already been tested with AncestryDNA.

in The Tree House by Steve Harris G2G6 Pilot (431k points)
retagged by Ellen Smith

12 Answers

+6 votes
I have taken several brands, and they all have their strengths and weaknesses for whom you match on their site with.  The good thing with FamilyTree, is it allows raw data from other brands to be uploaded into their site for matching for Free.  Huge plus there as you are not locked into just FamilyTree or Ancestry.  What good is a site that does not maximize the population it compares results against right?  Ancestry is the most locked down since it does not share outside their proprietary site.  I see no advantage to be with Ancestry unless your entire family is with them. But what you will be missing is everyone you don't know about, and cannot find you because you are behind the Ancestry Firewall!  Autosomal is Autosomal.  Its good for about six generations, and you will get lots of matching..  Yes Autosomal is the way to go.  I did MyHeritage and 23andme.  I uploaded my raw data to FamilyTree and don't forget to put it onto GEDMATCH.  If doing Ancestry DNA make sure you load your raw data to GEDMATCH so you can at least see who you match (if they took another brand), but more importantly, so they can match YOU.    I think I saw MyHeritage had two kits for $39/ea and free shipping.  If you then upload to FamilyTree and Gedmatch you will have everything you need and be visible by all (if you opt into share).

One other thing is 23andme uses spit and it takes quite a lot.  MyHeritage uses a cheek swab and it is up and down inside the mouth for a few motions.  Very easy..

ANcestry and FamilyTree and MyHeritage all allow GEDCOM trees to be uploaded and managed within their sites.  23andMe is working on a tree they build for you using their DNA test takers DNA.  Still in beta form.  Quite a few bugs still.  No Gedcom uploading of trees yet.

I don't need health benefits so 23andme, Ancestry may be your options if you want those additional reports at more cost.

Happy hunting
by Kirt Fetterling G2G6 Mach 1 (12.5k points)
Ancestry is the most locked down, but it is by far the largest.  If you can upload to other sites, why pay for their tests also?
As I understand it, AncestryDNA will not let you upload from other companies, but other companies will let you upload AncestryDNA to theirs, either for free or for a nominal one-time fee. So if you go with AncestryDNA, you can have your data basically everywhere.

Despite what all the commercials say, the real genealogical value is in the matches with other people. True, AncestryDNA does NOT tell you exactly what parts of which chromosome you match people on - which is frustrating - but, again, you can do that on the other sites you upload to, and you can at least try to get others on AncestryDNA to upload elsewhere as well.
23andme also allows downloads but not uploads, but has the tools that Ancestry lacks. There user base is very large, but still short of Ancestry DNA’s.
Yes, their user base is large, but it is mostly hidden.  While it is true that for beginners (or anyone, I suppose), the strongest matches are the most useful, I have occasionally found very useful matches on Ancestry in the 10-20 cM range.

Also, 23andMe's requirement that matches agree to "share" makes the matches much less useful, as most of the time as I recall my matches have not responded to my share requests.
Thanks everyone! So it looks like the recommendation through this thread may be to do the AncestryDNA testing, and upload to other tools such FamilyTree and GEDMATCH?
When it comes to the 94 year old grandfather, I'd be doing the Y DNA test at FTDNA, and the Family Finder autosomal test, and let some great grandchild or other join Ancestry DNA for their autosomal test and the multitude of cousin matches that come with it. The question is do you want the truth, or do you just want to spend less money. Autosomal testing has to be joined with Y DNA testing to be most useful, and Ancestry DNA doesn't do Y DNA testing.
As other have said, AncestryDNA is largest pool to fish in, but is the most frustrating - not only because it lacks a chromosome browser but b/c the vast majority of folks who test with them are not serious about genealogy. They take the test on a whim to try and learn their ethnicity breakfown and then never log in again after they get their results. Message after message goes unanswered...

BUT, if you test with AncestryDNA you can upload to other sites - GEDmatch, FamikyTreeDNA and MyHeritage.  

MyHeritage is interesting because not only does the site have a chromosome browser but also a triangulation tool.  However, I have found messaging matches to be very troublesome - for most I cannot send unless I upgrade (too expensive, yearly subscription not monthly) and those few I have messaged do not reply and I wonder if my messages were even received b/c of lack of subscription.
+6 votes
For autosomal testing, I think that AncestryDNA and 23andMe are probably your best bets. AncestryDNA apparently has the largest user base, but its DNA tools aren't very good. 23andMe apparently has the second largest user base and it has excellent DNA tools. Family Tree DNA also has great DNA tools, but a smaller user base -- but if you test somewhere else you can upload data from other sites to FTDNA for free and pay a modest fee to use their tools. And you can upload data from any of these testing services to Gedmatch.

Any autosomal test should include X-chromosome results. Both 23andMe and Gedmatch have tools for working with the X data; I'm not sure about the others.
by Ellen Smith G2G Astronaut (1.1m points)
Thanks Ellen!
+5 votes
Best bang for the buck auDNA is Ancestry for matching (most members) also good for resources and building a tree. I'd get kits for all three of you there for sure.  I'd add an additional kit from My Heritage for your grandfather.

Next, I'd join a Harris surname project on FTDNA and start with a Y37 test there, for just yourself, when they have a sale (may be now). The surname project will group you with (distant) male cousins in your particular paternal Harris line and so provide the best insight into its actual origins.
by Mike Wells G2G6 Mach 3 (32.3k points)
Yes you can do a YDNA test, but it is expensive, and matching and getting results is not easy.  It requires a lot of digging to find answers.  I took a YDNA-37 test and have not found anyone close enough yet.  Upgraded to a YDNA-111 marker test, and my base of matches was no better.   More markers better matching right?  WRONG.   Expensive and I need to wait for others to test.  Right now my closest match is 20 generations away at 98% confidence.  Now unless your tree is correct for 20 generations, good for you!.. Most of us are confident about maybe four generations?   Stick with Autosomal.   Now FamilyTree has a bundle kit on sale 40% off with lots of options.   https://www.familytreedna.com/products

My YDNA came back as Haplogroup I-M253 from familyTree.  Autosomal from MyHeritage was I-M253  but from 23andMe it is I-Z58 as subgroup of I-M253.   My $300 test of YDNA was not better than a $40 autosomal kit.   Or to explain in a better way.  I'm still on hold waiting for someone I match with my YDNA to respond.  I have 10 matches but the closest (so far) is 20 generations away.  The people are unresponsive to my requests to work to find our common ancestor.  Autosomal at least I find matches and based on the matching CM's you can tell if you just need to fill in a gap on a generation or two to get that person into your tree.  Hope that helps!
If a YDNA test is not done, it's literally impossible to verify one's paternal line back to its origins unless perhaps a YDNA-tested proxy (a brother or verified male cousin in same line) is used.  

As you pointed out, YMMV for # of matches with that test based on how many men in your direct paternal line who have taken the test with FTDNA. But your personal experience of course will not be indicative of everyone's experience.  Y37 is a good place to start to confirm your line, and of course Y111 is available to provide "better matching" IF and when refinement is needed - which it often is.

I happen to have plenty of matches for my Y111, but only at GD 3 and higher for now. No recent SNP matches for my Y700 - yet.
+6 votes

94 year old Grandfather is really special.  Get him tested!  Within the US, Ancestry.com has by far the largest database.  So, you will find more matches.

I've also used 23&Me and FamilyTree.

Your experience with Ancestry.com will be shaped by your relatives.  If several of your relatives have entered bogus geneaolgies, and other have followed suit, then you will not like the results.

However, in my situation, I've found that most of my relatives on Ancestry.com have entered accurate Trees.  As such, I found a lot of good matches.  My Ancestry.com results match almost exactly what is on WikiTree.

Both of my parents are on Ancestry.com.  I'm looking at 60,000 DNA matches for my dad, 1,350 4th cousins or closer and about 300 matches for who a common ancestor is identifiable.  Some go back to the late 1600's.

My Mothers results are not as impressive; 25,000 DNA matches, 350 4th cousins or closer, but still about 175 common ancestors.  That's because she does not have as many colonial american ancestors, but I have found a lot of relatives living in Sweden.
by Andrew Ross G2G6 Mach 2 (24.7k points)
Are you saying that if a proliferation of errors exist on Ancestry User Trees, that the AncestryDNA tests will link through them?
Steve, if there are erroneous trees on Ancestry, they can affect what you see on your ThruLines.  That can also happen on WikiTree, where if someone has mis-stated his own line, WikiTree could show you as related in a way that you are actually are not.
+6 votes
Everyone has given excellent advice for the autosomal testing. I would just like to add one suggestion. Since your grandfather is 94, I would spring for the Y-DNA test on FTDNA for him. Right now on the holiday sale, the Y-37 test is $99. If you add the FamilyFinder (autosomal) it is another $40 or $49. If you want to go with Ancestry for the autosomal, do the Y test for him on FTDNA and then do the autosomal on Ancestry along with your Dad. If you then need more Y testing later, you can use the same test and upgrade it for your grandfather and then test your Dad at a later time. That would be awesome for tracking the male lines in your family.

Good luck - lots of choices and a lot of good information to consider.
by Virginia Fields G2G6 Pilot (512k points)
I'd only get a YDNA for him if he didn't match the expected auDNA cM for his son. Otherwise, there is basically no benefit in getting him a YDNA test when his male descendants can do so.  The reason is simply because YDNA STR markers changes very, very slowly and would be identical or only slightly different than that of his male descendants.

Because his auDNA will provide the oldest MRCAs for the family, and therefore provide more genealogical clues and connections, I recommend he get an additional kit (or more) from other auDNA companies (such as My Heritage) that have somewhat overlapping but also different global member pools than Ancestry.  For example, through My Heritage I've found dozens of ancestors in countries like Finland (up to 45cM) that simply would have never been found using just Ancestry.

I'm not sure what you're getting at with "if he didn't match the expected auDNA cM for his son". If you're saying to make sure he's biologically related first, fine, but if he's NOT biologically related that would be a whole different issue, and you might not want to do a YDNA at all, depending on their goals.

But assuming that they ARE all related (and chances are high that they are) if they're interested in YDNA - which basically tells you which other guys with your surname are related to you, even if distantly - you'd STILL want to do grandpa FIRST. There could easily (about a 30% chance, I think) be a mutation between him and the next couple of generations, and if you don't test him NOW, you'd never catch that. Hopefully, the price will come down eventually, and you might be tempted to get the younger generations tested. You could do the grandson, and then if he's a 100% match, there'd be absolutely no need to do the generation between.

That's a great point about auDNA, though. As has been discussed on this forum, the testing companies don't test every bit of DNA, or even every bit of genealogically meaningful DNA, and what they DO test doesn't doesn't necessarily have as much overlap as we'd like. You could probably get good coverage with just a couple, though, if you pick the right two.

Supposedly, we can expect full testing in about 4 years, which will eliminate this issue. But I'd say grandpa can't wait around for that!
Frank, sure there can be maybe 1 mutation, but consider the net value in what that information would provide vs the net genealogical value of being tested by multiple auDNA companies. If you have x dollars to spend, then *no* don't have him do the Y-DNA test *vs* another company's auDNA test. That's where I'm coming from - and I've been in the exact same position as the OP.
Thanks Frank for confirming my point. From the experts on down, everyone says to test the oldest generation possible. Later generations can test to confirm the relationships. I was only able to test one aunt in my parent's generation but thank goodness, she agreed to the autosomal test and I will be forever grateful. For the Y-test I am testing cousins and 2nd cousins for one of my lines but DNA testing was not available before all the older generations had died. Don't pass up that chance!!
Steve,  Yes it seems we are all saying take autosomal for at least you and your dad.  I would also do GrandPa autosomal (x), but do keep in mind if a spit test or Swab and if that may make a difference for GrandPa.  Now my understanding is YDNA does not matter which male in the line takes it.  You all share that same gene which has hardly mutated over tens of thousands of years..   It is expensive, and unless you are willing to work hard to contact others to find out who your last know common ancestor is, do it later.  You loose nothing for the YDNA, but you loose GrandPa's matching autosomal DNA for the lines you and your father also should match.  Said another way, bad assumptions in your tree can be easily proven if you are all getting pretty much the same results.  You need 6th cousins for Triangulation matching to prove everyone.   YDNA will only help you on your father and grandpa and up the line.  It does nothing for all your other branches I believe you are trying to discover when you proposed your original question.  Your original question was not about a specific parental line but the most bang for the buck with matching I believe.   I could also have been understood as you are only interested in the male line of three consecutive living  males.    Knowing what I know now, use inexpensive Autosomal testing  to build out your tree.  Once that is done, and you want to prove your fathers line, do the YDNA test.  Wait until on sale again.  Plus they are always getting new SNP's, so waiting will not hurt.  $39 to test each SNP you are assumed to be in, but you need to purchase more testing to narrow your match's.  Grandpa's value is in half the X's he carries, that you may not have inherited from him and your father and grandmothers X's.    ALSO  if getting the Ancestry brand (or any other), make sure you upload all tests to FamilyTree and Gedmatch so others can match you, while you are also looking in Ancestry to match those behind the curtain.
+7 votes
Steve, from what I can tell from your profile, three of your grandparents may still be living.  If I were you, I would spend my entire budget on testing them (and Ancestry first, due to the largest user base; then make sure you upload their results to GEDmatch).
by Julie Kelts G2G6 Pilot (313k points)

Indeed! The mom's test has only half the DNA of her parents - it would definitely be worth having them tested. I guess he still needs his dad tested, since grandma is no longer with us on that side, but a test for himself is basically just for entertainment - and I guess to "check up on" his parents. wink 

Frank, so would it better to test the living relatives (grandparents and father) instead of me?

FYI - I am not concerned with confirming my relationship to my father (or mother for that matter). You can't fake the common characteristics that our family shares, lol.

Yes, Steve. Test your oldest living ancestors first and do it quickly. Their DNA is a family treasure that could be lost at any moment. Life is uncertain!
And, Steve, all your DNA came from one of your parents or the other.  If you can test them both, testing yourself adds nothing.  Likewise for your mother, if you can test both her parents (although I see she has already tested).
Also, by the way, if any of your grandparents have living cousins, they would be great test candidates too.  In every case, the farther back you can go, the better, and if you can't test a person's parents, then that person's siblings can provide new information.  For example, my mother took a DNA test, but her sisters have matches that she does not have.
+5 votes
Ancestry is far better for genealogy than any of the other DNA testing services due to database size and a customer base with more interest in genealogy. I have tested with Ancestry and 23andMe; both of my parents have tested with Ancestry. Data from all 3 of these Ancestry tests has been uploaded to FTDNA and MyHeritage as well as GEDmatch.

Total number of matches for my father: Ancestry, 94.7K; MyHeritage, 9714; FTDNA, 1525.

Total number of matches for my mother: Ancestry, 133.8K; MyHeritage, 10,010; FTDNA, 1438.

Total number of matches for me: Ancestry, 123K; MyHeritage, 9547; FTDNA, 1363; 23andMe, +/-1200 (the number fluctuates; the very last person on my match list on 23andMe (with the smallest shown amount of shared DNA) is a known 3C1R who shares 22cM.
by C Handy G2G6 Mach 9 (96.7k points)
Interesting stats! Makes me feel less bad about not having uploaded to those other sites yet.

Still, is it "apples and oranges"? I have 31.9k "Distant Matches" (meaning between 6cM and 20cM) on AncestryDNA, and it sounds like, from what you're saying, that 23andMe at least isn't giving you matches below 22cM.

If I have that right, the fairer comparison would be between my matches ABOVE 20cM on AncestryDNA (which happens to be 663). It's still an issue if they're not giving you lower matches, but I'd say it's a separate issue.

How low do these matches on MyHeritage and FTDNA go?
Frank, after being on Ancestry for a couple of years, I tested with 23andMe.  I was shocked by the meager number of matches they showed me.

Edit:  Ancestry gives me 61,375 total matches, 1297 of which are over 20 cM.  23andMe give me 1268 matches, down to 24 cM, although I had to search to even find the cM, due to their practice of stating the percentage of shared DNA more prominently than the cM.

Because the Ancestry "4th cousin or greater" matches are roughly equal to the total number of matches 23andMe reveals, I wonder if 23andMe actually could provide me around 60,000 matches, if they cared to.  I don't know the answer to that, but it seems logical.
MyHeritage goes down to 8cM, FTDNA goes down to 13cm. And 23andMe not showing me matches below 22cM may not be typical; my ancestry is mostly colonial American on both sides, so I have a fairly large number of matches above c. 20cM (I have 3300 above 20cM on Ancestry; my father has 3200, my mother has 4200...but I also have matches all the way down to 6cM whose shared match list includes as many as 50 people who share above 20cM).
Yes, the number of matches is highly dependent on the match threshold. In my experience with 23andMe, the match threshold has gotten higher as the number of members increases. It seems like the approximate number of DNA relatives has stayed constant, so the match threshold gets higher over time. Currently I'm also seeing only people at 22 cM or better. Most of the time, there's no discernible relationship to people at that match threshold or less (they might match me at, say, 16 cM, but they don't match any of my other known relatives and there's nobody in their family tree who looks remotely familiar). However, that's well above the match value I've seen between a couple of pairs of known relatives at approximately the 3rd cousin level, not to mention the 4th cousin level, so I think we're missing some good match data.
auDNA stats from me:

Ancestry: 46,195 matches - Can't figure out how to filter by cM, but only 1932 matches 4th cousin or closer

My Heritage: 7,800 matches (but *many* more matches outside of USA for some reason)

GEDmatch: 3,000+

FTDNA: 1,312 matches

23andMe: 1,235 matches
Thanks you two (Julie & C Handy)! Interesting!

Say - when I look at the match list for a 23andMe person on GEDmatch, I see "lots of red" - meaning lots of warnings that there's a low overlap between their test and everybody else's, icluding a lot of the folks who have ALSO done 23andMe. Seems like something's amiss here - maybe that has to do with the low match counts, despite their supposedly having a good number of people tested. Did somebody say they're #2 in that regard? Strange.
Ellen, that reminds me of what seems like has gone on at AncestryDNA. At some point, they used some algorithm to throw away especially common DNA, and people said their match counts went way down. I have to believe that it was a business decision - perhaps people were overwhelmed by so many matches. Of course, they have a technical rationalization - that YouKnowWho would defend - but if it weren't oerfectly good DNA, they wouldn't have tested it in the first place.

When I get below 40cM or 50cM it can start to get pretty tough to find out how you're related to people, even if they cooperate. Still, I've found perfectly good matches "down there". Sometimes the way I learn of them is that I see them on my brother's match list. Often, if I have a low match to someone, he'll have a relatively high one, and vice versa. It's great when you have access to a close relative's matches that way.
The red for 23andMe matches in comparison to Ancestry or FTDNA matches is because 23andMe v5 tests mostly different SNPs with only a small overlap with Ancestry/FTDNA's tested SNPs. Both Ancestry and 23andMe v5 test c. 650K SNPs (the overlap of SNPs in common between both tests is only in the range of 200K I think).

And as to your other comment, Ancestry uses an algorithm called "Timber" that they claim "reduces matching of DNA common to populations" but it is almost certainly stripping out genuine IBD; a comparison of a group of matches on 23andMe with known common ancestors (my 6th great-grandparents William Willett and Mary Griffith of Prince George's County, Maryland) on Ancestry and the same testers on 23andMe shows that Ancestry reduces the shared DNA for those matches by between 5cM and 15cM (longest segment on 23andMe is 45cM on chr8; triangulates with 41, 36 and 33cM segments on the same chromosome). The same matches on Ancestry show as sharing between 28 and 32 cM.
Yes, I agree Timber sometimes distorts results.  For some reason I can't begin to explain, Ancestry applied Timber differently to my mother's results than to mine, so often my match to a given person is stronger than hers (even though there is no second source of my shared DNA), a difference that disappears on GEDmatch.
Mike, the thing about cM levels on AncestryDNA is just that "4th cousin or closer" is really "20cM or higher".

BTW, I've got relatives as distant as 6C1R (6th cousin, once removed) in that category, so don't take the name of it too literally. A more accurate name would be "2C1R, or more distant"! Those categories they give are fairly accurate for 1C and closer though!
+3 votes

Test your father and grandfather both with 23andMe and Ancestry DNA.

Don't bother testing your father and grandfather with Family Tree DNA. It simply has too few users to be useful as an autosomal DNA test. If you want your Y-chromosome tested, do that with your own DNA on FTDNA after you find out your Y haplogroup using 23andMe. 

Y DNA testing is only worthwhile if you have a very specific, pre-planned test to undertake and only affects one line. It also requires at least 2-3 men to test for certainty. You have your father's Y-chromosome, so do it later if necessary.

23andMe phases your DNA using your parents, so this is the absolute best for 3 generations. It's database (like Ancestry) is substantial. But unlike Ancestry, it gives you segment data. That segment data, which is what you will want if you plan to use DNApainter.com (which is extremely useful in my own experience). 23andMe data is also the most compatible with MyHeritage if you choose to upload. 

23andMe does threshold your matches to the top 1000 to 2000, but this isn't usually a big deal: Those ones are quite likely going to be enough to start sorting out some mysteries. Still, if you can afford having them do both, include Ancestry too.

by anonymous G2G6 Pilot (127k points)
Hi JN,

In another thread, I think the recommendation was to do the AncestryDNA testing, and upload to other sites such FamilyTree and GEDMATCH to use their available tools.

Is your preference 23andMe due to the "segment data"?

Please forgive me on this - I understand and can explain the DNA concept at a high level, but when it comes to actual testing, I am not sure I understand the differences in each service and what they offer.

Hey Steve,

You can upload your DNA files to:

Personally, I would not recommend using GEDmatch, due to the security issues with the service and their eagerness to undermine their own agreements with users. The data model that they use (anyone can look at the matches of any kit) potentially allows for anyone to extract your raw data from the site, and that is not a good thing for personal, family, or national security. Moreover, you can get segment data from other matches who upload to MyHeritage or FamilyTree DNA, both of which have better-for-privacy data models (i.e. you only see your own kit's DNA matches), so why not use them instead? 

Of course, in both situations, you need to ask and persuade matches to upload to those services. That can be a major bottleneck. Hence my preference for 23andMe is due to segment data being immediately available (for those who have opted in), the ethnicity estimate, and also the comparisons that can be made. 

Ancestry's most useful feature is that so many people have family trees. But it's only useful if you're willing to pay a monthly subscription to obtain full access to those trees. Non-subscribers can only see up to a person's 2nd great grandparents -- at best you'll ID 3rd cousins, but not any 4th cousins. And on top of that, only 28% of my top matches have trees - and many are rather small or not extensive. Not many people at 23andMe have family trees, though one can link to a FamilySearch tree for free. Let's hope that they allow for linking to WikiTree too eventually.


Let's consider an example. Steve has four DNA matches all descended from his 4th great grandfather: Alice, Bob, Calvin, and David. Alice and Bob are siblings.  Calvin is 2nd cousins with Alice and Bob. David is descended from a different child of your 4th great grandfather, so he only matches 10 cM with Alice and 25 cM the others. Only David has a family tree, and it only goes back to his 3rd great grandparents. 

Ancestry would show Steve his  "Shared Matches" with Alice. It would just list Bob and Calvin with no additional information. But if these were on 23andMe, I would be able to see, immediately the probable relationships that I described above because 23andMe also tells me how much DNA Alice shares with each of her matches (the feature is called "relatives in common"). Additionally, because 23andMe has a lower "shared matches" threshold, Alice would also have David listed on her profile. 

That kind of information allows one to quickly draw a sketch of what the family tree might or should look like, e.g. 

So with Ancestry, you can probably build a little cluster of people (a number of people, including myself, do network graphs for this purpose), however with 23andMe, you can immediately work towards a good guess at your matches' relationships, without relying on them having an accessible tree. 

With the 23andMe information, I can roughly sketch out what their tree should look like, and start asking specific questions via messages: "Hey, Alice, I think that we might be 5th cousins. Do you have a family tree? Are any of your ancestors from Texas? I was wondering if you know how you're related to Calvin or David? etc..." 

Often, you'll end up having to do lots of family tree building on your own, in private. I sometimes add down to my matches' grandparents on WikiTree, which is part of why I'm pushing for a new mode for inviting users to join us

23andMe also has a new feature that attempts to predict a user's family tree. Essentially, they can lay out the connections and you fill in the names. It's still in beta, so the functions aren't 100% done yet, but it's promising.

Ancestry's ethnicity estimate is essentially a map with percentages. 23andMe takes it a step further and kind of shows you their match behind it, by showing you exactly which parts of your DNA their model has concluded originated in each place. (see the pictures in this blog post)

Additionally, 23andMe will use your father's and grandfather's results to phase your DNA and your father's DNA, so your ethnicity will have a much higher accuracy compared to Ancestry. You can use that information to help narrow down the origins of a given match to try to see if a segment matches up with a particular ethnicity; that can sometimes help narrow down the search to figure out common ancestors.

One other note on why segment info from 23andMe is important: If your family has any kind of endogamous populations, it's essential to figuring out the true connection. Ancestry usually tells me the wrong ancestor pair for French Canadian DNA matches, since with most French Canadians, I usually share about 20+ common ancestors. Ancestry just picks the one where we've both spelled names the same who is closest to both users. (There's a note in the comments here about this.)

If you are ready and willing to either (A) subscribe and continue paying Ancestry or (B) try doing a network graph, Ancestry could still be useful, but I generally have found greater utility from 23andMe. 

JN, how can "anyone" extract someone's raw data from GEDmatch?

Yikes! JN would you please explain to those of us not as deeply involved in DNA how it is possible to extract any information about someone else's DNA results from GEDmatch? Unless you can get the specific info from every chromosome and all the SNPs I don't see how it's possible. Also, saying that they don't honor their agreements with customers seems to me an extreme statement without some actual evidence of that actually happening. I'm sure we've all heard about some of the recent cases in law enforcement, but I believe they have addressed those issues. Just to be sure, I checked their privacy policy, updated today. Although no one can guarantee absolute security, I believe they're doing as good a job as anyone could. This sentence stood out:

The original Raw DNA and GEDCOM data you provide to GEDmatch is not kept in its original form. It is converted to a form that makes it more efficient for the software to perform searches and comparisons. The Genealogical Data is loaded into a relational database that might still be recognizable as text. The Raw DNA is converted to a compressed binary format in a process we call 'tokenization.' Although the Raw DNA is not encrypted in the usual sense of the word, it would be very difficult for a human to read it. Original uploaded files are deleted from the Site servers soon after they are processed and archived.

Julie and Lisa,

There are two problems with the GEDmatch data model. First, anyone can compare any two kits. Second, users' kit uploads are not checked against some physical proof of origin. With enough comparisons plus localized matching information (segment data), it's possible to determine an exact sequence. 

Last month, security researchers demonstrated that this was possible. While GEDmatch has made some changes in an attempt to prevent this ("As a result of the Washington and another study, we have made several changes and are working on others.  We appreciate the concerns these studies have brought to light."), the attempt will only stop an exact replication of the research which demonstrated the weakness. Moreover, this indicates that GEDmatch has not been proactive in applying security measures. 

This is a fundamental risk of GEDmatch's data model. With other matching services, one can, at best, determine the sequence only of a shared segment of DNA -- and that typically represents only half of a user's genome on that location. So, for instance, if I wanted to know which of my matches carried a particular gene, I could only determine that if I knew where and how they matched me. Thus I would also need to be a carrier for that gene. And they would also need to be a match of mine.

With GEDmatch, it's entirely feasible to make fake-but-apparently-real DNA kits which are undetectable, thereby being able to determine, through matching, who else shares a particular SNP or not. This can be extended beyond particular SNPs to entire genomes. 

While GEDmatch made the change after the research was publicly announced, it was something that people have known was theoretically feasible for a long time. We do not know what private or state actors have already engaged in this kind of hacking. Moreover, GEDmatch can't stop more sophisticated attacks of this kind without changing their data model. Ultimately, all that GEDmatch did was place a small speedbump on the roadway. 

Also, saying that they don't honor their agreements with customers seems to me an extreme statement without some actual evidence of that actually happening.

My earlier comment was alluding to providing police access to the entire GEDmatch database within 24 hours of receipt of the judicial order, despite this immediacy not being required. They had at least 20 days before compliance was required and did not bother to fight it in court. To me that is an action which "undermine[s] their own agreements with users" to limit law enforcement access to data which is not opted-in. 

Moreover, GEDmatch again changed their ToS today to not require that notice of ToS changes be sent to users by email and further to require you to give up further rights and indemnify additional parties. Not exactly a display of goodwill toward users.

Wow! Now I'm more curious. I willingly loaded my two kits and have no issue with those results being used by law enforcement. The other thing is, what can anyone actually do with my DNA?


I personally believe that it's important to consider not just ourselves when considering the law enforcement question. If I were to choose to do that, might it perhaps dissuade some of my cousins and relatives from being willing to test. And it also exposes their information in ways that might be analogous to exposing "family secrets" on WikiTree without consulting family members who are affected. 

But many companies are secretly compiling profiles of consumer behaviour, often in secret, with little to zero oversight as to how the data is used, stored, or applied. I don't know that such companies are not compiling genetic-related information as well. But given this issue with GEDmatch, it would be feasible. Would it not be interesting for an insurance company to secretly know who has Parkinson's disease... and to rig their algorithms to deny service? It could be very profitable, given the high costs associated with certain genetic diseases. 

There are both very subtle ways to discriminate (illegally) against people and very unsubtle (for instance, given American's lack of worker protections, any cause can be an excuse to fire an employee and an alternate cause can be fabricated in most instances). 

We've had this entire argument before.  Yes, we all know that information is being gathered about us all the time--every time we buy something on-line or with a credit card, or do a Google search, etc.  With that much information, I don't know what more the information gatherers might learn from our DNA, or why they would go to the expense to get it.  

If anyone wants my raw DNA data just let me know.


Lisa asked, so I followed up with her. And many others do not know anything about the potential risks, so it's generally good to repeat some information. 

As for your offer, I accept. Please upload your data to https://opensnp.org/ It's easier for me to download from there.

Please don't let your "failure of imagination" limit your ability to respond to knowledge about risks or speak out about them. DNA information provides for those studying human behaviours an underlying variable which might be influencing some of those actions which we take. It could also be used for racial profiling. Might not hurt you, Julie, but it could hurt others. Might not hurt me either, but I'll speak out for the sake of others and try to ensure that anyone who does choose to participate does not do so with their eyes in any way closed. 

OK, just a couple of responses:

As I have said before, no one has the legal or ethical right to deceive their insurance company.

As far as I know, racial profiling is usually done in face-to-face confrontations between law enforcement and people of color.  If you have an example of racial profiling based  on DNA, please provide it.

Edit was to correct typo.
J.N Just curious but what will you do with the kits uploaded to openSNP?


OpenSNP.org is an open website, meaning that anyone can download the data: Scientists (yay!), hobbyists (sure..), hackers (uh...), scammers (whaaa...?), and even foreign state actors (maybe time to rethink this...).

So my remark was meant to show the unlikely nature of people uploading their DNA data to a site where it can be readily accessed by anyone. But that aside, I do like to experiment with data and I have some experiments in mind that involve DNA data. 

Unfortunately, many people who have previously uploaded to GEDmatch, upon learning of the risk, experience cognitive dissonance (a psychological framework for understanding human behaviour), and engage in one or more of the 4 means of reducing cognitive dissonance. The easiest is #1: Don't upload or simply delete your DNA from GEDmatch. The other forms of cognitive dissonance reduction give me comments to reply to in G2G. 

Of course, I nonetheless await for Julie's data to show up on OpenSNP.org. And speaking of her...


As far as I know, racial profiling in usually done in face-to-face confrontations between law enforcement and people of color.  If you have an example of racial profiling based  on DNA, please provide it.

Per the article that I linked, it explicitly shows that racial profiling is not always done in face-to-face confrontations, nor is always done by law enforcement. I do not have evidence that someone is doing it using GEDmatch data, but rather this is a question of what is feasible. (Though Canadian Immigration has a situation which I believe falls in that category.) Just as last year I've warned people that it would be possible to download genetic raw data from GEDmatch, only to have it later published and demonstrated by security researchers, this is a possible use of the data. However, it will be much harder to catch, hence the need to proactively address the risks. 

As I have said before, no one has the legal or ethical right to deceive their insurance company.

Again, Julie, this is a failure of imagination on your part. In at least 3 ways:

1. Most people who are carriers for a high-risk genetic variant do not know. If they have a healthy, functioning copy, they may never know. Additionally, some tests intentionally scrub "medically relevant" SNPs from their data. But, if someone was a carrier and uploaded that data to GEDmatch, it is not only theoretically possible, but a demonstrated technique to determine carriers who might not know can be identified as carrying the high-risk variant. They do not even require to have the high-risk SNP listed in their data file.

2. I believe that there is room for debate on that issue. Canada's parliament definitively decided that Canadians do have an ethical (and now legal) right to keep such data private. But, that is a political opinion that you are entitled to keep.

3. This isn't just about your insurance company or healthcare providers. (Some of us, thank heavens, live in countries with cheap access to universal healthcare!) But corporate and foreign surveillance is a reality in today's world. Human beings are elements in risk models. And people have been discriminated against by employers over their genetic status. That's why both Canada and the US have Genetic Non-Discrimination Acts. 

Genetic screening examines the genetic makeup of employees or job applicants for specific inherited characteristics. It may be used to detect general heritable conditions that are not associated with workplace exposures in employees or applicants. For example, employers used genetic screening in the early 1970s to identify African Americans who carried a gene mutation for sickle cell anemia. Those carrying the gene mutation were denied jobs-even though many of them were healthy and would never develop the disease. In these cases, genetic screening to identify the sickle cell trait often occurred without the consent of the individuals.


But those acts can only address clearly stated discrimination (e.g. "We're firing you / not hiring you because you are at risk of X"), not that which results from secret surveillance (Publicly: "We're firing you because you came in to work late once last Tuesday." Privately: "She's a carrier for X; that is a risk to the company."). 

Or consider credit ratings agencies: Would you give credit to someone whose genes said they were likely to die by 40? That's a major risk of non-repayment! Is it deception to withhold your medical risks from a credit agency? Would you want a credit agency collecting and utilizing a genetic profile of you in determining whether you are "credit worthy"?

In demanding proof that something is already happening, you appear to be oblivious to the precautionary principle:

Two ideas lie at the core of the principle:[7]:34

  1. an expression of a need by decision-makers to anticipate harm before it occurs. Within this element lies an implicit reversal of the onus of proof: under the precautionary principle it is the responsibility of an activity-proponent to establish that the proposed activity will not (or is very unlikely to) result in significant harm.
  2. the concept of proportionality of the risk and the cost and feasibility of a proposed action.

We don't need to see that something bad is already happening before we take steps to prevent it. 

Understood JN. Excellent arguments about the safety of our most personal information going forward. However, "Julie, this is a failure of imagination on your part. " is a disservice to Julie. Consider instead, that she is giving the moral and legal implications of the fraud that could result.

Also, your definition of cognitive dissonance is slightly off. CogDis (abv.) is the name of my World of Warcraft guild populated by educators and educational administrators based on the definition suggested by our GM, (who was a PhD candidate at the time of its inception). Meaning: the state of having inconsistent thoughts, beliefs, or attitudes, especially as relating to behavioral decisions and attitude change

In our case, educators playing and exploring a MMORPG for its' educational possibilities instead of the expected revulsion one might expect from said group. 

JN, yes, I have heard of the precautionary principle, originally with respect to global warming, and I can not resist noting how striking the last line of your post is in that respect, as the human race marches foolishly towards its own destruction by climate change.  But irrelevant, I know.

Back to DNA.  I do not think the precautionary principle requires us to imagine every theoretically possible outcome and take steps as if it were likely without any evidence.

The example you cite of employer discrimination based on genetics was a woman who'd had a preventative double mastectomy.  Whether or not the employer actually fired her for that I have no idea, but if they did it wasn't based on analyzing her DNA that they obtained from GEDmatch (which is what I thought we were discussing).

If credit issuers were likely to deny credit to people calculated to have short life expectancies, I would expect to see them rescinding the credit extended to me and my many fellow senior citizens, as we are obviously nearing the point when we could die at any time, but I haven't heard of that happening.

You seem to be saying that many institutions are going to spend significant resources obtaining and analyzing DNA,  and then make irrational decisions based on the very complicated results.  I don't see that happening.

Edit was a minor clarification.
+4 votes
You already have a lot of responses, all with good information to assist you with making a decision.  I don't think anyone has mentioned that FTDNA stores the DNA sample.  I consider this a major point and reason to test your grandfather (and any other grandparents still alive) with FTDNA.  Today you may only be interested in autosomal, but down the road you may decide you are interested in mtDNA or Y-DNA.

I initially had my dad's mtDNA tested in 2003 at FTNDA.  It was to help prove a matrilineal line. Two years later I had them test his Y-DNA, and nine years after that his auDNA.   I need to see if they have any left or if I should get him to do another sample.

With the advances in DNA, we don't know what they'll be able to do in a few more years.  I prefer to have my dad's sample available for more tests.  Just something to consider . . .

If you really get into genetic genealogy, you'll ultimately want to be on as many sites as you can afford.  Once you have both of your parents tested, you'll want to work off of their tests for analysis.  Adding in your grandfather will be a huge help.  I've got my parents on FTDNA, ancestryDNA, 23andMe, MyHeritage and Gedmatch.  I don't see a 'con' to any service (other than the lack of a chromosome browser with Ancestry).  Ancestry has a huge database and more people with a family tree.

If funds are limited, I would focus on grandparents right now.  I would test all grandparents with both ancestryDNA and FTDNA.  As you can probably tell, we all have different ideas/opinions about what we would do.  The only reason I did 23andMe was to try and locate more matches, as my experience with people on that site is that they tested there for the health information and aren't as interested in the genealogical aspect of DNA.  As such, fewer of them have trees (or much knowledge of their ancestors).  That's just been my experience with matches, though; others may have different results.
by Darlene Athey-Hill G2G6 Pilot (401k points)
+4 votes

Many people here have provided excellent advice. Personally, since your Mom has already tested with Ancestry, I would probably start there.  Then as others have mentioned, you can transfer each to My Heritage and or FT DNA for additional tools. And of course upload the results to GEDMatch as well. GEDMatch is unique in that it allows you to compare your DNA to others who may have only tested at one of the other companies and may never transfer the results anywhere else.

One of the things to keep in mind is that each company will have slightly different "ethnicity" estimates.  Just keep in mind that these are just that... estimates. My Ancestry ethnicity results are different from FT DNA, and My Heritage, even though it was derived from the same Ancestry DNA test.

No matter what you decide to do, there is additional advice on DNA testing here: https://www.wikitree.com/wiki/Help:DNA_Tests#Ordering_a_Test

... and be sure to use one of the links on that page or here:

Ordering a Test

To order a test, or just check for sales, please follow these links:

If you are e-mailing a family member or friend or posting on Facebook or a forum, use one of these URLs to make sure that WikiTree gets credit for the referral.[2] Thank you!

And most of all GOOD LUCK!!!

by Ken Parman G2G6 Mach 3 (39.9k points)
+3 votes
Hi Steve, this is a good question.  The answer will rest in what are you trying to find out?

Autosomal is the least expensive of the tests.  A number of labs do this test.  The pros are it gives matches for both mother and father branches.  The cons are you often can't tell which side the match is on but since you are testing your father and grandfather you will be able to tell by who matches them and you who is coming from your mother's side (no match to your father or grandfather) and who is coming from their side.  

The difference among the labs testing autosomal is that some labs do not provide the specific chromosome and SNP levels needed to make what ISOGG standards set out for determining a real match.  Ancestry currently does not provide this.  FTDNA does if you dig below the chromosome browser.  Gedmatch does.  So in my opinion use Ancestry to get the hints to a wider database but confirm the matches in Gedmatch.  And to get a broader view pay $19 and upload to FTDNA as well.  The 3 give you a great view.  

Now if you want to trace down your surname route back farther than 4 generations you might want to do Y testing.  That is not provided by all labs but is provided by FTDNA.  There are varying levels of this testing at varying price levels.  Depending on what you are trying to learn you may want to consider doing a Y test and an autosomal test.  

Hope that helps.   There are sales going on for the Holidays.
by Laura Bozzay G2G6 Pilot (643k points)
+2 votes
I know there would be some issues doing this as each company does their testing differently, but if on a tight budget, I'd go Ancestry then download the Raw DNA data & upload to MyHeritage, FTDNA, GEDmatch etc etc.

(simple, quick, dirty, cheap solution)
by Kylie Glynn G2G3 (3.5k points)

Related questions

+10 votes
2 answers
+4 votes
2 answers
+7 votes
7 answers
186 views asked Jul 24, 2019 in The Tree House by Becky Troth G2G6 Mach 4 (42.8k points)
+6 votes
5 answers
+4 votes
1 answer
+17 votes
2 answers
+13 votes
4 answers

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright