Comparing results from actual recorded ancestry, to that predicted by Ancestry.com, 23andme, My Heritage, Living DNA, FT-DNA and more.

Recorded Ancestry

Results

During the Black Friday sales last December, I bought two Ancestry kits. I actually mean't to order one, but made a bit of a mess of it. Still, I thought, they were as cheap as I've seen a DNA kit, so I let the order process.

Some people might exclaim - but you've tested your DNA to death!  These kits were not directly intend for myself though.  I intended them for the art of Genetic Genealogy.  To help me verify my paper tree, biologically.

Although I enjoy the kick that I get from matching segments of DNA in strangers, to shared ancestors of the past on all lines, I'm particularly interested in one line, from one great grandfather.  You see, I had a very naughty great grandmother.  I have uncovered evidence of two bigamous marriages by her, as well as other relationships.  A second cousin of mine, through her, doesn't appear to have the amount of shared DNA that I would expect for a full second cousin.  It looks worse than even the old family rumours.

As I do have an extensive family tree down to that birth certificated great grandfather, even though I know full well that biological family isn't always as good family as non-biological, which the paper trail honours, I'd still like to know.  With Genetic Genealogy, I hope to verify - or otherwise, his biological relationship.

So... I used one kit to test one of my siblings, and the other to test my mother.  I've tested my mother before on 23andme.  Mistake.  I've learned a lot about DNA testing over the past few years or so.  Ancestry.com might seem like a heavy marketing, greedy big DNA company, with some slightly dishonest sales ploys (find out if your ancestor was a Viking!), and pressure to subscribe to more services in order to get the full benefit of the test - BUT ... it 1) has an awesome family tree building website for subscribers, that link to DNA tests, 2) has the largest customer database, and 3) through it's genealogy services, as well as marketing, has the most UK testers in it's database.

Okay, it's a little dumbed down.  The messaging system sucks (so I always send my email address), It doesn't provide a chromosome browser.  It doesn't provide segment locations on chromosomes.  But - for my uses - using DNA matches to verify a family tree pedigree, it serves extremely well.  I have had almost ten times more matches on AncestryDNA, than from 23andme, FT-DNA, and GEDmatch combined.  And many have online trees!

I've received my siblings results.  Wow.  I suspected it.  That the sibling has inherited some quite different DNA from the parents mean't that although we share some DNA matches, there are many that we don't!  Up to now, I've just used a spreadsheet to keep results of verified matches.  I could see that I now need something more powerful.  Something that I could search on - and filter different lineages.  When my mother's results arrived, I'll be able to divide all of my matches into maternal, or paternal sides.  On top of that, I have a 1C1R (first cousin once removed) on my father's side, that I can sometimes use to indicate some ancestry on his side.  I can look at all of my matches and their shared matches, and triangulate, where abouts they fit into my family tree.  I built a personal database for my DNA matches.

So I'm pretty pleased that I invested in those two kits during the sales.  It's kept me busy.

I used Open Office Base to build the database:

Okay it's basic and not pretty, but I can extend on it.  I've imputed our closest 187 DNA matches, nearly all from Ancestry, plus a few verified from FT-DNA and GEDMATCH.  It's a family match - I've included forms for imputing my mother's and sibling's matching segments - not just my own.  Any genuine matches that my sibling has - are also my cousins.  Just that I don't have personally share DNA segments with them.  I've also included a yes/no check box for that 1C1R.

I've used it to query an up-to-date list of "our" shared DNA matches that share a correlating common ancestor or two on their trees with ours.  My biological "verifiers".

Using the open source GRAMPS app, I produced a fresh family pedigree fan chart.  I then used open source GIMP to colour in the ancestors that I have verified with shared DNA segments.  The darker the tone, the more matches:

It's generally looking pretty verified isn't it.  My birth certificate grandparents were all very clearly, my biological grandparents.  The great grandparents, and the majority of great great grandparents are also looking pretty verified.  But what about that great grandfather?  The birth certificate version was my surname great grandparent, and biological version was my Y-DNA great grandparent.  Were they the same?

Well I still do not have evidence that I'd regard as overwhelming.  But I am gathering evidence that he may have been the same guy.  I have two DNA matches that strike directly through him.  Unfortunately, both were distant ancestry, with only a small shared segment each (around 7 cM).  That small, they could either belong to an undocumented relationship elsewhere, or even be identical but not by descent.  But it's evidence that I'm building, and it's more reassuring than if he'd had no DNA matches strike through his lineage to us.  The other supportive evidence was that my biological paternal line great grandfather carried an incredibly rare haplogroup: Y haplogroup L-SK1414 (L1b2c).  The only other L-SK1414 so far found in the British Isles, traced his paternal surname line back to Basingstoke, around 1740.  My documented surname line traces back in 1740 to Long Wittenham, Berkshire.  Only about 32 miles away from the Basingstoke L-SK1414 by road.  Could be a coincidence, but it supports that the Y-DNA could still correspond with the surname line back in 1740, and that my great grandaddy, was my DNA great grandaddy.

Such is the power of genetic genealogy.  Roll on the results of my mother.  That will reduce the number of matches that are likely to be on my paternal side.

I have new DNA cousin "matches".  This is a very important avenue of DNA testing for genealogy and ancestry that I have simply missed until recently.  Up to now, I've concentrated on DNA testing for general ancestry (or ethnicity as some businesses will call it).  The problem was that I first tested with 23andme, and simply, using their heavy USA customer base, and user unfriendly "experiences", I couldn't find any DNA relatives that actually had paper trails that could correlate to my own.

One of the problems is I feel, is that an awful lot of Eastern English migration to the Atlantic Coast of North America, occurred very early - late 16th to early 18th centuries AD.  As a result, although some generous matching systems (such as 23andme's) suggests much more recent shared ancestry, in reality, our links to our distant USA cousins are so old, that all they do is reflect that my distant cousins have Puritan, New England, and Virginian ancestry from Eastern England.  Even for those that do claim to trace ancestry to those pilgrim fathers - I can't.  Certainly not for the thousands of my direct ancestors for Generations 11 - 14.  I don't think any of us can.  Chuck in a bit of genetic folding, and all that these distant relationships is really telling us is, that we both have some ancestry from south east England between 300 and 600 years ago.

Then I tested with Ancestry.com, Ancestry.co.uk, AncestryDNA or whatever you want to call that genealogy mega-business.  Their matching system is dumbed down to the frustrating level.  No chromosome locations or chromosome browsers for painting.  Instead however, they have the fattest database of testers and customers - some of whom, will like myself, be subscription slaves to their family tree and documentary genealogical services.  Their matching systems may cut out chromosome data - but on the flip side, you can browse trees, surnames, ancestral locations, of your DNA matches.  As a consequence, I've found 14 matches that share DNA, with predicted relationships - that correlate to a paper trail relationship.

In addition I am now scouring GEDmatch, 23andme, and FT-DNA Family Finder for more relative DNA matches.  I'm recording everything (including chromosome locations when available) onto a spreadsheet.  The image at the top of this page demonstrates my DNA matches where they share ancestry so far.  The darker the shade, the stronger the verification.

I'm starting to see how this is a better tool to understanding, or verifying ancestry, than any stupid ethnicity / ancestry composition by DNA.  Family isn't always biological.  However, finding a genetic correlation is the ultimate evidence to strengthen a tree.  It's fascinating to see actual paper research turning up as segments of inherited DNA on matches.

This is an aspect of Genetic Genealogy that I'm sure is well known to some researchers, but that, I'm only just starting to appreciate.  I've been DNA testing for ancestry heavily for a year or two, but my prime interest has been older ancestry, admixture, and population genetics.  All of my early attempts to contact matches through 23andme and GedMatch, resulted in frustrated conversations with North American testers, that had no paper trail back before their ancestors emigrated.  Today, I matched on AncestryDNA with a third cousin.  The DNA prediction was fourth cousin, but the relationship on paper is third cousin.  This was my third match, confirmed by both shared DNA segments, and by a shared paper trail to common direct ancestors.  How cool is that?  Finding that yourself, and other researchers, share segments of the same DNA that appear to have been inherited through recent common descent.  Finding each other through the code of Life that is in our cells, and being able to see where that DNA came from in our family trees!

I know I should have smiled!  Me, myself sitting outside of the archaeology museum earlier this year, at Sofia, Bulgaria.South-West Europe.

FT-DNA Family Finder My Origins

I'm writing this post in response to a number of comments that I see online with regards to using a commercial DNA test, in order to ascertain ancestry.  Quite often, when someone asks how to find out their family history or ancestry, someone will come back with an answer in the form of "just spit in a vial, send it to Ancestry.com, and they'll tell you".  It's not really that simple, so I'm making this post, to explain how an ancestry DNA test can help, or not help, you discover your ancestry.  Nicely dumbed down I hope, for the beginner.

Traditional Genealogy

Traditional genealogists usually set out to create a genealogy (family history and tree), using interview techniques, artefacts, and oral memories, recorded from older relatives.  Artefacts might for example, include old family medals, or photographs.  They then extend the research, through documentary evidences, such as birth, death, and marriage certificates, church registers, census records, transcripts, electoral rolls, and military records. If they are interested in recording all ancestral information, and not merely a single line such as the surname line, then this research can go on for months, years, even decades.

What you cannot do, is to simply pay a small fee, and your entire family history drops through the letter box in a brown envelope.  It takes years of time to research, collate, and to verify a good family tree.  Most genealogy enthusiasts don't mind this, because they actually enjoy doing the research itself.  It becomes a hobby, even sometimes a passion.

However, a number of commercial DNA companies may give the general public the impression, that you now can simply pay a fee, spit or swab, and your ancestry magically appears for you on a website.  It's big business.  Does it work though?  Exactly what is genetic genealogy?

What is Ancestry and why do we care.

Ancestry can simply be defined as our descent from forbearers.  Why do we care who they were? Which forbearers or ancestors?  How many are there?  How far back?

Of course, not every one does care.  Not everyone cares about history.  But for other's how we define ourselves, our communities, and families, it does matter.  It tells us who we are, where we came from.  It defines us, gives us grounding.  It gives us identity.  Wars have often been inspired by ancestry.  At the same time, a deeper appreciation of the human family, and it's common ancestry, can be used to relate to those elsewhere.  One big family.  Discovering the immense poverty and hardships of our ancestors can help us to appreciate what we have, and to help others in need today.

So what ancestry can we discover?  For those few that merely concentrate on one patriarchal line, it's quite simple to define - the generations of a surname.  However, beyond that one narrow line of descent, few appreciate exactly how much total ancestry that we have.  Lets look at our biological ancestors at each generation:

• 2 parents
• 4 grandparents
• 8 great grandparents
• 16 great grandparents
• 32 g.g grandparents
• 64 g.g.g.grandparents
• 128 g.g.g.g grandprents
• 256 g.g.g.g.g grandparents.

These are only your 510 most recent direct ancestors, yet just those generations, will take you back to only around 250 years of family history.  Now add all of the recorded children of these direct ancestors - the great great uncles and aunts to the theoretical family tree.  You're probable going to have a tree of around 1,300 individuals.  That is just for 250 years.  You have a big family  Go back a few more generations, and it will explode before you reach far.  All of those direct ancestors though, are a part of your ancestry.  You'll most likely carry some DNA from most of them.  They are, from a biological perspective, who you are.

By the way, the number of biological ancestors will not continue to increase infinitely.  Because increasingly, you will find couples within your tree that are distant biological cousins of each other.  As this accelerates through thousands of years, that explains how all modern people around the world, all descend from a very small population around 100,000 years ago.

So before considering what DNA can do for genealogy, we need to consider which ancestors matter to us.  Do you just want to know who your biological parents, or grandparents were?  Do you want to know the names, places and social positions of your ancestors over centuries?  Do you want to know which parts of the world that your ancestors lived 500 years ago?  Do you want to know how some of your prehistoric ancestors moved across the globe, thousands of years ago?  Maybe you want to know everything.

Let's now turn to genetics for genealogy, and how DNA tests can answer some of these questions.

There are two main types of DNA tests for ancestry, although they are often incorporated together by commercial companies:

1. The haplogroups, the Y-DNA and mt-DNA
2. Autosomal DNA

The haplogroups are chains, or markers, that are carried on one of only two strict lines of descent.  They do not apply to your entire ancestry - just two lines.  As we saw above, we have 256 g.g.g.g.g grandparents (unless any of their descendants reproduced together).  Our haplogroups came from only two of them.  Your haplogroup does not define you.  Yet, it's quite odd, because very quickly, many genetic genealogists do relate to them, rather like a hereditary football club.  They do become an identity, only if you enthuse over them.

The Y or paternal haplogroup, follows the strict paternal line.  From father to son.  Women do not have a Y chromosome, so cannot pass it on.  It has to come from the biological father.  However, within this constraint, Y-DNA is particularly useful to genealogists.  It mutates often, both as STRs and less often, as SNPs (snips).  Because of these frequent mutations, it is useful for tracing shared descent with others.  It can also be aligned with surname studies.  The champion commercial DNA company for Y-DNA research, is Family Tree DNA.

The mt or mitochondrial (maternal) haplogroup, follows the strict maternal line.  From mother to children.  Both sons and daughter inherit their mt-DNA haplogroup from their biological mother.  However, only the daughters can pass it down.  Two downfalls to mt-DNA for genealogy.  1) The surname frequently changes, traditionally nearly every generation through marriage. 2) it doesn't mutate as frequently as the STRs of Y-DNA. It is still a useful tool, and can prove descent through the maternal line.  It can also still be used for studies of much deeper, ancient ancestry.

This is the bulk of you DNA.  All of the snips (SNPs), that make up who you are genetically.  You receive approximately 50% from each parent, 25% from each grandparent, 12.5% from each great grandparent.  This subdivision cannot go on forever, and indeed, once you go back much more than six generations, the approximates start to deviate, so that you may have no snips at all from a particular line that joined your family tree over 250 years ago.

The problem with autosomal DNA is that it can be such a mess.  It recombines randomly with every generation.  Therefore, it is much harder to track ancestry in the same way, that we can with the haplogroups.

So how can they be applied for genealogy:

Not everyone knows who their biological parents were, or where they came from.  This is the first use of DNA testing.  It can be used to find, test, or prove recent descent.  The first hurdle of genealogy.  Both haplogroup evidence, and autosomal evidence can be used to prove or determine relationship.

Cousins

Many genetic genealogists, use DNA to find distant, and sometimes not so distant cousins.  The hope is that they can link trees, share knowledge and research, perhaps copies of artefacts.  Therefore an awful lot of genetic genealogy is about tracing genetic relatives, and establishing common ancestry.

There are two main tools:

• Haplogroup Projects.  The Y haplogroup is favoured for it's frequent STRs, and also for it's link to surnames.  Family Tree DNA projects track the STR and SNP data of it's members, tracking families, relationship, known mutations.  Project administrators at FTDNA can predict relationship to other members in the project.  Your Y cousins.
• Shared segments.  Autosomal DNA can be used for finding distant cousins.  23andMe for example, have Relative Finder.  Alternatively customers of any commercial DNA company that gives them access to their raw data, can upload that data to GEDMATCH.  At GEDmatch, they can search for other kits, looking for lengths of shared segments (measured in cM - centimorgans) on the autosomes or X chromosomes.  The longer or more segments can be used to indicate shared ancestry.

It is important to understand, that this is not about directly tracing ancestry.  It is only about establishing shared biological ancestry, with other researchers, with which you may be able to share resources.  In the old days of genealogy, we would find distantly related researchers by browsing through annually printed surname interest directories.  Here, the same thing is happening, but we are finding people by comparing DNA.

Ancestry from Autosomes

Most commercial DNA companies providing ancestry information, now use their own propriety calculators to look at the autosomal DNA of their customers for patterns that they can relate to a number of reference populations.  23andMe for example, uses Ancestry Composition to determine what parts of the world, that the ancestors of their customers lived 500 years ago.  They predict from this in percentages of ancestry.

However, it is very much a developing art.  The problem is that genes have been randomly mixing and moving around ever since prehistory.  The customers of these DNA companies want hard facts.  They want their ancestry accurately pin pointed down to modern or ancient nation-states, or to historical populations such as the Vikings or Huns.  Ancestral DNA companies are under pressure to provide this deep ancestry.  However, can they?  Ancestral analysis of DNA can be very enlightening.  It can provide some surprises within a family history.  However, it's accuracy is exaggerated.  It is increasingly successful at predicting ancestry from a particular corner or end of a particular continent.  But it cannot for example, accurately tell French, British, and German ancestry apart to any high accuracy.  It can recognise some populations better than others.  It cannot tell anyone if they had Viking ancestry.

Ancient Ancestry

This is a particular value of the haplogroups.  As we accumulate more and more data on more mutations, as we expand the recorded database, and as we relate that to more ancient DNA extracted from referenced and dated ancient human remains, so we will be able to better explore the population genetics not only in history, but deep into prehistory.

However, it is also becoming increasingly realised, that patterns of ancient admixture can also be detected within the autosomes.  Although Autosomal DNA ancestry calculators claim to reveal relatively recent admixtures over the past 500 years, it is becoming clear that these are being confused by much older patterns of admixture.  These patterns can now be explored and probed on a number of GEDmatch programs.  People can compare their DNA with the kits from ancient DNA, or predict just how much of their ancestry was likely "Western Hunter-Gatherer, or "Early Neolithic Farmer".

In addition, more DNA companies are now measuring for much more ancient admixture with archaic populations such as the Neanderthals.

Conclusion

Genetic Genealogy is fun, great fun.  It is not however, a quick and easy replacement for traditional genealogy.  Unless you get lucky with some comparative Y-DNA in a project, it is not going to directly tell you the names or social status of any ancestors.  It can give you a phylogenetic tree, but not any kind of family tree that you can bore other family members with.

Genetic genealogy can provide some tools to some researchers.  It can test biological relationship.  It can be used to predict some of your ancient history.  For most researchers, particularly those that are able to interview many local family members, search local grave yards, access archives and records - it has no, or little value to the pursuit of collecting ancestors.

I personally love to explore my genetic genealogy. But it is documentary research that provides the names.  Genetic genealogy for myself, is more about the long and ancient journey.

The above photo at A Capela dos Ossos (the bone chapel) in Évora, Portugal. The entire chapel is covered with human bones.  Every wall and pillar is decorated with skulls and bones.  On another wall hangs the mummified remains of a man and child, said to have been cursed. There is a sign at the entrance of the chapel which states "Nós ossos que aqui estamos, pelos vossos esperamos" (Our bones here, await yours).

Genetic Genealogy

I was a sceptic of genetic genealogy, I'll admit it.  Now I'm hooked.  Not because I feel that it has been a way of hooking up with distant cousins, that can help me extend my family tree.  That's not the way that I've used it so far.  Instead, it has provided very different kind of information, that helps me understand who I am, and how I can link my ancestry to known heritage.

I might not have been so hooked, but I've had so many surprises with my 23andMe results.  If my results had been perhaps, dire and boring, then maybe I would have retreated to traditional genealogy and regarded the technique as predictable and uninteresting.  However, what ancestry related surprises did I have?

• I have a very rare Y haplogroup for NW Europe.  So far predicted to L1b M317.  It will be shared by my brother, my son, one cousin (and his son, and grandson).  Today I sent away a further FTDNA Y111 swab test.  The L haplogroup is mainly concentrated in Southern and Western Asia, from Afghanistan down to Southern India.  My L1b M317 sub clade is concentrated in Western Asia, including Eastern Turkey, Armenia, Georgia, Azerbaijan, the South Caucasus, and Western Iran. A faint trace of it along the length of the Med in Southern Europe, and across Italy, and a slight cluster in central Europe - which apparently, I don't belong to.
• Autosome Ancestry composition by 23andMe, gave me a very low percentage of "British & Irish", and high percentages of "French & German" and "Scandinavian".  I've explored the possibility that this could reflect early medieval admixture from across the North Sea.  I've looked at the typical Ancestry Compositions of people with a strong recorded English ancestry, and compared them to the results from people with strong Irish ancestry.  That SE English people typically sit somewhere between the Irish, and typical Dutch in Ancestry Composition reinforces my view that this is the case.
• My mtDNA was H6a1.  Not the most exciting haplogroup, but not the most boring neither.  It allows me to relate to the latest evidences for Eurasian Steppe admixture into Western Europe during the Early Bronze Age.

A Southern European Enigma

I captured the above photo at Cabo Espichel, Portugal.

There was a fourth, further surprise in my 23andme results.  It lay in the autosome.  23andMe AC (Ancestry Composition) on speculative mode, suggested 2.4% Southern Europe, including a prediction of 0.5% Iberian ancestry.  On speculative mode again, it falls on five pairs of chromosomes - but never on both sides.  On standard mode, 0.1% remains, just on one side of pair 21.  This suggests that all of it comes from just one of my parents.

I might think that this was just "background noise", an error in AC.  However, it keeps popping up.  Indeed when I upload my raw data to the program at DNA.land, they predict only 80% North/Central European, and a whopping 15% South European.  It doesn't stop there.  On GEDMATCH, the Eurogene calculators keep suggesting Iberian or South European admixture on their mixed population oracles.  Eurogenes K9 for example, gives me 61% North European, 29% Mediterranean, and 6% Caucasus.

Let's just refer back to my recorded paper ancestry.  I have 190 recorded ancestors, all in England, with English surnames.  No sign of any Roman Catholicism.  I have all sixteen of Generation 6 (G.G grandparents) named.  All born and named English.  No sign of any South European even in the 1,490 people on the entire family tree for my kids.

However, I think that all of the autosome ancestry calculators could be telling me a truth, that I can't see in my known family tree.  If I have a South European ancestor somewhere, whether Iberian or not, then either a) I have not yet found them, or b) they were the biological ancestor of a NPE (non-parental event), a cuckoo.  I have 3 out of my 32 Generation 7 ancestors unnamed - all absent fathers.  I have 15 missing ancestors in Generation 8.  Above that, the representation really starts to decline, although I have some ancestors named up to Generation 11.  Could a South European be in there?  23andMe in speculative mode suggested 2.4%.  That would seem "average" for an ancestor in Generations 7 or 8 (3 to 4 x G grandparent level)  Of course from around that point, "averages" become pointless, and subject to a randomness that can delete entire lineages further up from any surviving DNA.  None-the-less, I could have a South European from around that period - either one of the 18 "missing" ancestors, or a NPE cuckoo.

I'm commissioning a 23andme test for my mother.  Three reasons.  1) she wont be here for ever.  Recording her genome feels valuable and worthy.  2) I want to see how her very dense 100% recorded Norfolk ancestry projects on Ancestry Composition and on GEDMATCH.  3) I want to phase her results against mine.  It will tell me for example, where my "South European" DNA came from - which parent.  It will help me further understand my own genetic ancestry.