Friday 30 April 2010

The new Family Finder test from FTDNA

In February this year Family Tree DNA announced the introduction of an exciting new DNA test called the Family Finder. The test is being rolled out in a phased release, and is currently available to existing customers only. It is scheduled to be launched to the general public in the coming weeks. Traditionally family historians have used Y-chromosome (Y-DNA) tests to look for surname matches in the direct paternal line, and mitochondrial DNA (mtDNA) tests to find genealogical matches in the direct maternal line. The Family Finder looks at the 22 autosomal chromosomes to find matches on all your ancestral lines, but is best suited for finding matches with close cousins up to the fourth or possibly fifth cousin level. The test works by locating shared segments of DNA and predicting relationships based on the number and length of shared segments. FTDNA have provided an excellent set of FAQs (frequently asked questions) on the new test which can be found here. The FAQs are updated at regular intervals so it is worth checking back from time to time. A demo can be seen here. If you already have an FTDNA account make sure you are logged out before accessing the demo page.

Autosomal DNA is shuffled up and becomes diluted with each new generation. The following figures show the average amount of autosomal DNA shared with close relatives:

50% mother, father and siblings
25% grandfathers, grandmothers, aunts, uncles, half-siblings
12.5% first cousins
6.25% first cousins once removed
3.125% second cousins, first cousins twice removed
0.78% third cousins
0.20% fourth cousins

The percentages can vary. A son might for instance inherit 53% of his DNA from his mother, but only 47% from his father. While the Family Finder test is sensitive enough to detect shared DNA for most third cousins, some fourth cousins will test and not have a match. It is recommended wherever possible to test the oldest generations in your family to maximise your chances of finding matches with your more distant cousins. Rather than test myself I have therefore had tests done on my mother and father as part of the beta-testing programme. My dad's results have now come through. My mum unfortunately sent off the vials without including the brushes and is having to redo her test so it will be a while before I get her results! When the results came in I was very surprised to discover that my dad had 12 matches despite the fact that the test is still very new and there can be no more than a few thousand people in the database at present. One of his matches is in the second to fourth cousin range. The remaining matches are what FTDNA terms "speculative" and are in the fourth/fifth to distant cousin range. The screenshot below shows how the matches are presented with the names of the matches obscured for privacy. You will need to click on the picture to enlarge it to see the details.
As can be seen, the Family Finder picks out surnames in common and highlights them in bold. In this case it matched Reid with Rudd and Peden with Paddon, though as far as I can establish there is no known genealogical connection. A chromosome browser is provided so that you can see the location of the shared segments. The screenshot below shows the chromosome browser view with my dad's presumed second to fourth cousin. Names have been removed for privacy.
In this case my dad shares a chunk of autosomal DNA on chromosome 16 with his newly found cousin. If the relationship was closer there would be more matching segments. With a more distant match the segments would be smaller. It is also possible to compare the matching segments for up to three people at a time. The screenshot below shows a comparison between three matches in the fifth to distant cousin range who all have correspondingly smaller chunks of matching DNA, again with the names blanked out.
The Family Finder test will help to identify relatives but establishing where they belong in your family tree can only be achieved by traditional genealogical research. To get the best out of the test you will therefore need to do your own research on all your lines for at least three or four generations. You must also hope that the people you match will have done the equivalent research too wherever possible. I have now contacted most of my matches but we have not been able to find any connections as yet. The closest match lives in America. All her lines are from Germany and Denmark apart from one line which is from England. Unfortunately she has not yet been able to establish where in this country her English line originated, and the surname is not one which appears in my tree.

The speculative matches at the fifth to distant cousin level will in most situations not be worth pursuing because of the difficulties in researching every line back this far. I have already done a substantial amount of research on my father's line but it was quite sobering when I looked at his pedigree to see how far back I could trace all the branches.  Although I can trace some lines back to the 1600s, and in some cases back to the 1200s, I have been less successful with my research in some of the other lines. I can currently identify fourteen of my dad's sixteen great-great grandparents, but just eight of his thirty-two great-great-great grandparents. In most instances I've taken the line back beyond the censuses to a specific parish, but a substantial amount of parish register reconstruction in many different counties would now be required to trace these lines further back in time. I also have a substantial brick wall with a certain William Hunter who was born c.1798 in Scotland. He married in 1828 in Stepney and appears in both the 1851 and 1861 censuses in Limehouse in the East End of London where he rather unhelpfully tells us that he was born in Scotland with no indication as to the county or parish. There are rather a lot of William Hunters born around this time on the Scotland's People website and it would be a huge undertaking to trace and eliminate each one. I suspect that because of the large number of Scottish emigrants to America William Hunter will be responsible for the majority of my dad's matches with cousins on the other side of the Atlantic.

There are further Family Finder features in development. The test will eventually give percentages based on ethnic origin, though this feature would probably not tell me very much as I anticipate that I would be 100% European on both my maternal and paternal lines. More interestingly, I understand that results will eventually be reported for the X-chromosome, and there are plans for a separate X-chromosome browser. The X-chromosome has a special inheritance pattern which makes it easier to pinpoint the ancestor contributing the shared X-chromosome so this should be a very useful tool.

The Family Finder test is not a replacement for Y-DNA and mtDNA testing, but it can be a useful complement to those tests, and can be particularly useful for proving relationships in the last four or five generations. As with all DNA testing it is a tool which needs to be used in conjunction with the paper records. Its value will grow as the database becomes larger and there is more chance of finding meaningful matches. I've seen a lot of orders going through my FTDNA projects, and I'm sure that it will not take long for that to happen.

Updates
With effect from July 2013 the price of the Family Finder test has been reduced to just $99 (£60). See my blog post Autosomal DNA testing is now affordable for all for further details.

An X-chromosome matching service was added to the test in January 2014. For details see my blog post on Family Finder X-chromosome matching.

Read my article An autosomal DNA success story to understand how the test can work in practice.

See also my article Tracking DNA segments through time and space.

© Debbie Kennett 2010

26 comments:

Mavis said...

Debbie,

Thanks for the review of the Family Finder test. I'm anxiously awaiting my dad's results. Will test my mom at a later date.

Debbie Kennett said...

Thank you Mavis and Andrew for your kind comments. It's a very exciting new test, and it will be interesting to share some of the success stories.

Nancy Rivers said...

I have received my results - and had several matches. 2 at the 4th cousin level, although we have yet to find the connection. I did have a cousin contact me who matched at the 5th to distant cousin level and I found 6 ways that he and I are connected! I've also been blogging about it in case you are interested: http://nancycrivers.com/blog/

Debbie Kennett said...

Thanks Nancy. You've done an interesting write-up and you did well to locate the possible match in the paper trail.

Bernadette Olden said...

I had my dad's DNA tested through FTDNA right before he died at age 90, 2-1/2 years ago. He was an only child and so am I, so I was hopeful of finding connections to his line through DNA. So far I haven't found any. Do I understand correctly that if I wanted to have his DNA tested for the family finder I would need another sample, which is of course impossible?

Debbie Kennett said...

Hi Nona,

FTDNA keep all the DNA samples for 25 years so they should be able to do a new test on your dad's sample if that's what you want. Get in touch with them and explain the situation. I'm sure they'll be able to help.

Unknown said...

Excellent review all stistics are dead on.

Each child always gets 50% of their autosomal DNA from Mom and the other 50% from dad.

a given grandparent, however, may not contribute the round 25% of their grandchild's autosomal DNA because of the phenomenon of crossing over. That is why the predicted relationship will become less precise with each generation of separation.

Still this review nails how exciting the program is and the statistics are inordinately accurate!

Debbie Kennett said...

Benjamin

The exciting thing about these new tests is that if you test close family relations you can actually see for yourself the percentages of shared DNA. Children do not always inherit exactly 50% of their DNA from each parent. Tim Janzen has compiled a spreadsheet with results from known relations who have tested which you can find here:

http://kquilting.homeserver.com/23andme/ref.html

As you can see, the percentages vary from about 47% to 53%. These data are derived from the 23andMe test which is very similar to FTDNA'S Family Finder test. 23andMe customers tend to take the test mostly for medical reasons, and it is less useful for finding genealogical matches as a lot of the matches aren't interested in genealogy and don't want to correspond.

Anonymous said...

Regarding the apparent 47/53 split... Isn't it more likely that this represents the inevitable inaccuracies of the methodology, rather than requiring an exotic uneven split of parental DNA ?

Debbie Kennett said...

The methodology is correct, but autosomal DNA is inherited on a random basis. The 47/53 split I quoted is perhaps an exaggeration, but there can sometimes be small differences. There are some statistics on this ISOGG wiki page which you might find of interest:

http://www.isogg.org/wiki/Autosomal_DNA_statistics

Andrea said...

A son receives an X chromosome from his mother and a Y chromosome from his father. In terms of size, the X is much bigger than the Y. The X is about 3% of the genome, which is why some companies will report that a son shares 53% of DNA with his mother and 47% with his father. This is due to the X chromosome and not because, for example, a child inherited two copies of "Gene A" from his mother and none from his father (that never happens). A child gets one complement of chromosomes from each parent, but for a son, the mother's X means that she contributes more physical DNA than the father. 23andme members can look at this thread for more info about DNA statistics -- https://www.23andme.com/you/community/thread/5936/ :)

Debbie Kennett said...

Many thanks Andrea for the interesting link. It's fascinating to see the variations in the percentages between different relations.

GKS said...

I'd like to know what the ancestry percentages mean. I am told I am 92 percent Western European and 8 percent Middle Eastern with Palestinian, Jewish, Mozabite, Bedouin and Druze, in that order. Did the Neolithic migrations of middle easterners to Europe have a lasting effect on the autosomal gene pool or is this necessarily more recent? All of my UK and French lines have been in North America for 11-12 generations.

Debbie Kennett said...

Don't read too much into your Population Finder results. It does not mean that you have ancestors from the Middle East. These are simply your closest matches from the reference populations. There are no British reference populations in the database other than Orcadians (people from the Orkney Islands). My dad also came out with a small percentage matching the Middle East. All his lines have been in England for hundreds of years. You can read more in the Population Finder FAQs here:

http://www.familytreedna.com/faq/answers.aspx?id=22

The tests will eventually provide more meaningful information. The People of the British Isles Project has found marked regional differences in the British population:

http://www.peopleofthebritishisles.org/

Preliminary results were presented by Professor Sir Walter Bodmer at a conference in Cardiff recently. It should eventually be possible to give percentages of ancestry from say Devon and Cornwall. The south-east of England is more homogenous. Bodmer also found a clear divide between North and South Wales.

Anonymous said...

Hi Debbie,

Nick again. In your last response to GKS you say all your father's ancestors have been in England for hundreds of years, but in your original post you say that you can only find 8 of his 32 3X Greats. Depending on how old he is I would suspect is 3X greats were born between say 1780 and 1820.

This leaves a huge possibility for finding non-English (even those possible Mid-Eastern indications) in the ancestry of his other 24 3x greats.

As far as the Orcadian population goes, I've read mixed messages on whether this is generic Orkney or a mixed UK group. Here is what researcher Prof. Doug McDonald said about it 3 May 2011:
"when FTDNA started their PF, the only academic
samples from the British Isles were Orkney (from the HGDP)and Ireland
(from a medical study, but available from Illumina).

However, very recently a good set of real English became available.
The result of using English instead of Orkney is that quite
a bit more people test as 100% English instead of say 90% English
and 10% something more eastern or southern."

Nick

Debbie Kennett said...

Nick,

In the 1700s there was minimal immigration into England. The chances of someone from the Middle East turning up in a sleepy English village at this time and marrying a local English girl must be virtually nil!

I understand that the Orkney samples are indeed from Orkney and are not a mishmash. In the People of the British Isles study they even found distinct variations within the Orkney Islands. The English samples that are now available in the Human Genome Project are from the People of the British Isles Project. The researchers had the whole genomes sequenced of about 100 people in the project, but I don't think they provided the detailed geographical information which would make the results more meaningful. FTDNA do not appear to have added this new reference sample to their database as far as I'm aware.

Anonymous said...

Debbie,

It doesn't have to be limited to someone arriving from the Middle East in the 1700s. It could be 1600s or 1500s or even before.

Look at Benjamin Disraeli and his predecessors. Sephardic Jews who ultimately trace back to Middle East roots who arrived in England circa 1748. Do you think none of the descendants of this family ever married into English families? These connection will and do show up in the DNA.

The Romani (Gypsy) people are ultimately from South Asia (India and surrounds). We know for a fact that they were in Scotland and England quite early and that they certainly, at times, intermarried with Anglo Saxon and other "historically" English people. This will show up in the DNA of descendants to some extent.

I don't know if we ultimately agree, or are working at cross purposes. I just think one needs to take a long view with these DNA results and not limit oneself to thinking that a small Middle Eastern influence in your DNA results is wrong.

Nick

Debbie Kennett said...

I agree that traces of exotic DNA could potentially turn up in people of British ancestry but I just don't think there are sufficient data yet to draw such inferences. It would be a different matter if there were a sizeable reference population from the British Isles on which the percentages were based. I know a lot of people with British ancestry who are getting these small matching amounts with Middle Eastern populations. This is simply a symptom of the lack of adequate reference samples. It doesn't mean that large sections of the British population have Sephardic Jewish ancestry. You shouldn't read the results too literally. Population Finder is in any case still in beta-testing.

Anonymous said...

Firstly, let me specify that I am an Aleut, certified by the Bureau of Indian Affairs, born and raised in Alaska. Being male, I have paid for the Y-DNA 111, mtDNA full sequence, and the autosomal DNA testing called family finder. I wanted a broader understanding of my ancestral history, well beyond what I was able to grasp from my rather extensive family tree.

At this point, I am very unsatisifed with the results, especially where cost for this extensive testing is considered. As far as determing Haplogroup, FTDNA seems just fine, but provides only the most generic information on origin, lacking related ethnicity. For example, my mtDNA was classified as D2A1A, which is specific of Aleuts. However, the report I received (if it can even be called that), just said that the Haplogroup first emerged about 20,000 years ago. No additional information was provided.

I haven't received the Y-DNA 111 results yet, although I already know the Haplogroup. Both my father and grandfather have done the testing, but only at 37 markers. My decision to test at the maximum level was to determine most common ancestor by those people who share my family name, currently engaged in a large testing project at FTDNA. It was already determined that my direct male lineage was originating from the Brigantes (Celt).

Anonymous said...

Now for the family finder, AKA autosomal DNA testing. It was not only expensive, but provided horrible results.

Oddly, my overall classification was European with a sub-continent of Russia at more than 96% with an additional 13+% cited as Middle Eastern. This was all wrong, so I have sent numerous e-mails to FTDNA to get this problem resolved, without success.

Just recently, I learned some valuable information that everyone should know. The most important is that if you are of Native American / Native Alaskan origins, just like me, this test is a waste of time and money and will only upset you. FTDNA has proven that not only are they incapable of identifying us, but have no problems classifying us the wrong way. All you can expect to hear, as a result of your concerns, is "sorry you aren't satisfied". I think this is unacceptable.

Furthermore, out of the 700,000 SNPs in the test, FTDNA has told me that they only match about 1/3rd of them, using a worldwide population genepool of a mere 62 ethnic groups. Their reasoning is that they believe that some people are so genetically similar, it is not worth the effort to attempt and identify them.

The only upside to my testing with FTDNA was the fact that I was able to download the raw genome data. With that, I paid $50 to a company called DNA Tribes, who performs third-party SNP analysis, and has a current reference population of almost 1,300 ethnic peoples, including nearly 900 that are considered "Native" to their particular area. They then provided me a 40 page report, matching my SNPs to the groups in their database. I will honestly say, I was not only suprised at how accurate it was, but how detailed. They were able to identify my father's primarily Northwest European background (not Russia like FTDNA said), in addition they showed my relation to several ethnic Russian groups on my maternal side, including my Aleut heritage, plus identification of Mesoamerican and Native American DNA.

In my humble opinion, based upon price and services offered, you would do far better to get tested at 23andme, which provides Y-DNA, mtDNA, and autosomal DNA results, all for less than the cost of 1 test at Family Tree DNA. Afterwards, if you crave more specific information, do the third-party SNP analysis at DNA Tribes and you will be set.

Not only do I not recommend Family Tree DNA, I do not believe that they are adequately prepared to provide reasonably accurate autosomal results or family finder services, due larger to their small genetic reference population and lack-of matching to SNPs. Not sure how they expect to find ample relations when only 1/3rd of the SNPs are matched, leaving the rest without analysis.

Hope this information helps someone else to avoid the disappointment I felt and aids in making a better selection when it comes to your DNA testing needs, whether that is for genealogy of the determination of close genetic relationship.

Debbie Kennett said...

Dear Anonymous

Thank you for sharing your experiences. You have to remember that we are still in the very early days of DNA testing. The costs of the tests is coming down all the time. The interpretation of test results is very difficult because scientists have not yet published studies to provide the answers that we would like to have. With the Y-DNA and mtDNA tests at Family Tree DNA you will learn most about your origins by joining the relevant haplogroup project. I agree that the reports provided by FTDNA are not very helpful.

You will find a list of Y-DNA haplogroup projects here:

http://www.isogg.org/wiki/Y-DNA_haplogroup_projects

You will find a list of mtDNA haplogroup projects here:

http://www.isogg.org/wiki/MtDNA_haplogroup_projects

All the haplogroup projects are run by volunteer administrators so the support you receive very much depends on how active and knowedgeable the respective admins are.

You might not want to spend any further money but I highly recommend ordering a custom mtDNA report from Dr Ann Turner. She will be able to do a literature search for you and see what studies have been done on your particular subclade. She only charges a very modest fee. You can find further information here:

http://www.isogg.org/wiki/Custom_mitochondrial_DNA_reports

FTDNA has the largest full sequence mtDNA database in the world, and has thousands of samples that have never appeared in the scientific literature and can only be found in the haplogroup projects or as matches in the FTDNA database.

I'm afraid it is not possible to assign any haplogroup to the Celts or any other tribes for reasons which are clearly explained in this article published on the Sense About Science website:

http://www.senseaboutscience.org/resources.php/119/sense-about-genetic-ancestry-testing

Debbie Kennett said...

Dear Anonymous

The cost of the FTDNA tests has come down considerably since you tested there. As you say, Family Finder was previously more expensive than the 23andMe test but they have now dropped the price and the Family Finder test now costs $99, the same price as 23andMe. For Americans the prices are comparable but for everyone else FTDNA now actually works out much cheaper. 23andMe charge very high shipping costs, and also the 23andMe test is only available in a limited number of countries, mostly within Europe and North America. See:

 http://www.isogg.org/wiki/Shipping_DNA_kits.

FTDNA have also reduced the price of their entry-level Y-DNA and mtDNA tests to just $49 each. The advantage of the FTDNA Y-DNA and mtDNA tests is that your results go into a genealogical matching database. Just knowing your haplogroup doesn't really tell you very much.

I agree that the Population Finder component of FTDNA's Family Finder test is not that brilliant, but the same can really be said for all these admixture tests. There are only a limited number of reference populations available, and all the companies are using the same publicly available datasets. 23andMe use the ancestry information from their own customers to inform the results of their Ancestry Composition results. Even then, they still only have a limited number of customers with four grandparents born in the same country because the majority of their customers are in the US. FTDNA's Population Finder test is still in beta-testing and there is supposed to be an improved version in development so you can expect your results to change over time. I don't yet know when the new version will be launched. Population Finder is only telling you which reference populations you have the closest match with. If your population is not included in their dataset then all they can do is match you with the nearest equivalent.

The reality is that none of these admixture tests are really that helpful at present beyond the Continental level. They can distinguish between European, Asian and African but it becomes much harder to distinguish between, for example, European populations. I have the same problem as you with all the companies. I live in the UK and all my ancestry on all my lines as far back as I can trace it is from the British Isles. 23andMe's Standard Estimate only reckons that I am 17.4% "British and Irish". Even when I go up to the Speculative Estimate I am still only 56.7% "British and Irish". This is probably not surprising because most of 23andMe's customers are not Brits but Americans of British and Irish origin. The British component probably gets drowned out by the Irish as there are so many Irish-Americans. According to FTDNA I am 86.23% Western European and 13.7% European, which again doesn't really tell me anything useful at all. According to Ancestry I am 58% Central European, 25% British Isles, 13% Eastern European and 4% unknown. Yet bizarrely a lot of the Americans I am matching at Ancestry, many of whom have ancestors from a range of European countries, are coming out at 80% or 90% "British". I understand that Ancestry are also planning to improve their ethnicity results and we can probably expect changes later this year.

Debbie Kennett said...

Dear Anonymous

I've not tried DNA Tribes but I'm puzzled by the information you provide. DNA Tribes do analyse SNP results from other testing companies but according to their website they have 280 populations in their SNP database not 1300:

http://www.dnatribes.com/snp.html

I think the information you found about their 1300 "ethnic peoples" refers to their autosomal STR database. Autosomal STRs are the markers used by the police for forensic work. The DNA Tribes test only uses 22 or 26 autosomal STRs.

There are a number of free admixture analyses services available. There is a list in the ISOGG Wiki:

http://www.isogg.org/wiki/Admixture_analyses


I suggest you contact Doug McDonald as I'm sure he would find your results of interest.

In my opinion DNA testing is best used for genealogical matching purposes. The 23andMe test is also very interesting if you want the health and trait reports. The ethnicity percentages received from all the testing companies are very vague and often highly speculative. They are more like a genetic horoscope and should not be taken too literally.

Anonymous said...

I wholeheartedly agree with the 'anonymous' comment above in reference to FTDNA.

I recently had my DNA results transferred from 23andMe to FTDNA to locate more relatives and they have me at 99% Orcadian. I know my ancestry and I'm Native, African, Asian, Swedish and Russian. It's amazing to me that FTDNA has me at 99% Orcadian. It's ridiculous. (BTW, all of the other DNA testing companies show my African, Asian and Native, as well as my Swedish and Russian.)

Debbie Kennett said...

FTDNA's Population Finder currently only includes a very limited range of reference populations. It is in the process of being revamped and a new Population Finder should be launched in the next few months.

23andMe's Ancestry Composition is generally considered to be the best on the market but still leaves a lot to be desired. I come out much less "British and Irish" than most Americans. You should take all these admixture results with a very large pinch of salt.

Unknown said...

FYI...coming from a miner's daughter of Cornish Decent..I have family from Barbados, Africa, Mexico, Chile and Portugal, North America Austrailia and New Zealand. Most of us are lost ancestors of family. I cannot wait to find out more!