Thursday, June 23, 2011

What do my Family Finder test results mean?

A slight oversight on my part, I realize that I never did the post that I initially wanted to do for my series on FTDNA's Family Finder test, and that is what information do you get from this test and what does it all mean?

What is the Family Finder test?
Family Finder (FF) is a product from Family Tree DNA that can identify genetic relatives both close and distant up to 5 generations.  The test uses autosomal DNA (Chromosomes 1-22) and the X chromosome to identify regions of your DNA that are "Identical By Descent" (IBD) with other members of the database.  Based on the amount of IBD (as opposed to "Identical By State" (IBS), which is DNA shared between two people that is coincidental) DNA shared between two individuals the algorithms created by FTDNA can determine a relationship range as well as in close genetic relatives, a suggested relationship.


What information do I get from this test?
Personalized website: MyFTDNA
After you order a kit, once it's mailed to you and you swab your cheeks and return the kit to FTDNA, you will get a confirmation email that will identify your Kit # and password.  Then you log into MyFTDNA with your kit # and password and you will be brought to your homepage where once your results come in this will be where everything will be located.

This is what my MyFTDNA homepage currently looks like with my FF results back.

When you first get your kit back you will need to go to where it says "Pending Lab Results" (marked in green). This will tell you what batch # you are in and when your results are expected back.







Once your results are returned the heading (marked in blue) where it says "Family Finder Illumina OmniExpress" will have listed the following tabs (click to take you to that section of this post):


These are the tools that you're going to use to identify your paternal relatives and possibly your biological father.


Matches
Obviously your matches are the most important aspect of this test, but in order to gain the most from them I'm going to explain a few things.

When you first get your results and you click on "Matches" you will be brought to a page that looks like this:


This is a list of your genetic relatives.  However, the default of what is shown is only your close and immediate genetic relatives.  If you do not have any make sure you change the "Relation" bar to include ALL MATCHES.

My close matches take up 2 pages, and all my matches take up 7 pages (I have 12 close matches and 66 total matches).

What you see here is the name of your genetic cousins and the following pieces of information.
Relationship range - this is a range based on FTDNA's algorithms that determines about how close or distant they believe you and this person to be based on the amount of shared DNA.
Suggested relationship - this is again, based on the algorithm devised by FTDNA and is the "most likely" relationship based on DNA. Only close and immediate relatives have a suggested relationship.
Shared cM - cM is "centimorgans" and it's the standard measurement of distance for DNA.  It's actually the recombinant frequency, so if 1cM corresponds to 1 million base pairs, that means that if two markers or SNPs in the genome are 1cM apart, there is a 1% chance of them being separated through crossing-over in a single generation.  So the more shared cM's two people have means they are more likely to be closer related.
Longest block - Again based on cM, while the Shared cM looks at the total amount of shared DNA across the genome, the Longest Block looks at the longest stretch of shared DNA in a row.  Obviously, a long stretch of identical alleles between two individuals suggests that they are very closely related.
Ancestral Surnames - This is a SELF-REPORTED list of surnames that they know are a part of their family (though due to non-parental events - infidelity, secret adoption, donor conception, etc - these are not always accurate).  However,  this will give you an idea as to the potential surnames of your biological father, his possible ethnic groups, and the location(s) that his family may live or have lived.

NOTE ABOUT SUGGESTED RELATIONSHIPS:
It came to my attention after I got my results and started talking to my matches (it never dawned on me before), that the majority of people who order genealogical DNA tests tend to be older (Boomers and up), because they tend to be the ones with the time on their hands to be doing hardcore genealogy research, and have the money to spend on such tests.  However, many donor-conceived adults who are thinking about or have already ordered this test tend to be younger (Boomers and down) than the average genealogist on FTDNA.  So the suggested relationships for those of us who identify as Gen X or Gen Y is going to be skewed.  For example, my #1 match had a relationship range of 1st to 3rd cousin and a suggested relationship of 2nd cousin.  But he is an older gentleman, likely in my grandmother's generation.  So while FTDNA suggests we are 2nd cousins, it is unable to take into account that there is a 2 generation gap between us, so we are probably "closer" cousins than that.  Turns out we are actually 1st cousins 2x removed.  But the algorithms cannot differentiate that.  So just FYI, if your matches are significantly older than you are, it's possible that the most recent common ancestor (MRCA) is not, for example suggested 2nd cousins, your g-grandparents.  In my example, our MRCA is his grandfather, but my gg-grandfather, thus 1st cousins 2x removed.


Chromosome Browser
This is a tool where you can select up to 5 of your matches and see where on the genome you have shared DNA.  This is especially helpful if you want to see perhaps if one of your matches might also be related to another one of your matches.  Though remember, if you all share the same segment of DNA with two different matches it's highly likely that all 3 of you are related, but if there is no shared DNA between the 3 of you it does not suggest that you are NOT all related.


Known Relationships
The KR tab is really sort of a crappy tool.  It's something that honestly should be available for all of your matches, but has been relegated to only those with whom you supposedly can identify with a paper trail. What it does is that for matches that you identify as a "known relationship", you are then able to see what matches you have in common.  Ironically, this feature is available on the free GEDMatch program (but the downfall is that it's much less accurate).


Population Finder

The PF tool is still in beta, but for those of us who are donor-conceived, it might be the best chance we have at determining a general locale that our biological father's ancestry might lie.

As you can see at the left, this is my PF results.  They are as accurate as they could be with the very limited reference samples that FTDNA uses.  Here is a list of their sample set.  As you can see there are only a handful of different ethnic groups for each large population.


The European population set is:
Northern European (Finnish and Russian)
Southeast European (Romanian)
Southern European (Italian, Sardinian, and Tuscan)
Western European (Basque, French, Orcadian, Spanish)

Obviously there are some really huge ethnic groups missing just in Western Europe, not to mention the rest of the continent and all the other population groups!!  What this means is that your PF results are going to take your data and match it to the closest continent(s) and population(s) as possible, but they do not mean that those are your ethnic groups.  Obviously I have no French ancestry, yet PF says I'm 72.4% Western European with French and Orcadian (the Orkney Islands are a part of Scotland).  I'm also a quarter Armenian, yet my Middle Eastern results (27.6%) came up as Iranian (right next door), Adygei (a Caucasus people near Russia - Armenians are also a part of the Caucasus population), and Druze (an eclectic religious sect thats beliefs are a combination of all the Abrahamic religions (Judaism, Christianity, and Islam) found throughout the Levant, mostly Syria, Lebanon, and Israel - not really sure about this one, but same region thus probably similar DNA).

Bottom line, the ethnic groups that I belong to are not population groups that are used in this sample set, so they picked ones that most closely resemble my true ethnic heritage.

My suggestion if you want more accurate ethnic breakdowns, after you get your results email your raw data to Dr. McDonald.  His population sets are much larger and more inclusive and he will give you a much more definite idea as to your ethnic makeup.  Dr. McDonald's analysis of my DNA was pretty much dead on.

Here's what he said about my ancestry:

Most likely fit is 70.5% (± 3.8%) Europe (all Western Europe) and 29.5% (± 3.8%) Mideast (various subcontinents).
The following are possible population sets and their fractions, most likely at the top:
English= 0.677 Armenian= 0.323 
French= 0.768  Iranian= 0.232 
French= 0.726 Georgian= 0.274 
French= 0.696   Adygei= 0.304 
English= 0.660 Georgian= 0.340 
These are in fact exactly what I would expect given what you say. 
Czech comes out an (English or French)/(Armenian or Georgian) mix or an English/Romanian. Thus its no surprise that English-Armenian fits best. 
The Mideast on the chromosomes is very weak indeed, very close to Europe, as befits just a little Armenian, which itself is rather European.
 I'm going to do a post in the near future just on McDonald analysis, so stay tuned for more information about what Principle Component Analysis is and what information you get from Dr. McDonald.


Download Raw Data
Fairly self-explanatory.  You can download your raw data and then use it in some of the programs and tools I talked about last week.

***************************
I will continue this series with more information and advice on how to use your FF results to best of your advantage, including how to best contact matches, how to gain information from matches that don't respond, and more!

4 comments:

Richard McMurtry said...

HI Lindsay,
I very much appreciate the depth of thought you put into your postings.

I am struggling with the power and the impossibility of the family finder test. I was led to believe that this test could tell if someone was a 3rd or a 4th or a 5th cousin. So I invested huge amoungs of time and raised $2400 from genealogists all interested in some unasnwered questions about their respective ancestors born between 1785 and 1822 all converging upon Rutherford County Tennessee.

What I am finding is that I am getting large numbers of matches for which despite trails back 5 or 6 generations can find no paper matches even though the results say 3rd to 5th cousins.

So I thought I'd get two samples from each of the individuals born in the 1780s to 1820s and see if a consistent prediction of distance of relationship emerged or an average predicted relationship emerged that could be correlated against those whose relationship is known to extrapolate to the relationship between those who are not known.

I am still waiting for results from the second batch.

But I just got an email today from someone who says one of their predicted matches turned out to be a 7th cousin! So even if the predicted match (not stated by my correspondent) was a 5th to distant cousin, the prediction was 2 generations off from the real value!

I'm not so sure that FTDNA has been very straightforward with telling people about the limitations and difficulties of getting useful conclusions from this test. A lot of us, inspired by the tremendously powerful and useful results of the Y-DNA test, read the claims of being able to trace across female lines and lept to the opportunity to resolve thorny long standing genealogical puzzles.

It seems like what we are going to get from this is the capacity to tell when people are not closely related (dispelling family tradition myths), but not be able to tell the degree of relatinoship that caused us to seek the help of this test in the first place. (I hope this pessimism does not cause me to be labelled a troll!)

I think what is needed is a realistic assessmnent of the error bars on the estimates based on statistical analysis of the difference between predictions of the algorithm and the actual known relationship between donors. If the estimate is 2nd to 4th, we need to know if 20% of the time this could be 5th cousin or is it 5% and we need to know if 10% of the time it could be a 6th cousin.

I'm beginning to think that actual relationship outside the predictd relationship is actually more like a 30% chance.

It would be wonderful if you could be a catalyst for getting FTDNA to publish this analysis even though it might not help their marketing if people realized the limitations of the test and the difficulty or interpeting the results.

Lindsay said...

Richard,

I'm not 100% sure what you think I can do about this. I am in no way affiliated with FTDNA, other than as a project admin. I simply have a background in genetics and try to educate on these tests.

Though, perhaps you've been misled but there is a reason they call them speculative matches as 4th-distant or 5th-distant. They are unable to determine with any accuracy how two individuals are related at this level, other than they are related distantly.

The issue I believe you are running into is that there is that you're dealing with more than likely an endogamous population, and therefore the amount of intermarriage among cousins makes more distant cousins appear to be closer.

They have acknowledged this in the Ashkenazi population, as it's easy to identify. But southern colonials also have the same situation (though no as far back as Ashkenazi, where their founder populations go back hundreds of years in Europe versus Colonials we're only dealing with 1600-1700s). Unfortunately FTDNA is unable to change their algorithms for southern colonials to remedy this issue. So you must go into the testing understanding that if this is your ancestry that this is a situation that may make things more difficult.

I discovered my bio-father was/is southern colonial, and while the plus side is that I have TONS of matches, the downfall is many of the matches have no clue how they are related to each other.

As for percentages, again if it's a 4th or 5th-distant suggested relationship they have already explained that these are speculative and not exact. For relationship ranges of 3rd-5th, in dealing with an endogamous population again, you need to understand that their algorithms currently do not take that into account except for those with obvious Ashkenazi ancestry. So any suggested relationship or relationship range needs to be taken with a grain of salt.

DNA does not lie, but the technology is still cutting edge and we currently do not have the abilities to read discriminate against two individuals from an exogamous population that share x amount of DNA and two individuals from an endogamous population that share the same amount of DNA. While the exogamous population pair might be 3rd cousins, the endogamous population pair might be 5th, or 7th cousins multiple/many times.

Paul said...

Very interesting. I "just" got my Family Finder results. I looked at the Population Finder,which said I was "100% Orcandian". My actual genetic background is E-V13 (Greek) and English/Norse on my paternal side and Separdic/Ashkenazi on my maternal side. Any thoughts? Looks like FTDNA messed up to me! Thanks

Lindsay said...

Paul,

I have no idea, but yes it does seem like they messed up! But what I'm not sure.

What about your matches? Are they representative of your actual ancestry? Do you have a notation above your results about Ashkenazi endogamous populations? If you don't have that and none of your matches are reflective of Greek and Jewish ancestry I'd say there are two possible explanations: 1) contamination, or 2) someone at FTDNA accidentally tested the wrong vial.

You may not have many greek matches (some ethnic groups are less interested in DNA testing), but if you're maternal line is jewish you would most definitely have jewish matches, for sure.

I would email FTDNA just to see what they say. Their customer service people are fantastic and sometimes you might even get a response from Bennett Greenspan himself!