46 matches that don't match either parent

[ausgenjourneys]

Hi,
Both my parents have now taken the Family Finder test. I've realised I have 46 Family Finder matches that are not matches for either of my parents. I expected (from reading the Family Finder FAQs) that I might have a few different matches...but 46? 3 of these 46 are matches for a maternal uncle.
I'd appreciate any thoughts from others who have more experience with autosomal dna testing of family groups. Thanks.

[T E Peterman]

They are probably distantly related to you on both your father's & your mother's side of the family. The cm count for each isn't great enough to be reported for them for each of your parents.

But for you, since you are related on both sides, happen to have enough centimorgans for them to be reported.

Most people with colonial American ancestry have a few in this category. But 46 in this category is a lot.

This is probably the most likely explanation. But you might want to send a message to Family Tree DNA to determine what the real reason is.

Timothy Peterman

[travers]

I don't know about the explanation that you are related to them on both sides of your family. Usually the most likely scenario is that these are matches that fall right at the border of being a match or not. A few SNPs from the other parent can trigger a match in the child that neither parent shows as matching. I wouldn't contact FTDNA as suggested unless these matches are sharing a significant amount as this is in the FAQ.

[ausgenjourneys]

Thanks Timothy Peterman for your reply. I thought perhaps it might be because both my parents had some Scots Irish ancestors from neighbouring counties in Northern Ireland. Looking at the surnames and locations listed for a number of the 46 matches Scots Irish ancestry seems to be the common factor. In fact, some of the 46 list my mother's Scots Irish ancestor's surname! One even with a location in Scotland...could there possibly be a link that far back ?? I was surprised that they did not match my mother. I will contact FTDNA just in case there is something else wrong.
Thanks again.

[ausgenjourneys]

I don't know about the explanation that you are related to them on both sides of your family. Usually the most likely scenario is that these are matches that fall right at the border of being a match or not. A few SNPs from the other parent can trigger a match in the child that neither parent shows as matching.

Thanks Travers,
They are all in the 4th/5th to remote category, the largest cM shared is 12.27
, smallest 7.69. I was mostly interested because some actually listed one of my known surnames from my mothers family, and was surprised that they didn't match her.

I wouldn't contact FTDNA as suggested unless these matches are sharing a significant amount as this is in the FAQ.

You are right, I guess I will leave it for now.

Thanks Timothy and Travers for your replies.

[vinnie]

I have one similar case for my own results, and I would very much like the ability to compare this person with both of my parents. Hopefully he'll upload to Gedmatch so I can do that.

[JohnG]

Distant matches

I have identified common ancestors for 10th cousins who have minimal matching DNA. They have been from early New England or New Amsterdam and with lots of reinforcement by intermarriage at the time. I have suspicions of similar distant matches in Northern Ireland and in Norway. Before the DNA tests I had no idea I had any Norwegian ancestry, but things are pointing that way.

[T E Peterman]

Another thing to check is the number of centimorgans you share with each parent. If there are a few minor gaps between your results & each of their results, it probably represents a few misread SNPs; maybe just 3 or 4, scattered across several chromosomes (Illumina has about a 99.95% or so accuracy rate; Matt could give you the exact percent). The impact of these misread SNPs can be huge when dealing with speculative matches (ie, those hovering near the 20 cm minimum threshhold), especially if the chromsome involved for the match is the same one with the misread.

Timothy Peterman

[sjadelson]

In certain circumstances, this is pretty common. I've tested my dad, thinking it would help determine which side of the family a match came from. You'd think that anyone from his side of the family would have more shared DNA with him than with me, and that it would be (roughly) 100% more.

What I found was that even with nearly 1100 common matches between us, only 32 had even 50% more DNA, and only 9 have 75% more. Two known, close cousins (2nd and 2nd 1x to my father), are between 57% and 64% more.

Point being, there's nothing "wrong" here, you're just getting common ancestry spoofing the algorithm into a closer match than is really there.

[T E Peterman]

Sjadelson's point is right on the mark. I suspect that, especially in cases of colonial American ancestry, many of the more distant matches are in fact multiple cousins, sharing not just two lines, but possibly 3, 4 or 5 from back in the 1700s or 1600s.

Two participants that share say 5 lines of ancestry are 5 times more likely to see each other as matches than those that share just one line of ancestry. Estimated kinships for those with multiple shared lines are always closer than they really are.

I suspect that 13 generation gedcoms that were reasonably complete would reveal this quite well.

Timothy Peterman

[mkdexter]

this is the reason you have 46 matches that don't match your parent(s). Copied from one of my older posts.

First.. This is not a problem with the relationship. This is very normal for the Family Finder test as it relates to 1/2 IBD testing and blocking of SNPs in the matching process. It means these people are riding the tolerances of the test setup to eliminate false matches and noise. Unfortunately this is a consequence of having to set limits yet those limits are absolutely necessary in order to provide accuracy.

The Family Finder software clusters the SNPs into sets that are about 50 to 100 SNPs long. See this link in the FAQ: http://www.familytreedna.com/faq/answers.aspx?id=17#798

Sometimes one person has homozygous readings and the other person has heterozygous readings (CC to AC for example). This allows a match even if by the opposite relation.

I have an 8th cousin who matches me and not my mother. The common ancestor is listed on our paper trails. The REAL match is less than the threshold of Family Finder however my mother does match this person when compared behind the scenes.

This is what happens. Even though there are limits of consecutive SNPs in the Family Finder matching process, they are blocked off by 50 at a time. If I match my 8th cousin by 1000 SNPs and my mother matches by 995 SNPs, the FF tests sees that as a 950 SNP match and the threshold is not met. The cM of 50 SNPs is larger than the cM of 5 SNPs.

Why would I match by an extra 5 SNPs, because my 8th cousin has 5 SNP matches to my father's side too. That's not enough to say there is a relation to my father but that is enough to put my test 50 SNPs over my mother's compared to the 8th cousin. 995 SNPs = 950 in the test and 1000 = 1000 in the test.. And that's why you see a difference.

Now.. is 10cM really 3 cM over the tolerance in your case? That really depends on the location of the match. Some chromosome positions have high cM levels and some have just a few. It is possible a difference of 50 SNPs in some locations is a 3 or 4 cM difference. It all depends on the locations and chromosome.

Remember the key to all matches is the reference SNP, the longest block. The other matching segments have more of a supporting role. If you don't match by the longest block as it is determined by the FF test, you are not going to have a match.

Matt.

[Ann Turner]

Hi,
Both my parents have now taken the Family Finder test. I've realised I have 46 Family Finder matches that are not matches for either of my parents. I expected (from reading the Family Finder FAQs) that I might have a few different matches...but 46? 3 of these 46 are matches for a maternal uncle.
I'd appreciate any thoughts from others who have more experience with autosomal dna testing of family groups. Thanks.

People have suggested various scenarios to explain this discrepancy. One that hasn't been mentioned yet is that these are "pseudo-segments." They're not actually one long consecutive run of half-identical SNPs inherited from one parent, but a patchwork stitched together from a few paternal SNPs alternating with maternal SNPs that make it appear to be one long run.

I described this in my 'Satiable Curiosity column in JoGG "Identity Crisis: Identical by State or Identical by Descent?"

http://www.jogg.info/72/files/Turner.htm

This is a consequence of looking at genotypes, where you don't know which allele came from your father and which from your mother. Since you have a father/mother/child, you can phase the data (sort out which allele is which) into haplotypes. David Pike has a utility to do this:

http://www.math.mun.ca/~dapike/FF23utils/trio-phase.php

If your match is willing to share, you could then compare the child's paternal haplotype and maternal haplotype separately.

[ausgenjourneys]

Thanks everyone for your help and ideas. I will have to try to phase my data.

Thanks again.

[Jackd]

They are probably distantly related to you on both your father's & your mother's side of the family. The cm count for each isn't great enough to be reported for them for each of your parents.

But for you, since you are related on both sides, happen to have enough centimorgans for them to be reported.

Most people with colonial American ancestry have a few in this category. But 46 in this category is a lot.

This is probably the most likely explanation. But you might want to send a message to Family Tree DNA to determine what the real reason is.

Timothy Peterman

I observe what I call overlap. I'll try to explain what I mean.

For example, I'll put in the name of one of my matches in the "in common" feature. Or what is called a triangulate on gedmatch. This match will have 10 people listed. What happens sometimes is I'll have 4 out of 10 who I predict are related to my mother and 6 out of 10 who I predict are related to my father. This is just a handful of matches. Nevertheless, it is confusing.

According to gedmatch, my parents share 0.33 centimorgans witch amounts to 12th cousins. So that is vitually no relation at all.