Saturday, February 11, 2012

Discordant SNPs


I keep a database of which segments are shared with each DNA cousin here for all my profiles.

I use this database to determine which branch people are connected to.  I have noticed that most if not all of my predicted 4th cousins, with 2 segments, arent likely 4th cousins, as they have a segment from my mothers line and one from my fathers line.

Since receiving my fathers results, I have noticed some alarming false matches (ie where I can see they dont match my father, but do match me and I also match a maternal relative in that segment who doesnt match the cousin).

Looking at one particular predicted 4th cousin, (2 segments, 14 cM total)  there is one segment on chr 2 which matches me in a 5.5 cM and also my maternal grandfather.  Peachy, no problem there, clearly this relative would be thru my mother and then thru my grandfather.  This cousin also matches my maternal half brother in the same segment . Again, peachy, exactly what I would expect since I also match my half brother there and we both match my maternal grandfather there.

BUT, in the 2nd segment, the larger one, its on chr 7.  apparently 14M-19M (8 cM).  Now on chr 7 I match my maternal half-sister 8M-44M. Neither my sister or myself match our maternal grandfather in this segment, so the logic is that that segment is from my mothers mother.   My half sister doesnt match the cousin  on chr 7 (or anywhere else).  Now, until recently this was also peachy, as I then assumed that the chr 7 segment was from my father.  Trouble is, now I have my fathers results, and he DOESNT match this cousin at all.   So this would be I guess an 8 cM false match?

But it got me thinking, soI ran David Pikes tool for discordant SNPS between myself and my father.  How many is average for parent/child?  I have 164, but 26 of those dont count because they are on the X or mtDNA, which is expected to be discordant.  So we really have about 138 of them.

And sure enough one of them is on chr 7 at 16607786.  This would be within that "false match".   segment.   I have looked there and see:

CT maternal half bro
CC me
CT maternal grandfather
TT maternal grandfathers full sister
TT father
CC maternal half sis

So one of my Cs would be from my mother....and my father is either a genotyping error and should be CT or....I could have the genotyping error and really be CT or....?   Ive been reading some on copy number variations but dont fully understand them yet.

This is the only discordant SNP within this particular false matching segment.  Would ONE SNP different cause my father to not match if all the others did?  I will be contacting this cousin to beg him to upload to gedmatch so I might compare there, but does anyone have thoughts on this?

I have fantasies about phasing all my data and reconstructing a DNA profile for my mother, but these little glitches really throw me for a loop!