PDA

View Full Version : Imaginary gaps in parent-child chromosome comparison



geebee
04-29-2017, 06:00 AM
In reality, the only "segments" a parent shares with a child should be whole chromosomes, so the number should always be exactly 23.

However, I'm also well aware that sometimes -- for various reasons -- a "gap" may be reported in the sharing on a chromosome, which will result in a higher number of segments being shared.

Between my father and two of his offspring, chromosome 9 shows as two shared segments.
They are exactly the same for both. One is from 1-135032890, and the other is from 136275141-141088641. The puzzling thing, however, is that matching continues on all SNPs through the entire "gap" -- with just one exception. At position 135139064 (rs11787894), both offspring are reported to have GG, while our father is reported to have TT. The remaining four offspring all have GT.

A single SNP should not by itself cause a gap to be reported, and from the data below you can see that it did not do so in the case of my other siblings. All of them had discordances with our father on chromosome 9, but outside of the supposed "gap". None of their discordances caused any gap to be shown.


Bernie (v4)
9 rs7863242 90780784 TT CC


Curt (v2/v3)
9 rs771667 1565742 AA GG
9 rs2782434 27095072 CC TT
9 rs7863242 90780784 TT CC
9 rs17810151 100966840 GG TT
9 rs10816719 111537522 TT CC


Gary (v2/v3)
9 rs771667 1565742 AA GG
9 rs2782434 27095072 CC TT
9 rs7863242 90780784 TT CC
9 rs17810151 100966840 GG TT
9 rs10816719 111537522 TT CC
9 rs11787894 135139064 TT GG -- within the "gap"


Kim
9 rs771667 1565742 AA GG
9 rs2782434 27095072 CC TT
9 rs7863242 90780784 TT CC
9 rs17810151 100966840 GG TT
9 rs10816719 111537522 TT CC


Luci
9 rs771667 1565742 AA GG
9 rs215189 135671064 AA GG
9 rs4842226 138078679 CC TT


CJ (v4)
9 rs11787894 135139064 TT GG -- within the "gap"

Only CJ and I showed a gap on chromosome 9 in the comparison with our father.

Here are my daughter's results, as well:


Kathryn (v2/v3)
9 rs10973917 3866465 AA GG
9 rs641274 15357727 CC AA
9 rs10965805 23364154 TT GG
9 rs17810151 100966840 TT GG
9 rs10901242 136077053 CC TT

And as rs11787894, Kathryn has GG. As a final note, it's pretty clear that all of Kathryn's paternal chromosome 9 is a match for my paternal chromosome 9 -- or in other words, Kathryn received no DNA from my mother on this chromosome. Of course, this means Kathryn also shares this apparent gap. Since my father is her grandfather, a gap would be possible.

However, if she actually didn't share with my father in this region, then she'd share with my mother in this region. In the region in question, I do match some of my siblings on my maternal copy of chromosome 9 -- but my daughter does not.

So, anyone have any clues as to why 23andMe is showing a gap on in these parent-child comparisons on chromosome 9?

geebee
05-03-2017, 02:46 AM
I've now received a reply from 23andMe's imaginary support desk. Basically, this is their explanation:


It is not uncommon to see slight breaks in a chromosome a child shares with their parent. This is likely due to no calls in the parents' results.

Now, the flaw in this explanation is that there's one parent here for six offspring. Two of us show the break, but the other four do not. It seems as if, were the parent's no calls a factor, if should affect either all six of us, or none of us. Of course, it could be a combination of parent and child no calls that's the problem.

So I used David Pike's "Inspect a Shared DNA Segment in Two Raw Data Files" (http://www.math.mun.ca/~dapike/FF23utils/strand.php) to take a closer look at the region in question. What I found is that the comparison between my father and my sister Kim -- one of the ones with no break in chromosome 9 -- actually yielded more no calls than between my father and me. In fact, a lot more: 98 between my father and Kim, versus 55 between my father and me.

Furthermore, the comparison with the least no calls between them involved my father and my sister CJ. That was only 13 no calls. However, my father's results are from v2_v3, where CJ's results are from v4. So much, perhaps, for the reduced number of SNPs not affecting comparisons!