What are the sporadic red/green lines in GEDmatch chromosome pictures?

Question

I just compared my Mom and her brother in GEDmatch and realized I maybe don't understand the graphical depictions of chromosomes.  The blue segments in the lower half of a chromosome depiction are clearly the shared segments. 

Edit: I just found a blog saying that in the upper half, green vertical lines are positions where the pair match on both chromosomes, yellow are where they "half-match" on only one chromosome, and red is where there is no match.  

Question 1: Why do I see a few  segments marked as matching segments with a couple of red vertical lines in them?  Does GEDmathch "excuse" a few non-matching SNPs in the middle of a matching segment?  Perhaps at SNPs known to have high mutation rates?

Question 2: why are there "sporadic" green and red lines separated from others?  It seems to me there are too many of these to have had crossover events at every single position where there is a color switch?  If some of these are really "false positives" for crossover, how does one identify the actual crossover points?

Answer 1

Okay, I'm answering my own question: but please comment if I'm wrong or missing something.

Lets imagine looking at the copies of chromosome 11 received by my Mom and by her brother from their father.  In sperm formation, we expect roughly one crossover in each chromosome, and these positions would correspond to color changes in the corresponding chromosome picture, so we'd expect two color changes comparing my Mom and her brother on chromosome 11, corresponding to the one crossover that Dad gave them.  Similarly, we'd expect a few more color changes due to crossovers received from their mom.

Since there aren't many crossovers, we'd expect a good portion of half match where, say, both match on the chromosome from their Dad but not from their Mom.  But: sometimes by chance the two different genomes they got from Mom will just happen to match by chance, (especially since there are only four basic DNA molecules).  These would produce smatterings of green among the sea of yellow.  Similarly, we would expect smatterings of yellow among seas of red, where there is no match by descent but by chance both my Mom and her brother match on one chromosome.  On the other hand, we wouldn't expect smatterings of color among green bars with this explanation.

And so looking again, that's what I see.  Solid green bars that aren't broken up like the yellow or red bars.  And red bars are broken up by more yellow then yellow bars are broken up by green because there are two ways for a red position to happen to be a half match.

Now the only trouble is why there is an occasional yellow blip in the green bars, and I guess that's because of a mutation.

Answer 2

EDIT: didn't refresh page before I answered, so didn't see Barry's answer first.

Good question, Barry!

Regarding your question 2, can you describe a bit more? I.e., green and red lines separated from others -> by what, or in what region? Probably a limit of my understanding of your question.

Regarding your question 1, I've seen the same thing, and I suspect it is due to the "mismatch-bunching limit." It would appear that a mismatch (red line) is allowed "within" certain limits. I'm guessing that allows for errors in DNA processing and/or the matching process (false "negatives"). Dunno, just sayin'

From https://www.gedmatch.com/gedwiki/index.php?title=One-to-one

Size (in SNPs) of Mismatch Evaluation window

If two conflicting SNPs are found to be closer to each other than this number of SNPs, then the matching segment will end.

SNP count minimum threshold

Size (in SNPs) of Mismatch-Bunching limit

Determines how close single mismatching SNPs may occur without terminating a matching segment.

1/2 of mismatch evaluation window

Comments

Thanks Bruce.  I hadn't considered mess-ups in processing.  Now my ever-curious brain can't stop wondering which is more common:  these or mutations?

---

If you're REALLY curious, take another DNA test from the same company, and compare to yourself one-to-one, chromosome graphics. Don't know whether there are minor differences in the actual DNA in 2 samples from the same person, but -assuming your own DNA doesn't vary- a visual comparison would provide you with an idea of the combined errors of the processing + matching processes. ;-)

---

Mishaps wins over mutations, at least for me.   Use the DNA file diagnostic at GEDMatch.  It says for my ftDNA result that 3.9% of the SNPs had no-calls, which gave a warning message in the diagnostic.  My mother's was 2.2% no-calls  which it implied was OK.

Now, no-calls seem to be ignored for comparison purposes rather than generating a mismatch, but when there are so many ambiguous results you might expect a few outright errors.

---

I guess each genome has 100-200 mutations.  On the other hand, FTDNA, for instance, tests about 700000 SNPs, so I guess you're looking at 15,000 or more no-calls.  So even if the error-rate would have to be two orders of magnitude lower for mutations to compete.  I'm inclined to agree with you.  And thank you for the reminder that I should run the diagnostic utility on my kits!