Dual peak for binary trait mapping

[Maggie Johansen]

I've spent some time creating a linkage map and performing binary trait mapping for the colour locus in a species of land snail (either yellow or not yellow). I've had some very promising results, but I'm curious about one thing. For the genome scan I two peaks on chromosome 11 (figure attached), and during thew find_peaks step when looking for multiple peaks I get two intervals that correspond to the two peaks from the genome scan:

lodindex    lodcolumn    chr    pos    lod        ci_low    ci_hi
1        yellow        11    7.429    12.05108    7.429    7.429
1        yellow    11    31.636    10.04166    30.294    32.307

However, the first peak does not have a corresponding (or at least not very strong) corresponding trait effect (figure attached). Furthermore, it doesn't correspond to the expected allele pattern for this genotype where we would expect non-yellow individuals to have a AA genotype (figure attached). From looking at all the results and independent evidence, I'm very confident that the true location of this locus is within the second interval (even though the LOD score is lower). 

My question is, what could be the reason for the multiple peaks on this chromosome? 

Thanks, 

Maggie 

[Dan Gatti]

I’m not entirely sure. Could the linkage map be wrong? Or is there correlation between the marker genotypes at the two peaks? That might lead to a “ghost peak”, but it’s strange that it’s LOD is higher than the LOD for the causal locus. Or there may be a second locus that is causal and you’re discovering it right now, but you’d have to rule out the first two questions before going down that path.

--
You received this message because you are subscribed to the Google Groups "R/qtl2 discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rqtl2-disc+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl2-disc/7cc1cf20-f670-4101-9ea5-2b6ffbde01d7n%40googlegroups.com.

---

The information in this email, including attachments, may be confidential and is intended solely for the addressee(s). If you believe you received this email by mistake, please notify the sender by return email as soon as possible.

[Maggie Johansen]

Hi Dan, 

thanks for the quick response.

" Could the linkage map be wrong?"

Yes, I think it could, however I find it unlikely. One of the resons I'm happy with the second interval is the presence of contigs (draft reference genome isn't to chromosome level) that we know from previous analysis are probably linked to this particular gene. 

"Or is there correlation between the marker genotypes at the two peaks?"

No, not really. At the second peak all positions (bar one) shows BB associated with yellow and AA not associated with yellow. Which is what we expect to see, yellow being recessive to not yellow. While the singular location at the first peak don't follow this pattern (though BB is mostly associated with yellow, apart from three individuals). 

"Or there may be a second locus that is causal and you’re discovering it right now"

Yes, there could be, considering we know very little about the genes of this species. However, the locus we're looking for is in a tightly linked supergene, but I suppose there could be a modifying locus outside of that. I've been looking into trying to do a QTL as well which accounts for variation within colour morphs but have not been able to to come to a good solution using the spectrophotometer readings we have. 

Thanks for your help, trying to resolve this issue is last step before the results are publishable. 

[Dan Gatti]

Hmm. I’m not sure then. The LOD plot looks like there are a set of markers in LD around the first peak. So it’s not one marker being off.  What does the effect plot look like at the first locus?

 

When I look at your LOD/founder effects plot, the effect size a the first locus is small, but you still get a large LOD. Can you see what the sample size is in each genotype class at the first peak? Maybe you have unbalanced allele groups sizes and that’s throwing things off?

 

Dan

To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl2-disc/3b94fa51-e3df-4a1d-958d-0b020a39ecb5n%40googlegroups.com.

[Maggie Johansen]

The LOD plot looks like there are a set of markers in LD around the first peak.

It does, I do however find it strange that it's only giving a single position rather than an interval with both high and low as 7.429. 

What does the effect plot look like at the first locus?

Sorry, not sure what plot you mean? 

Can you see what the sample size is in each genotype class at the first peak?

The sample sizes are: AA = 22, AB = 20, BB = 34. Could that be enough to throw it off? 

There is also several markers at this postion, 26 markers from 13 unique contigs. 

[Karl Broman]

Between the left peak and the valley, some set of individuals had a recombination event. And then between the valley and the right peak, there were further recombinations. I think the key question is: were there many double recombinants, You want to understand the relationship among the phenotype, the marker at the left peak, the marker at the valley, and the marker at the right peak.

karl

[Dan Gatti]

The sample sizes look OK. And I saw the you had the allele effects in your first post.

 

Karl had a suggestion about looking at the recombinants in detail. Maybe that’s that way to go.

To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl2-disc/d0181d35-96d2-4948-9adc-20bd2a79d531n%40googlegroups.com.

[Maggie Johansen]

Thanks both, I really appreciate the help/insights. 

I've done some looking into the recombinants, and several individuals have recombination events between the two peaks, and as far as I can tell only two individuals have two recombinations in the interval between the two peaks.