Dear Karl Broman,
I am dealing with a transgressive phenotype where a new phenotype appeared in RIAILs that is not in both parents used to establish the lines.
Recombinant inbred advanced intercrosses lines (RIAILs) were produced in a microscopic free-living nematode. Nematodes live in a population of nematodes containing males, females and hermaphrodites. The main feature of this nematode is the production of a low number of males, where males are scarce < 2% of the population. When males are crossed with a female the outcross population would have a low percentage of males < 18%. In my research, I am investigating the reason why the production of males from an outcross falls short from the Mendelian inheritance, where the ratio of males to females is 1:1.
However, this ratio can be drastically changed by mixing genomes of two nematode stains (with scarce number of males) using RIAILs. At the end of the inbreeding, more than 50% of the lines had a high male population, a phenotype that is not observed in any of the parents. Males from those lines (with high male population) also produced a high number of males when crossed with a wild type female, again a phenotype that is not in the parent.
I am using rqtl to identify genomic region associated with the production of high number of males in the RIAILs lines. In addition to the markers set the data also includes phenotype set of two columns; percentage of males produced and covariate (whether the line is HM or LM). I have identified a number of qtls from a single scan and multiple qtl scan. Two way scan didn't work because of the large number of markers. I am not sure which qtls to consider in my final result.
1. Single scan identified
Q1, Q2 and Q3, with qtl peaks.
2. Multiple qtl scan
using function addint() I assessed the interaction between those qtls. Then refined the qtl location using refineqtl (). So now Shall I consider those peaks and disregard peaks from the single scan?
3. I scanned for QTLs in lines with a high number of males (HM) and lines with a low number of males (LM) separately.
I identified a QTL in HM only (Q4).
4. I scanned for a qtl with the covariate.
- As additive and interactive covariant separately, found a QTL in a similar location to (Q4).
- Then scanned for QTL phenotype interaction by finding the difference between the LOD score and the permutation test with the phenotype as an interactive covariate and the LOD score with the phenotype as an additive covariate. Found a new qtl (Q5).
5. Scanned for interaction between Q1, Q2, Q3 and the covariant.
addint(A.fre_genop, qtl=qtl, method="hk", covar = Covariate, formula = y~Q1*Q2*Q2*Q3+Phenotype)
6. Scanned for interaction between Q1, Q2, Q3, Q4, Q5 with and without the covariate.
- addint(A.fre_genop, qtl=qtl, method="hk")
- addint(A.fre_genop, qtl=qtl, method="hk", covar = Covariate, formula = y~Q1*Q2*Q2*Q3*Q4*Q5+Phenotype)
7. Refined the QTLs estimated from a single scan for Q1, Q2, Q3, Q4, Q5.
New peaks for Q1, Q2 and Q3. Wich peaks to consider, peaks from step 2 or those peaks from 7?
Finally, after running addqtl() found an LOD peak in a new location.
I am not sure which peaks to consider after running all of the above-mentioned scans? especially I ran multiple qtl scan twice and got different results depending on the number of qtls.
I have an R markdown file I can send to you separately through email if you would like to have a look.