high LOD in empty spaces

[Vianey B.E]

Dear Karl

I am working with a 4-way cross. I have got some QTL using scanone (em method) but when I take a closer look I notice that some QTLs have high LOD values in areas without any marker. I would like to understand what is happening here.

Thanks

[Karl Broman]

Take a look at chapter 4 of the R/qtl book.

https://rqtl.org/book/rqtlbook_ch04.pdf

karl

[Vianey B.E]

Thanks, Karl. I understand better what is happening. I checked manually the data and used the effectplot function to confirm if the involved markers have similar behaviour (phenotype vs genotype) and it was as the book described.

I was wondering if there is a limitation in the size of the gaps of missing information and what precautions should be taken when this kind of QTL appear?  Currently, I am using a step size of 0.5 cM

--
You received this message because you are subscribed to a topic in the Google Groups "R/qtl discussion" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/rqtl-disc/Z06pIZxJmtU/unsubscribe.
To unsubscribe from this group and all its topics, send an email to rqtl-disc+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl-disc/34e5f6a6-33ea-4bea-ba24-a14b13f301b9n%40googlegroups.com.

[Karl Broman]

the marker gaps you have don’t look big, but you haven’t said what sort of cross this is.

the step argument just concerns where calculations are done. step=0 means just do calculations at the markers, but otherwise that choice shouldn’t influence things.

the usual solutions are to find the source of the problem (skewed phenotype?) and fix that (take logs?), or to switch to a different method, like haley-knott regression or imputation.

karl

[Vianey B.E]

Thanks for your reply, Karl

It is a 4-way cross, two heterozygous diploid parents with contrasting phenotypes. 

I have other QTLs with gaps of 20cM more or less, but checking the phenotypes I have found that those QTLs appear in skewed phenotypes. I will try your suggestions.  

You received this message because you are subscribed to the Google Groups "R/qtl discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rqtl-disc+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl-disc/60adc9c6-f1c5-4337-9e94-9d1c681eacbbn%40googlegroups.com.