n.marcovar

[Ian W.]

Hi Karl, 

I understand that the CIM function is a slightly crude method for composite interval mapping, I've been using it as a sort of preliminary scan for multiple QTLs.

I have a query about how increasing n.marcovar increases LOD scores and hit number? I found this from even changing n.marcovar  from 3 to 4, which increases the number of QTLs identified and the LOD score of those hits. However, it seems like I'm artificially increasing the significance of my results? I have not taken into account how the permutations results would change, maybe this is accounted for. Is there a maximum/minimum number or a way of determining the most appropriate n.marcovar ?   

Thanks for your time (and help in my previous query).  

[Karl Broman]

I think you want n.marcovar to be equal to the number of detectable QTL. But this is not something that you'd know in advance, and in my mind it's the main disadvantage of this sort of CIM approach and why I prefer the penalized LOD score criterion encompassed in stepwiseqtl(). See Broman and Speed (2002) https://www.biostat.wisc.edu/~kbroman/publications/jrssb.pdf

But for sure you'll want to re-do the permutations for the chosen value of n.marcovar, but this won't account for the search over different values of n.marcovar, if you're going to try different values and then pick the strongest results.

karl

> --
> 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 post to this group, send email to rqtl...@googlegroups.com.
> Visit this group at https://groups.google.com/group/rqtl-disc.
> For more options, visit https://groups.google.com/d/optout.