comparing penalized lod scores

[maridel.fr...@gmail.com]

Hello team, 

I have a question related to interpreting and comparing penalized LOD scores. 

I understand that the stepwiseqtl() function reports the model with the maximum pLOD score among all models tested. I would like to also report the second highest pLOD score for comparison (which I can find in the attributes of the stepwiseqtl output). However, is there some kind of relative likelihood estimate for these pLOD scores (something akin to AIC weight?) which would indicate a measure of our confidence that the model with the maximum pLOD score is actually the best one?

Thank you very much for the help.

[Karl Broman]

I don't have any formal way to compare penalized LOD scores.

If the two models are "nested", the differences in LOD scores without the penalties would be meaningful.

"...is actually the best one" depends on the meaning of the term "best". If you're seeking the model with maximum penalized LOD score, which is the goal for stepwiseqtl(), then the penalized LOD scores themselves tell you which one is "better". If you're thinking of "best" as "closest to the truth" or "contains as many true QTL as possible, with as few extraneous terms", well that's not knowable.

karl