The new summary.scantwo and plot.scantwo

In R/qtl version 1.04, the functions summary.scantwo and plot.scantwo have been changed quite substantially. Also, the permutations with scantwo have been changed to match the new format for summary.scantwo. In this document, I describe the revisions and how to use the new functions. We'll look at the hyper data as an example. First we need to load the package and the data. > library(qtl) > data(hyper) I'm going to use scantwo with method="em". First I run calc.genoprob, and then scantwo as before. > hyper <-calc.genoprob(hyper, step=2.5) > out2 <-scantwo(hyper) The output, in this new version of R/qtl, is slightly different. The LOD scores for the full model (two QTL plus interaction) are there as before, but in place of the epistasis LOD scores, I save the LOD scores for the additive QTL model. Also, we now always run the single-QTL analysis with scanone, since the results are necessary to make sense of the output of scantwo. The big change is in summary.scantwo. Consider a pair of positions in the genome, s and t. We consider four models. Full: y = µ + β 1 q 1 + β 2 q 2 + β 3 (q 1 × q 2) + ǫ Add: y = µ + β 1 q 1 + β 2 q 2 + ǫ One: y = µ + β 1 q 1 + ǫ Null: y = µ + ǫ Let l f (s, t) be the log 10 likelihood for the full model with QTL at s and t, l a (s, t) be the log 10 likelihood for the additive model with QTL at s and t, l 1 (s) be the log 10 likelihood for the single-QTL model with the QTL at s, and l 0 be the log 10 likelihood under the null (with no QTL). Define the LOD scores as follows.