CI issue with find_peaks

[Richard Radcliffe]

Hello,

We're mapping mouse F2's derived from consomic x background crosses so we've only genotyped the single consomic chromosome for each F2. On chr 1, we've phenotyped 169 mice and genotyped 7 markers (~20 cM to ~84 cM) and have mapped a highly significant peak using the standard HK procedure (figure attached). The analysis included pseudomarkers spaced at 1 cM. I've tried finding the confidence intervals using find_peaks and no matter what I do, ci_lo and ci_hi always end up on a marker. For example, this is the code I used for the figure (ci shown as vertical turquoise lines):

find_peaks(out, map, threshold=4, peakdrop = 3, drop = 1)

ci_lo = 29.2

ci_hi = 68.5

I can change the values which can change ci_lo/hi, but they always end up on a marker. Same with lod_int and bayes_int.

Any thoughts?

Thanks,

Richard

(qtl2, v 0.24)

[Karl Broman]

There's an argument expand2markers that defaults to TRUE. So try

    find_peaks(out, map, threshold=4, peakdrop=3, drop=1, expand2markers=FALSE)

karl

[Richard Radcliffe]

That did the trick. Thanks and thanks for the speedy response.

-Richard