mergeWindow {mixer} | R Documentation |
Grouping identified windows with significant scores into non-overlapped peak regions
mergeWindow(windowX, wl, probeX, pl, radius, lambda=1/500, window.cut=0.05, probe.cut.type = "lfdr", probe.cut=0.1, n.probe=3)
windowX |
window level score. Output of funciton win |
wl |
local FDR for winodw level score, output of fucntion lfdr |
probeX |
probe level signal |
pl |
local FDR for probe level signal, output of fucntion lfdr |
radius |
half of the window size |
lambda |
incident rate in the Poisson point process modeling the sonicantion process. If on average, there is one break up of DNA sequence per 500 bp, lambda = 1/500. |
window.cut |
cutoff of window level local FDR |
probe.cut.type |
if probe.cut.type="lfdr", use lfdr as probe level cutoff, otherwise use probe signal as cutoff |
probe.cut |
cutoff of probe level local FDR |
n.probe |
claim on window as a peak if at least n.probe in one window has local FDR smaler than probe.cut |
it is not precisly "merge". We simply pick up one significant peak and eliminate related probes and then re-evaluate window level score and pick up another peak if there is one.
data.frame, each row for one peak
start |
start position of the peak |
end |
end position of the peak |
score |
score of the peak |
n.probe |
the total number of probes in one peak |
lfdr |
local FDR |
Wei Sun