You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
94 lines
4.0 KiB
R
94 lines
4.0 KiB
R
##################################################
|
|
#################### edspk #######################
|
|
##################################################
|
|
edspk <- function(eds.exp, kerpk = 1, fitmaxiter = 50) {
|
|
|
|
eds.base <- baselinefit(eds.exp, tau=2.0, gam=1.0, scl.factor=3.0, maxwdth=0.20)
|
|
|
|
# This loop deals with the output from baselinefit()
|
|
# It makes a "melted" dataframe in long form for each
|
|
# separated peak for some baseline parameters
|
|
eds.pks <- data.frame()
|
|
eds.pks.basl <- data.frame()
|
|
eds.pks.pmg <- data.frame()
|
|
eds.pks.spl <- data.frame()
|
|
peaks <- 1:length(eds.base$npks)
|
|
for (s in peaks) {
|
|
# recorded data in long form by separated peak
|
|
eds.pks <- rbind(eds.pks, # column names assigned after loop
|
|
data.frame(peak = factor(peaks[s]),
|
|
kernel = NA,
|
|
eds.exp[eds.base$indlsep[s]:eds.base$indrsep[s], ]))
|
|
# the calculated baseline in long form by separated peak
|
|
eds.pks.basl <- rbind(eds.pks.basl,
|
|
data.frame(peak = factor(peaks[s]),
|
|
kernel = NA,
|
|
x = eds.exp[eds.base$indlsep[s]:eds.base$indrsep[s]],
|
|
y = eds.base$baseline$basisl[eds.base$indlsep[s]:eds.base$indrsep[s]]))
|
|
# the taut string estimation in long form by separated peak
|
|
eds.pks.pmg <- rbind(eds.pks.pmg,
|
|
data.frame(peak = factor(peaks[s]),
|
|
kernel = NA,
|
|
x = eds.exp[eds.base$indlsep[s]:eds.base$indrsep[s]],
|
|
y = eds.base$pmg$fn[eds.base$indlsep[s]:eds.base$indrsep[s]]))
|
|
# the weighted smoothed spline in long form by separated peak
|
|
eds.pks.spl <- rbind(eds.pks.spl,
|
|
data.frame(peak = factor(peaks[s]),
|
|
kernel = NA,
|
|
x = eds.exp[eds.base$indlsep[s]:eds.base$indrsep[s]],
|
|
y = eds.base$spl$reg[eds.base$indlsep[s]:eds.base$indrsep[s]]))
|
|
}
|
|
# Column names assigned to d.pks
|
|
names(eds.pks) <- c("peak", "kernel", "x", "y")
|
|
|
|
|
|
# This loop calls pkdecompint() on each peak separately
|
|
# It makes a "melted" dataframe in long form for:
|
|
eds.fit <- list() # holds pkdecompint output
|
|
eds.fit.fitpk <- data.frame() # contains fitting curves
|
|
eds.fit.parpk <- data.frame() # physical parameters by peak and kernel
|
|
eds.nobasl <- data.frame() # data with baseline removed
|
|
peaks <- 1:length(eds.base$npks)
|
|
for (s in peaks) {
|
|
######## PKDECOMPINT ########
|
|
if (length(kerpk) > 1) {
|
|
# set number of kernels per peak manually
|
|
eds.fit[[s]] <- pkdecompint(eds.base, intnum = s,
|
|
k = kerpk[s], maxiter = fitmaxiter)
|
|
} else {
|
|
# use number of kernels determined by baselinefit()
|
|
eds.fit[[s]] <- pkdecompint(eds.base, intnum = s,
|
|
k = eds.base$npks[s], maxiter = fitmaxiter)
|
|
}
|
|
# Setup the dataframe that makes up the peak table
|
|
for (kernel in 1:eds.fit[[s]]$num.ker) {
|
|
eds.fit.parpk <- rbind(eds.fit.parpk,
|
|
data.frame(peak = factor(eds.fit[[s]]$intnr),
|
|
kernel = factor(kernel),
|
|
x = eds.fit[[s]]$parpks[kernel, "loc"],
|
|
height = eds.fit[[s]]$parpks[kernel, "height"],
|
|
area = eds.fit[[s]]$parpks[kernel, "intens"],
|
|
fwhm = eds.fit[[s]]$parpks[kernel, "FWHM"],
|
|
m = eds.fit[[s]]$parpks[kernel, "m"],
|
|
accept = eds.fit[[s]]$accept))
|
|
eds.fit.fitpk <- rbind(eds.fit.fitpk,
|
|
data.frame(peak = factor(peaks[s]),
|
|
kernel = factor(kernel),
|
|
x = eds.fit[[s]]$x,
|
|
y = eds.fit[[s]]$fitpk[kernel, ]))
|
|
}
|
|
eds.nobasl <- rbind(eds.nobasl,
|
|
data.frame(peak = factor(peaks[s]),
|
|
x = eds.fit[[s]]$x,
|
|
y = eds.fit[[s]]$y))
|
|
}
|
|
|
|
|
|
|
|
return(list(eds.base = eds.base,
|
|
eds.peaks = eds.pks,
|
|
eds.fit.parpk = eds.fit.parpk,
|
|
eds.fit.fitpk = eds.fit.fitpk,
|
|
eds.nobasl = eds.nobasl))
|
|
|
|
} |