Nothing
#' Signal grouping
#'
#' Euclidean distances between m/z signals are hierarchically clustering using
#' the average method and the composite spectrum groups determined by an
#' absolute
#' error cutoff
#'
#' @param spectrum.df a dataframe or matrix with two or more columns:
#' 1. Mass/ Mass-to-charge ratio
#' 2. Intensity
#' @param mzError interpeak absolute m/z error for signal grouping
#' (Default = 0.001)
#' @param minPeaks numeric minimum number of peaks to integrate
#' @usage signalGrouping(spectrum.df = NULL, mzError = 0.8, minPeaks = 5)
#' @return dataframe of m/z grouped signals, the m/z values of the input
#' dataframe/ matrix peak groups are averaged and the signal intensities
#' summed.
signalGrouping <-
function(spectrum.df = NULL,
mzError = 0.8,
minPeaks = 5) {
# error handling
if (is.null(spectrum.df)) {
stop("No spectrum matrix/dataframe supplied")
} else {
hr <-
fastcluster::hclust.vector(spectrum.df[, 1], method = "median",
members = NULL)
# cut tree according to absolute m/z error
spectrum_group <- cutree(hr, h = mzError)
# calculate weighted mean of the m/z and
#sum signal within each peak group
mass <-
do.call(c, as.list(by(spectrum.df, as.factor(spectrum_group),
function(x) {
weighted.mean(x[, 1], x[, 2])
})))
grouped.df <- data.frame(
mass = mass,
intensity = tapply(spectrum.df[, 2],
as.factor(spectrum_group), sum),
stringsAsFactors = FALSE
)
# average any additional columns i.e. retention time
if (ncol(spectrum.df) > 2) {
groupedCols <- apply(spectrum.df[, 3:ncol(spectrum.df),
drop = FALSE], 2,
function(x)
tapply(x, as.factor(spectrum_group),
mean))
grouped.df <- cbind(grouped.df, groupedCols)
}
if (nrow(grouped.df) > minPeaks) {
if (!is.null(colnames(spectrum.df))) {
colnames(grouped.df) <- colnames(spectrum.df)
}
return(grouped.df)
} else {
return("Less than minPeak")
}
}
}
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.