R/fillback.R

In sneumann/RMassBank: Workflow to process tandem MS files and build MassBank records

# TODO: Add comment
# 
# Author: stravsmi
###############################################################################

#' Fill back reanalyzed / refiltered peak info into spectra
#' 
#' This method takes the info which is added to the aggregated table in the reanalysis and 
#' multiplicity filtering steps of the workflow, and adds it back into the spectra.
#' 
#' @export
setGeneric("fillback", function(o, ...) standardGeneric("fillback"))

#' @export
setMethod("fillback", c("msmsWorkspace"), function(o, ...)
    {
      for(i in seq_len(length(o@spectra)))
        o@spectra[[i]] <- fillback(o@spectra[[i]], o@aggregated)
      o
    })

#' @export
setMethod("fillback", c("RmbSpectraSet"), function(o, aggregated)
    {
      for(i in seq_len(length(o@children)))
        o@children[[i]] <- fillback(o@children[[i]], o@id, aggregated)
      o
    })

#' @export
setMethod("fillback", c("RmbSpectrum2"), function(o, id, aggregated)
    {
      .fillback(o, id, aggregated)  
    })

.fillback <- function(o, id, aggregated)
{
  peaks <- selectPeaks(aggregated,
      (cpdID == id) & (scan == o@acquisitionNum))
  curPeaks <- getData(o)
  # Check that our data processing assumptions are correct: the peaks that are rawOK are the peaks
  # that are in the aggregate table. (This is a very rough check: nrow identical)
  stopifnot(nrow(peaks)==nrow(curPeaks[curPeaks$rawOK,,drop=FALSE]))
  # If dppmBest is in the aggregate table, drop the one in the current spectrum, because the refiltering
  # can remove peaks with better dppm!
  if(("dppmBest" %in% colnames(peaks)) & ("dppmBest" %in% colnames(curPeaks)))
    curPeaks <- curPeaks[,which(colnames(curPeaks) != "dppmBest"),drop=FALSE]
  # Find all columns that we don't have yet in curPeaks
  colnames(peaks)[colnames(peaks) == "mzFound"] <- "mz"
  colsRemove <- c("scan", "cpdID", "parentScan", "index")
  colsRemoveIndex <- which(colnames(peaks) %in% colsRemove)
  peaks <- peaks[,-colsRemoveIndex,drop=FALSE]
  colsNew <- setdiff(colnames(peaks), colnames(curPeaks))
  for(col in colsNew)
  {
    if(col != "dppmBest")
      o <- addProperty(o, col, class(peaks[,col]))
  }
  peaksNew <- merge(curPeaks, peaks, all=TRUE)
  # Check that no stray "new peaks" were added by incorrect merging. If this happens, we have to write cleaner code
  stopifnot(nrow(peaksNew) == nrow(curPeaks))
  o <- setData(o, peaksNew)
  #browser()
  return(o)
}