R/SignalProcessing.R

In MSclassifR: Automated Classification of Mass Spectra

########################
## Spectral treatment ##
########################

SignalProcessing <- function (x,
                              transformIntensity_method = "sqrt",
                              smoothing_method = "Wavelet",
                              removeBaseline_method = "SNIP",
                              removeBaseline_iterations = 25,
                              calibrateIntensity_method = "TIC",
                              alignSpectra_NoiseMethod = "MAD",
                              alignSpectra_method = "lowess",
                              alignSpectra_halfWs = 11,
                              alignSpectra_SN = 3,
                              tolerance_align = 0.002,
                              referenceSpectra = NULL,
                              minFrequency = 0.5,
                              binPeaks_method = "strict",
                              keepReferenceSpectra = FALSE,
                              ...){
  ## Signal treatment ##

  # Transform intensity
  print(paste0("1- Transform intensities with ",transformIntensity_method," method"))
  spectra <- MALDIquant::transformIntensity(x,
                                            method = transformIntensity_method)
  # Smoothing
  print(paste0("2- Smoothing with ",smoothing_method," method"))
  spectra <- MALDIrppa::wavSmoothing(spectra,
                                     method = smoothing_method, ...)
  # Remove Baseline
  print(paste0("3- Remove baseline with ",removeBaseline_method," method"))
  spectra <- MALDIquant::removeBaseline(spectra,method = removeBaseline_method,
                                        iterations = removeBaseline_iterations)
  # Calibration intensities
  print(paste0("4- Calibrate intensity with ",calibrateIntensity_method," method"))
  spectra_before <- MALDIquant::calibrateIntensity(spectra,
                                                   method = calibrateIntensity_method)
  #Reference spectrum for alignment
  peaks_before <- MALDIquant::detectPeaks(spectra_before,
                                          halfWindowSize = alignSpectra_halfWs,
                                          method = alignSpectra_NoiseMethod,
                                          SNR = alignSpectra_SN)
  
  if(is.null(referenceSpectra)){
  print(paste0("5- Creating a reference spectrum to align spectra"))
  referenceSpectraAlign <- MALDIquant::referencePeaks(peaks_before,
                                                      minFrequency=minFrequency,
                                                      method = binPeaks_method)
  }else{print(paste0("5- Reference spectrum to align the spectra has been provided"))
    referenceSpectraAlign <- referenceSpectra
    }
  
  # Aligment
  print(paste0("6- Align spectra with ",alignSpectra_method," method"))
  spectraPro <- try(MALDIquant::alignSpectra(spectra_before,
                                             halfWindowSize = alignSpectra_halfWs,
                                             SNR = alignSpectra_SN,
                                             tolerance = tolerance_align,
                                             warpingMethod = alignSpectra_method,
                                             allowNoMatches = TRUE,
                                             emptyNoMatches = FALSE,
                                             reference = referenceSpectraAlign))
  
  if(is.character(spectraPro)){
    warning("Processed spectra are available but were not aligned due to an error (minFrequency is maybe too high)!")
    spectra <- spectra_before
  }
  
  if(is.null(referenceSpectra)){
    if(keepReferenceSpectra == TRUE){
    spectra = list(
      spectra = spectraPro,
      RefS = referenceSpectraAlign)}else{spectra <- spectraPro}
    }else{spectra <- spectraPro}
  
  return(spectra) # List of spectra after signal processing
}