SignalProcessingUltra: Optimized signal processing for MALDI-TOF spectra (parallel + optional C++ alignment)

Description

This function performs post-acquisition processing for lists of MALDIquant MassSpectrum objects. It parallelizes per-spectrum steps and can align with either MALDIquant's warping (cubic/lowess) or a fast landmark-based C++ algorithm ("landmark_cpp").

Usage

SignalProcessingUltra(
  x,
  transformIntensity_method = "log",
  smoothing_method = "Wavelet",
  removeBaseline_method = "TopHat",
  removeBaseline_iterations = 25,
  calibrateIntensity_method = "PQN",
  alignSpectra_NoiseMethod = "MAD",
  alignSpectra_method = c("cubic", "lowess", "landmark_cpp"),
  alignSpectra_halfWs = 11,
  alignSpectra_SN = 3,
  tolerance_align = 0.002,
  ppm_align = FALSE,
  referenceSpectra = NULL,
  minFrequency = 0.7,
  binPeaks_method = "strict",
  keepReferenceSpectra = FALSE,
  n_workers = NULL,
  ref_sample_n = NULL,
  verbose = TRUE,
  ...
)

Value

A list of MassSpectrum objects, or a list with $spectra and $RefS if keepReferenceSpectra = TRUE.

Arguments

x: list of MALDIquant MassSpectrum objects.
transformIntensity_method: character, intensity transform (default "log").
smoothing_method: character, smoothing method ("Wavelet" UDWT).
removeBaseline_method: character, baseline method ("TopHat" default; "SNIP","ConvexHull" supported).
removeBaseline_iterations: integer, SNIP iterations if removeBaseline_method = "SNIP".
calibrateIntensity_method: character, intensity calibration ("PQN" default, or "TIC","median").
alignSpectra_NoiseMethod: character, noise estimator for peak finding pre-alignment ("MAD").
alignSpectra_method: character, alignment engine: "cubic" (default), "lowess", or "landmark_cpp".
alignSpectra_halfWs: integer, half window size for peak detection.
alignSpectra_SN: numeric, SNR for peak detection.
tolerance_align: numeric, tolerance for matching anchors to the reference during alignment. Use consistent units across your pipeline (Da by default here).
ppm_align: logical, set TRUE if tolerance_align is in ppm (then interpreted as ppm).
referenceSpectra: optional MALDIquant MassPeaks object to use as alignment reference.
minFrequency: numeric, minimum peak frequency to build reference if not provided (default 0.7).
binPeaks_method: character, "strict" (default) or "relaxed" for reference peak binning.
keepReferenceSpectra: logical, if TRUE and no reference provided, returns list(spectra=..., RefS=...).
n_workers: integer, number of parallel workers (default: all cores minus one).
ref_sample_n: integer or NULL, if set, build the reference from a random subset of this many spectra.
verbose: logical, print progress.
...: passed to MALDIrppa::wavSmoothing (e.g., n.levels).