R/calibrateIntensity-functions.R

Defines functions .calibrateProbabilisticQuotientNormalization .calibrateIntensitySimple .scalingFactor

## .scalingFactor
## calculate scaling factor for TIC/median ranged calibration
##
## params:
##  object: AbstractMassObject
##  method: character
##  range: double
##
## returns:
##  double
##
.scalingFactor <- function(object, method=c("TIC", "median"), range) {

  method <- match.arg(method)

  if (!missing(range)) {
    object <- trim(object, range=range)
  }

  switch(method,
    "TIC" = {
      totalIonCurrent(object)
    },
    "median" = {
      median(object@intensity)
    }
  )
}

## .calibrateIntensitySimple
## calibrate intensity values by offset and scaling factor
##
## params:
##  y: double, intensity values
##  offset: double
##  scaling: double
##
## returns:
##  double, calibrated intensity values
##
.calibrateIntensitySimple <- function(y, offset=0L, scaling=1L) {
  if (scaling == 0L) {
    warning("Scaling factor is zero. No calibration applied.")
    y
  } else {
    (y - offset) / scaling
  }
}

## .calibrateProbabilisticQuotientNormalization
## calibrate intensity values by Probabilistic Quotient Normalization
##
## F. Dieterle, A. Ross, G. Schlotterbeck, and Hans Senn.
## "Probabilistic quotient normalization as robust method to account for
## dilution of complex biological mixtures. Application in 1H NMR
## metabonomics."
## Analytical Chemistry 78, no. 13 (2006): 4281-4290.
##
## 1. Perform an integral normalization (typically a constant
##    integral of 100 is used).
## 2. Choose/calculate the reference spectrum (the best approach
##    is the calculation of the median spectrum of control samples).
## 3. Calculate the quotients of all variables of interest of the test
##    spectrum with those of the reference spectrum.
## 4. Calculate the median of these quotients.
## 5. Divide all variables of the test spectrum by this median.
##
## params:
##  l: list of MassSpectrum objects
##  range: double
##
## returns:
##  list of calibrated MassSpectrum objects
##
.calibrateProbabilisticQuotientNormalization <- function(l, range) {
  ## 1. integral normalization (==TIC)
  l <- calibrateIntensity(l, method="TIC", range=range)
  ## 2. median reference spectrum
  if (missing(range)) {
    reference <- .averageMassSpectra(l, fun=.colMedians, mergeMetaData=FALSE)
  } else {
    reference <- .averageMassSpectra(trim(l, range=range), fun=.colMedians,
                                     mergeMetaData=FALSE)
  }

  lapply(l, function(x) {
    ## 3. quotient calculation
    q <- approxfun(x)(reference@mass) / reference@intensity
    ## 4. median
    m <- median(q, na.rm=TRUE)
    ## 5. divide by median
    .transformIntensity(x, fun=.calibrateIntensitySimple, offset=0L, scaling=m)
  })
}
sgibb/MALDIquant documentation built on Jan. 26, 2024, 10:04 a.m.