R/IsoCombinations.R

In IsoCorrectoR: Correction for natural isotope abundance and tracer purity in MS and MS/MS data from stable isotope labeling experiments

#' @importFrom dplyr last
#' @importFrom dplyr arrange_
#' @importFrom stringr str_c
#' @importFrom tibble as_tibble
#' @importFrom magrittr '%>%'

IsoCombinations <- function(Fragment, x, Element, ElementInfo, AvailablePlacesTotal, IsoCluster, CalculationThreshold) {

  # NumberIso is the number of isotopes per element.
  # The number of nested loops used depends on the number of
  # isotopes of the current element.

  NumberIso <- nrow(ElementInfo[[Element]][["Isotopes"]][[1]])

  Combinations <- vector()
  Combinations[1] <- 1

  # Calculation of Combinations[IsotopeNo+1], the number of isotope combinations
  # when considering IsotopeNo different isotopes (also including combinations
  # where the sum of isotopes exceeds the number of available places IsoCluster).

  for (IsotopeNo in seq_len(NumberIso)) {
    Combinations[IsotopeNo + 1] <- Combinations[IsotopeNo] * ((IsoCluster) + 1)
  } # IsotopeNo

  TotalCombinations <- dplyr::last(Combinations) # get last element

  # Computation of the isotope combinations list `CombinationsArrayPreList`.
  # At this point, combinations where the sum of the different isotopes exceeds
  # IsoCluster are still considered, leading to invalid combinations that are
  # later removed. In its columns, CombinationsArrayPreList contains the number
  # of occurences of a given isotope in a given combination.
  # Each row corresponds to a combination.

  CombinationsArrayPreList <- list()

  for (IsotopeNo in seq_len(NumberIso)) {
    p <- 1
    tmpCombinationsArrayPre <- vector()

    # first, for each column (isotope) the repetitive column element is computed

    for (IsoCount in 0:IsoCluster) {
      tmpCombinationsArrayPre[p:(p + Combinations[IsotopeNo] - 1)] <- IsoCount
      p <- p + Combinations[IsotopeNo]
    } # IsoCount

    # Then, the repetitive column element is repeatedly inserted until the
    # number of rows matches `TotalCombinations`

    for (Repeat in seq_len((TotalCombinations / Combinations[IsotopeNo + 1])) - 1) {
      if (Repeat > 0) {
        tmpCombinationsArrayPre[p:(p + Combinations[IsotopeNo + 1] - 1)] <- tmpCombinationsArrayPre[seq_len(Combinations[IsotopeNo + 1])]
        p <- p + Combinations[IsotopeNo + 1]
      }
    } # Repeat

    CombinationsArrayPreList[[IsotopeNo]] <- tmpCombinationsArrayPre
  } # IsotopeNo
  names(CombinationsArrayPreList) <- stringr::str_c("Isotope_", seq_len(NumberIso))

  # CombinationsArrayPreList is sorted to get the most probable combinations on top

  CombinationsArrayPreList2 <- CombinationsArrayPreList %>% as_tibble() %>% dplyr::arrange_(.dots = names(CombinationsArrayPreList)) %>% as.list()

  CombinationsArrayPre.df <- as.data.frame(lapply(CombinationsArrayPreList2, function(x) rev(x)))

  RowSum <- apply(CombinationsArrayPre.df, 1, sum)
  j <- 1

  # CombinationsArray is derived from CombinationsArrayPreList by removing
  # all combinations that do not have a row sum of IsoCluster

  CombinationsArray <- NULL

  for (i in seq_len(TotalCombinations)) {
    if (RowSum[i] == IsoCluster) {
      tmpA <- vector()

      tmpA <- as.numeric(CombinationsArrayPre.df[i, seq_len(NumberIso)])
      tmpA[1] <- tmpA[1] + (AvailablePlacesTotal - IsoCluster)
      CombinationsArray <- rbind(CombinationsArray, tmpA)

      j <- j + 1
    } # RowSum[i]==IsoCluster
  } # i

  # The number of combinations in CombinationsArray is derived,
  # CombinationsArray is stored in Isotopes

  NumberIsoCombMax <- nrow(CombinationsArray)

  Isotopes <- CombinationsArray

  # The probabilities and mass shifts of the isotope combinations in
  # CombinationsArray are computed using IsoCombinationsProbMassShift()

  IsoComb <- 1
  ProbArray.vec <- vector()
  MassShiftArray.vec <- vector()
  NumberIsoComb.vec <- vector()

  for (IsoCombPre in seq_len(NumberIsoCombMax)) {
    icpms <- IsoCombinationsProbMassShift(
      Element = Element, NumberIso = NumberIso, ElementInfo = ElementInfo, CombinationsArray = CombinationsArray,
      IsoCombPre = IsoCombPre, AvailablePlacesTotal = AvailablePlacesTotal
    )

    # Due to CalculationThreshold which limits calculations to probability
    # values large enough to be relevant, probability and mass shift are
    # usually not computed for all isotope combinations.

    if (icpms[["ProbIsoComb"]] > CalculationThreshold) {
      ProbArray.vec[IsoComb] <- icpms[["ProbIsoComb"]]
      MassShiftArray.vec[IsoComb] <- icpms[["MassShiftIsoComb"]]
      NumberIsoComb <- IsoComb
      IsoComb <- IsoComb + 1
    } else {
      if (IsoComb == 1) {
        NumberIsoComb <- 0
      }

      IsoCombPre <- NumberIsoCombMax + 1
    }
  } # IsoCombPre

  return(list(Isotopes = Isotopes, ProbArray = ProbArray.vec, MassShiftArray = MassShiftArray.vec, NumberIsoComb = NumberIsoComb))
} # IsoCombinations