usecases/kcomparison/functions.R

In neobernad/evaluomeR: Evaluation of Bioinformatics Metrics

getFormattedK <- function(k) {
  return(gsub("^.*_","", k))
}

getDfQualityData <- function(qualData, k.range=NULL, digits=2) {
  lengthQuality = length(qualData)
  qualRangeStart = getFormattedK(names(qualData)[1])
  qualRangeEnd = getFormattedK(names(qualData)[lengthQuality])
  Metric = NULL
  kValues = list()
  for (i in seq(qualRangeStart, qualRangeEnd, 1)) {
    curQual = as.data.frame(assay(getDataQualityRange(qualData, i)))
    if (i == qualRangeStart) {
      Metric = as.character(levels(curQual$Metric))
    }
    kValues[[i]] = as.numeric(as.character(curQual$Avg_Silhouette_Width))
  }
  qualDf = as.data.frame(Metric)
  for (i in seq(qualRangeStart, qualRangeEnd, 1)) {
    values = round(kValues[[i]], digits = digits)
    newColname = paste0("k_", i)
    k = as.numeric(getFormattedK(newColname))
    if (!is.null(k.range) && (k < k.range[1] || k > k.range[2])) {
      next
    }
    qualDf[[newColname]] = values
  }

  if (!is.null(k.range) && (k.range[1] < qualRangeStart || k.range[2] > qualRangeEnd)) {
    # Input k.range is not a subset of the stabData k ranges
    stop("Input k.range [", k.range[1], ", ", k.range[2], "] is not a subset of range [",
         qualRangeStart, ", ", qualRangeEnd, "]")
  }

  rownames(qualDf) = qualDf$Metric
  qualDf = qualDf[, -1] # Remove "Metric" column, metrics are rownames now
  qualDf <- qualDf[ order(row.names(qualDf)), ]
  return(qualDf)
}

#
# It transforms the output of stabilityRange method
# into a dataframe like this:
# (rownames)       k_2       k_3       k_4
# RIN              0.6171262 0.6278294 0.4882649
# ...
# So that the input of getOptimalKValue has always a
# standardized dataframe to process.
#
getDfStabilityData <- function(stabData, k.range=NULL, digits=2) {
  stabDf = as.data.frame(assay(stabData))
  lengthColnames = length(colnames(stabDf))
  toRemove = list()
  for (i in seq(1, lengthColnames, 1)) {
    colname = colnames(stabDf)[i]
    newColname = gsub("^.*_.*_.*_","k_", colname)
    colnames(stabDf)[i] = newColname
    if (i != 1) { # Skip Metric column
      k = as.numeric(getFormattedK(newColname))
      if (!is.null(k.range) && (k < k.range[1] || k > k.range[2])) {
        toRemove = append(toRemove, newColname)
        next
      }
      stabDf[newColname] = round(
                            as.numeric(as.character(stabDf[[newColname]])),
                            digits = digits
                           )
    }
  }

  for (columnName in toRemove) {
    stabDf[, columnName] = list(NULL)
    lengthColnames = lengthColnames-1
  }

  inputStartRange = as.numeric(getFormattedK(colnames(stabDf)[2]))
  inputEndRange = as.numeric(getFormattedK(colnames(stabDf)[lengthColnames]))
  if (!is.null(k.range) && (k.range[1] < inputStartRange || k.range[2] > inputEndRange)) {
    # Input k.range is not a subset of the stabData k ranges
    stop("Input k.range [", k.range[1], ", ", k.range[2], "] is not a subset of data range [",
         inputStartRange, ", ", inputEndRange, "]")
  }

  rownames(stabDf) = stabDf$Metric
  stabDf = stabDf[, -1] # Remove "Metric" column, metrics are rownames now
  stabDf <- stabDf[ order(row.names(stabDf)), ]
  return(stabDf)
}