comparison waveica_wrapper.R @ 4:8b55efc7d117 draft

"planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit aa8206a01efc1813f2586584782f28a73a17fe86"

3:e3726251a055

  # get input from the Galaxy, preprocess data
  data <- read.csv(data, header = TRUE)

  required_columns <- c("sampleName", "class", "sampleType", "injectionOrder", "batch")
  if (anyNA(data)) {
    stop("Error: dataframe cannot contain NULL values!
Make sure that your dataframe does not contain empty cells")
  } else if (!all(required_columns %in% colnames(data))) {
    stop("Error: missing metadata!
Make sure that the following columns are present in your dataframe: [sampleName, class, sampleType, injectionOrder, batch]")
  }

  # sort data by injection order
  data <- data[order(data[, "batch"],
    data[, "injectionOrder"],
    decreasing = FALSE
  ), ]

  # separate data into features, batch and group
  feature_columns <- colnames(data)[!colnames(data) %in% required_columns]
  features <- data[, feature_columns]
  group <- enumerate_groups(as.character(data$sampleType))
  batch <- data$batch

  # run WaveICA
  features <- WaveICA::WaveICA(
    data = features,
    wf = get_wf(wavelet_filter, wavelet_length),
    batch = batch,
    group = group,
    K = k,
    t = t,
    t2 = t2,
    alpha = alpha
  )

  data[, feature_columns] <- features$data_wave

  # remove blanks from dataset
  if (exclude_blanks) {
    data <- exclude_group(data, group)
  }

  return(data)
}


# Match group labels with [blank/sample/qc] and enumerate them
enumerate_groups <- function(group) {

4:8b55efc7d117

  # get input from the Galaxy, preprocess data
  data <- read.csv(data, header = TRUE)

  required_columns <- c("sampleName", "class", "sampleType", "injectionOrder", "batch")
  verify_input_dataframe(data, required_columns)

  data <- sort_by_injection_order(data)

  # separate data into features, batch and group
  feature_columns <- colnames(data)[!colnames(data) %in% required_columns]
  features <- data[, feature_columns]
  group <- enumerate_groups(as.character(data$sampleType))
  batch <- data$batch

  # run WaveICA
  features <- recetox.waveica::waveica(
    data = features,
    wf = get_wf(wavelet_filter, wavelet_length),
    batch = batch,
    group = group,
    K = k,
    t = t,
    t2 = t2,
    alpha = alpha
  )

  data[, feature_columns] <- features

  # remove blanks from dataset
  if (exclude_blanks) {
    data <- exclude_group(data, group)
  }

  return(data)
}


waveica_singlebatch <- function(data,
                                wavelet_filter,
                                wavelet_length,
                                k,
                                alpha,
                                cutoff,
                                exclude_blanks) {

  # get input from the Galaxy, preprocess data
  data <- read.csv(data, header = TRUE)

  required_columns <- c("sampleName", "class", "sampleType", "injectionOrder")
  optional_columns <- c("batch")
  verify_input_dataframe(data, required_columns)

  data <- sort_by_injection_order(data)

  feature_columns <- colnames(data)[!colnames(data) %in% c(required_columns, optional_columns)]
  features <- data[, feature_columns]
  injection_order <- data$injectionOrder

  # run WaveICA
  features <- recetox.waveica::waveica_nonbatchwise(
    data = features,
    wf = get_wf(wavelet_filter, wavelet_length),
    injection_order = injection_order,
    K = k,
    alpha = alpha,
    cutoff = cutoff
  )

  data[, feature_columns] <- features

  # remove blanks from dataset
  if (exclude_blanks) {
    data <- exclude_group(data, group)
  }

  return(data)
}


sort_by_injection_order <- function(data) {
  if ("batch" %in% colnames(data)) {
    data <- data[order(data[, "batch"],
      data[, "injectionOrder"],
      decreasing = FALSE
    ), ]
  } else {
    data <- data[order(data[, "injectionOrder"],
      decreasing = FALSE
    ), ]
  }
  return(data)
}


verify_input_dataframe <- function(data, required_columns) {
  if (anyNA(data)) {
    stop("Error: dataframe cannot contain NULL values!
Make sure that your dataframe does not contain empty cells")
  } else if (!all(required_columns %in% colnames(data))) {
    stop("Error: missing metadata!
Make sure that the following columns are present in your dataframe: ", paste(required_columns, collapse = ", "))
  }
}


# Match group labels with [blank/sample/qc] and enumerate them
enumerate_groups <- function(group) {