QuantileNorm.train: QuantileNorm.train

In saeyslab/CytoNorm: Normalisation of cytometry data measured across multiple batches

QuantileNorm.train

Description

Learn the batch effects from control samples. This function computes quantiles to describe the distribution of the data, and infers spline functions to equalize these distributions over the files. Typically, you will use the function QuantileNorm.normalize after this function to reverse the batch effects on other files.

Usage

QuantileNorm.train(
  files,
  labels,
  channels,
  transformList,
  nQ = 99,
  limit = NULL,
  quantileValues = NULL,
  goal = "mean",
  verbose = FALSE,
  plot = FALSE,
  plotTitle = "Quantiles",
  ...
)

Arguments

files	Full paths of to the fcs files of the control samples
labels	A label for every file, indicating to which batch it belongs, e.g. the plate ID.
channels	Names of the channels to normalize
transformList	Transformation list to pass to the flowCore transform function
nQ	Number of quantiles to use. Default = 99, which results in quantiles for every percent of the data.
limit	These values will be modelled to map onto themselves by the spline
quantileValues	If specified, it should be a vector of length nQ with values between 0 and 1, giving the percentages at which the quantiles should be computed. If NULL (default), the quantiles will be evenly distributed, including 0 and 1.
goal	Goal distribution. Default "mean", can also be nQ numeric values or one of the batch labels.
verbose	If TRUE, progress updates are printed. Default = FALSE.
plot	If TRUE, a plot is generated (using the layout function) showing all quantiles. Default = FALSE.
plotTitle	Title to use in the plot. Default = "Quantiles".
...	Additional arguments to pass to read.FCS

Value

A list containing all the splines and quantile information. This can be used as input for the QuantileNorm.normalize function.

See Also

CytoNorm.train, QuantileNorm.normalize

Examples

dir <- system.file("extdata", package = "CytoNorm")
files <- list.files(dir, pattern = "fcs$")
data <- data.frame(File = files,
                   Path = file.path(dir, files),
                   Type = stringr::str_match(files, "_([12]).fcs")[,2],
                   Batch = stringr::str_match(files, "PTLG[0-9]*")[,1],
                   stringsAsFactors = FALSE)
data$Type <- c("1" = "Train", "2" = "Validation")[data$Type]
train_data <- dplyr::filter(data, Type == "Train")

ff <- flowCore::read.FCS(data$Path[1])
channels <- grep("Di$", flowCore::colnames(ff), value = TRUE)
transformList <- flowCore::transformList(channels,
                                         cytofTransform)

png("nQ99.png",
    width = length(channels) * 300,
    height = (nrow(train_data) * 2 + 1) * 300)
model_nQ_99 <- QuantileNorm.train(
  files = train_data$Path,
  labels = train_data$Batch,
  channels = channels,
  transformList = transformList,
  nQ = 99,
  plot = TRUE)
dev.off()

png("nQ99_limited.png",
    width = length(channels) * 300,
    height = (nrow(train_data) * 2 + 1) * 300)
model_nQ_99 <- QuantileNorm.train(
  files = train_data$Path,
  labels = train_data$Batch,
  channels = channels,
  transformList = transformList,
  nQ = 99,
  limit = c(0,8),
  plot = TRUE)
dev.off()

png("nQ_2.png",
    width = length(channels) * 300,
    height = (nrow(train_data) * 2 + 1) * 300)
model_nQ_2 <- QuantileNorm.train(
  files = train_data$Path,
  labels = train_data$Batch,
  channels = channels,
  transformList = transformList,
  nQ = 2,
  quantileValues = c(0.001, 0.999),
  plot = TRUE)
dev.off()

model_goal_mean <- QuantileNorm.train(
  files = train_data$Path,
  labels = train_data$Batch,
  channels = channels,
  transformList = transformList)

model_goal_batch1 <- QuantileNorm.train(
  files = train_data$Path,
  labels = train_data$Batch,
  channels = channels,
  transformList = transformList,
  goal = "PTLG021")

model_goal_fixed <- QuantileNorm.train(
  files = train_data$Path,
  labels = train_data$Batch,
  channels = channels,
  transformList = transformList,
  nQ = 99,
  goal = seq(0.01, 0.99, by = 0.01))

saeyslab/CytoNorm documentation built on Nov. 2, 2024, 12:39 p.m.