FlowSOM: Run the FlowSOM algorithm

In saeyslab/FlowSOM: Using self-organizing maps for visualization and interpretation of cytometry data

Run the FlowSOM algorithm

Description

Method to run general FlowSOM workflow. Will scale the data and uses consensus meta-clustering by default.

Usage

FlowSOM(
  input,
  pattern = ".fcs",
  compensate = FALSE,
  spillover = NULL,
  transform = FALSE,
  toTransform = NULL,
  transformFunction = flowCore::logicleTransform(),
  transformList = NULL,
  scale = FALSE,
  scaled.center = TRUE,
  scaled.scale = TRUE,
  silent = TRUE,
  colsToUse = NULL,
  nClus = 10,
  maxMeta = NULL,
  importance = NULL,
  seed = NULL,
  ...
)

Arguments

input	a flowFrame, a flowSet, a matrix with column names or an array of paths to files or directories
pattern	if input is an array of file- or directorynames, select only files containing pattern
compensate	logical, does the data need to be compensated
spillover	spillover matrix to compensate with If NULL and compensate = TRUE, we will look for $SPILL description in FCS file.
transform	logical, does the data need to be transformed with the transformation given in transformFunction.
toTransform	column names or indices that need to be transformed. Will be ignored if transformList is given. If NULL and transform = TRUE, column names of $SPILL description in FCS file will be used.
transformFunction	Defaults to logicleTransform()
transformList	transformList to apply on the samples.
scale	logical, does the data needs to be rescaled. Default = FALSE
scaled.center	see scale
scaled.scale	see scale
silent	if TRUE, no progress updates will be printed
colsToUse	Markers, channels or indices to use for building the SOM. Default (NULL) is all the columns used to build the FlowSOM object.
nClus	Exact number of clusters for meta-clustering. Ignored if maxMeta is specified. Default = 10.
maxMeta	Maximum number of clusters to try out for meta-clustering. If NULL (default), only one option will be computed (nClus).
importance	array with numeric values. Parameters will be scaled according to importance
seed	Set a seed for reproducible results
...	options to pass on to the SOM function (xdim, ydim, rlen, mst, alpha, radius, init, distf)

Value

A list with two items: the first is the flowSOM object containing all information (see the vignette for more detailed information about this object), the second is the metaclustering of the nodes of the grid. This is a wrapper function for ReadInput, BuildSOM, BuildMST and MetaClustering. Executing them separately may provide more options.

See Also

scale, ReadInput, BuildSOM, BuildMST, MetaClustering

Examples

# Read from file
fileName <- system.file("extdata", "68983.fcs", package = "FlowSOM")
flowSOM.res <- FlowSOM(fileName, compensate = TRUE, transform = TRUE,
                      scale = TRUE, colsToUse = c(9, 12, 14:18), nClus = 10)
# Or read from flowFrame object
ff <- flowCore::read.FCS(fileName)
ff <- flowCore::compensate(ff, flowCore::keyword(ff)[["SPILL"]])
ff <- flowCore::transform(ff,
         flowCore::transformList(colnames(flowCore::keyword(ff)[["SPILL"]]),
                                flowCore::logicleTransform()))
flowSOM.res <- FlowSOM(ff, 
                       scale = TRUE, 
                       colsToUse = c(9, 12, 14:18), 
                       nClus = 10)

# Plot results
PlotStars(flowSOM.res,
          backgroundValues = flowSOM.res$metaclustering)

# Get metaclustering per cell
flowSOM.clustering <- GetMetaclusters(flowSOM.res)

saeyslab/FlowSOM documentation built on April 20, 2024, 8:30 p.m.