processFCS: Input, filter, normalise, and transform FCS expression data.

Description Usage Arguments Details Value Author(s) Examples

View source: R/processFCS.R

Description

Input, filter, normalise, and transform FCS expression data.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
processFCS(
  files,
  assayname = "scaled",
  metadata = NULL,
  filter = TRUE,
  bgNoiseThreshold = 1,
  euclideanNormThreshold = 1,
  transformation = TRUE,
  transFun = function(x) asinh(x),
  asinhFactor = 5,
  downsample = 1e+05,
  downsampleVar = 0.1,
  colsDiscard = NULL,
  colsRetain = NULL,
  newColnames = NULL,
  emptyValue = TRUE,
  verbose = TRUE
)

Arguments

files

A vector of FCS files.

assayname

Name of the assay slot in which data will be stored.

metadata

Metadata associated with the FCS files specified in 'files'. A strict rule is enforced requiring that rownames(metadata) matches files in both name and order.

filter

Boolean (TRUE / FALSE) to enable filtering (per sample) for background signal / noise.

bgNoiseThreshold

Threshold for background noise. Used when filter == TRUE.

euclideanNormThreshold

Euclidean norm threshold for background noise. Used when filter == TRUE.

transformation

Boolean (TRUE / FALSE) to enable data transformation after filtering.

transFun

The function to apply (per sample) for transformation. Typically, for flow and mass cytometry, this is hyperbolic arc sine (asinh(x)). User can supply any function.

asinhFactor

The factor to apply when transforming via asinh(). For flow cytometry, this is usually 150; for mass cytometry and CyTOF, it is 5. Note that this is not used if the user has supplied their own function to transFun.

downsample

Downsample to this number of random variables. This is performed on the final merged dataset, i.e., after all samples have been bound together. NULL to disable.

downsampleVar

Downsample based on variance. Removes this proportion of cells based on lesser variance. This is applied per sample. If user wishes to apply this globally on the final merged dataset, then set this to 0 and remove based on variance manually.

colsDiscard

Columns to be removed from the final merged data. These names are literal and must match exactly.

colsRetain

Retain these columns only. This is the same as colsDiscard but in reverse. Technically, it is possible to activate both colsDiscard and colsRetain, but colsDiscard will be executed first.

newColnames

A named vector of new marker names to assign to each sample. The values of this vector should be the new marker names; the names of this vector should represent the ooriginal marker names. This operation is performed AFTER any operation involving colsDiscard and colsRetain.

emptyValue

boolean (taken from ?flowCore::read.FCS indicating whether or not we allow an empty value for keyword values in TEXT segment.

verbose

Boolean (TRUE / FALSE) to print messages to console or not.

Details

Input, filter, normalise, and transform FCS expression data.

Value

A SingleCellExperiment object.

Author(s)

Kevin Blighe <kevin@clinicalbioinformatics.co.uk>

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
# create random data that follows a negative binomial
mat1 <- jitter(matrix(
  MASS::rnegbin(rexp(50000, rate=.1), theta = 4.5),
  ncol = 20))
colnames(mat1) <- paste0('CD', 1:ncol(mat1))
rownames(mat1) <- paste0('cell', 1:nrow(mat1))

mat2 <- jitter(matrix(
  MASS::rnegbin(rexp(50000, rate=.1), theta = 4.5),
  ncol = 20))
colnames(mat2) <- paste0('CD', 1:ncol(mat2))
rownames(mat2) <- paste0('cell', 1:nrow(mat2))

metadata <- data.frame(
  group = c('PB1', 'PB2'),
  row.names = c('mat1', 'mat2'),
  stringsAsFactors = FALSE)

kevinblighe/scToolkit documentation built on Sept. 25, 2021, 11:29 p.m.