sample_random_X: Sample random X

In MarianSchoen/DTD: Digital Tissue Deconvolution

Sample random X

Description

The loss-function learning digital tissue deconvolution approach published by Goertler et al 2018 estimates cell compositions for a given reference matrix X (supervised deconvolution).
Basically, there are two methods to specify the reference profiles in X. Either they are selected using external knowledge (e.g. additional measurements) or they are randomly selected out of the complete data set. The sample_random_X function is an implementation for the second method.

Usage

sample_random_X(
  included.in.X,
  pheno,
  expr.data,
  percentage.of.all.cells = 0.1,
  normalize.to.count = TRUE
)

Arguments

included.in.X	vector of strings, indicating types that will be in the reference matrix
pheno	named vector of strings, with pheno information ('pheno') for each sample in 'expr.data'. names(pheno)' must all be in 'colnames(expr.data)'
percentage.of.all.cells	0 < float < 1, which percentage of all possible cells should be use to generate a cell type profile?
normalize.to.count	logical, normalize each profile?

Details

For each entry of 'included.in.X', 'percentage.of.all.cells' are randomly selected. Then, the reference profile is built by adding up all selected profiles of a type. Afterwards, each reference profile is normalized to a total number of counts.

For examples see the DTD vignette: browseVignettes("DTD")

Value

list with two entries:

• X.matrix: numeric matrix with as many rows as 'expr.data', and as many columns as 'length(included.in.X)'

• samples.to.remove: vector of strings, all samples that have been used for generating X.

MarianSchoen/DTD documentation built on April 29, 2022, 1:59 p.m.