Skip to contents

Remove high correlated cell deconvolution features

Source: R/cell_deconvolution.R
removeCorrelatedFeatures.Rd

If two deconvolution features within a specific cell type are found to be highly correlated, one feature is kept randomly for further analysis.

Usage

removeCorrelatedFeatures(
  data,
  threshold,
  name,
  n_seed,
  corr_method = "spearman",
  batch = NULL
)

Arguments

data

Deconvolution matrix

threshold

Threshold for defined high correlated features

name

Cell type name corresponding to the given matrix in 'data'

n_seed

Seed to ensure reproducibility regarding the choice of the feature.

corr_method

Correlation type whether "spearman" or "pearson".

batch

Optional batch covariate used to compute partial correlations.

Value

A list containing

  • Deconvolution matrix with only one deconvolution feature per high-correlated pair.

  • Highly correlated features found

  • Cell type name