feature_correlation: Get features whose expression is correlated or...

In Close-your-eyes/scexpr: Lots of helpers and wrappers to analyze scRNAseq data sets

Get features whose expression is correlated or anti-correlated across all cells or groups of cells

Description

Find out the expression of which features is correlated or anti-correlated using simple correlation metrics like pearson or spearman. The analysis may be applied to a subset of cells or subset of features (see arguments). Due to dropouts in some scRNAseq technologies this analysis is not super-clean but may still provide a valid, relative, comparison of feature correlations. Other methods considering the dropout-effect exist: COTAN.

Usage

feature_correlation(
  SO,
  features,
  assay = c("RNA", "SCT"),
  method = c("pearson", "spearman", "kendall"),
  cells = NULL,
  min.pct = 0.1,
  limit_p = 1e-303,
  bar.fill = c("correlation_sign", "ref_feature_pct", "none"),
  theme = ggplot2::theme_bw(),
  split.by = NULL,
  min.group.size = 20,
  topn = c(10, 10),
  ...
)

Arguments

SO	Seurat object
features	which features to calculate correlations for (must be rownames in the selected assay)
assay	which assay to obtain expression values from; the data slot will be used in any case
method	which metric of correlation to calculate
cells	vector of cell names to consider for correlation anaylsis; if NULL (default) all cells are used
min.pct	minimum percentage of expressing cells (> 0 UMIs) to include a feature in correlation analysis
limit_p	p-value which p-values of 0 will be set to; this avoids obtaining INF when deriving -log10(p-val)
bar.fill	which bar fill to apply
theme	which ggplot theme to set as basis
split.by	groups for correlation analysis; must be a categorical column in meta.data of SO
min.group.size	required number of cell in one group to be considered in analysis
topn	numeric vector of length two; number of top anti-correlated and correlated features, respectively
...	additional arguments to psych::corr.test and to ggplot2::theme

Value

a default plot (ggplot2 object) and the underlying data frame with correlation values

Close-your-eyes/scexpr documentation built on April 21, 2023, 10:27 a.m.