Tests differential expression/accessibility using logistic regression

This function implements the test proposed by Ntranos et. al. (doi: 10.1038/s41592-018-0303-9) for SingleCellExperiment objects. Compared to the traditional approach of using cell labels as covariate, logistic regression is carried out to predict cell labels from quantification of associated features. Those features can be transcripts or accessibility peaks associated with e.g. genes or regulatory regions.

lrde_test(object, response, exprs_values = "logcounts",
  group_by = NULL, reference = NULL, features = NULL, ...)

Arguments

object	A SingleCellExperiment object.
response	Character vector indicating the response (class) variable. In case of length of one, it should determines the column name of the colData slot.
exprs_values	String indicating which assay contains the data that should be used to perform grouping and testing.
group_by	Character vector indicating the grouping variable for expression values. In case of length of one, it should determines the column name of the rowData slot.
reference	The level of response that should be used as reference.
features	A character vector (of feature names), a logical vector or numeric vector (of indices) specifying the features to test. The default of NULL will use all features.
...	Additional parameters passed to p.adjust for multiple testing correction.

Value

A tibble with the following columns: - group - the grouping variable - pvalue - the p-value from comparison of the two LR models - p.adjusted - the multiple testing corrected pvalue.