differential_activation: Find Differentialy Activated Features (One vs All)

In vallotlab/ChromSCape: Analysis of single-cell epigenomics datasets with a Shiny App

Find Differentialy Activated Features (One vs All)

Description

Based on the statement that single-cell epigenomic dataset are very sparse, specifically when analysis small bins or peaks, we can define each feature as being 'active' or not simply by the presence or the absence of reads in this feature. This is the equivalent of binarize the data. When trying to find differences in signal for a feature between multiple cell groups, this function simply compare the percentage of cells 'activating' the feature in each of the group. The p.values are then calculated using a Pearson's Chi-squared Test for Count Data (comparing the number of active cells in one group vs the other) and corrected using Benjamini-Hochberg correction for multiple testing.

Usage

differential_activation(
  scExp,
  by = c("cell_cluster", "sample_id")[1],
  verbose = TRUE,
  progress = NULL
)

Arguments

scExp	A SingleCellExperiment object containing consclust with selected number of cluster.
by	Which grouping to run the marker enrichment ?
verbose	Print ?
progress	A shiny Progress instance to display progress bar.

Details

To calculate the logFC, the percentage of activation of the features are corrected for total number of reads to correct for library size bias. For each cluster ('group') the function consider the rest of the cells as the reference.

Value

Returns a dataframe of differential activation results that contains the rowData of the SingleCellExperiment with additional logFC, q.value, group activation (fraction of cells active for each feature in the group cells), reference activation (fraction of cells active for each feature in the reference cells).

See Also

For Pearson's Chi-squared Test for Count Data chisq.test. For other differential analysis see differential_analysis_scExp.

Examples

data("scExp")
res = differential_activation(scExp, by = "cell_cluster")
res = differential_activation(scExp, by = "sample_id")

vallotlab/ChromSCape documentation built on Oct. 15, 2023, 1:47 p.m.