BuildCCInx: Build cell-cell interaction predictions between cell types

In BaderLab/CCInx: Predict cell-cell interaction networks from single-cell data

Build cell-cell interaction predictions between cell types

Description

This function takes a list of gene statistics per cluster to predict cell-cell interactions between each cell-type (cluster). If the GeneStatistic argument is provided, this function will assume the gene statistics represent differential expression between experimental conditions, and will weight the predicted interactions accordingly. Otherwise, predictions will be weighted by expression magnitude per cell type. The output of this function can be explored interactively with ViewCCInx, or static figures can be generated with PlotCCInx.

Usage

BuildCCInx(
  GeneStatList,
  GeneMagnitude = "MeanNormGeneExpr",
  GeneStatistic,
  Species = "hsapiens"
)

Arguments

GeneStatList	A named list of dataframes. Each list element should represent a cell type / cluster to be included in the interaction network, and should be named accordingly. List elements should contain data frames where each row is a gene with official gene symbols as row names. Variables should be appropriately named statistics or annotations to be included in the resulting node metadata. Variable names should be consistent between list elements. The function BuildGeneStatList can be used to generate this list from Seurat or SingleCellExperiment objects when generating predictions not involving differential gene expression.
GeneMagnitude	Default = "MeanNormGeneExpr". A character vector of length 1 representing the variable name in the GeneStatList data frames carrying information on the magnitude (and direction of the change) of expression for the node (gene) in each cell type. This is either a measure of expression (generally mean expression or detection rate) or a measure of change (signed log expression ratio a.k.a. logFC). Default assumes GeneStatList is output from BuildGeneStatList, and uses mean normalized gene expression to weight nodes and edges.
GeneStatistic	Optional. A character vector of length 1 representing the variable name in the GeneStatList data frames carrying information on the statistical significance of expression change. This is generally a corrected p-value.
Species	Default='hsapiens'. The species of the source data. One of 'hsapiens' or 'mmusculus'. Note that the ligand-receptor database was built for human, and the mouse version is generated by homology mapping (only using uniquely mapped homologues).

BaderLab/CCInx documentation built on Jan. 31, 2023, 5:52 p.m.