GenerateCCIM: Calculate cell-cell interaction matrix

In BlishLab/scriabin: Single-cell resolved interaction analysis through binning

Calculate cell-cell interaction matrix

Description

Calculate cell-cell interaction matrix

Usage

GenerateCCIM(
  object,
  assay = "SCT",
  slot = "data",
  species = "human",
  database = "OmniPath",
  ligands = NULL,
  recepts = NULL,
  senders = NULL,
  receivers = NULL,
  weighted = F,
  nichenet_results = NULL,
  pearson.cutoff = 0.075,
  scale.factors = c(1.5, 3),
  weight.method = "sum"
)

Arguments

object	A seurat object
assay	Assay in Seurat object from which to pull expression values
slot	Slot within assay from which to pull expression values
species	character. Name of species from which to load ligand-receptor databases. One of: "human", "mouse", "rat". Default: "human"
database	Name of ligand-receptor database to use. Default: "OmniPath" When species is "human", one of: OmniPath, CellChatDB, CellPhoneDB, Ramilowski2015, Baccin2019, LRdb, Kirouac2010, ICELLNET, iTALK, EMBRACE, HPMR, Guide2Pharma, connectomeDB2020, talklr, CellTalkDB When species is "mouse" or "rat", only "OmniPath" is supported. To pass a custom ligand-receptor database to this function, set database = "custom"
ligands	Character vector of custom ligands to use for interaction graph generation. Ignored unless database = "custom" When ligands is supplied, recepts must also be supplied and equidimensional.
recepts	Character vector of custom receptors to use for interaction graph generation. Ignored unless database = "custom" When recepts is supplied, ligands must also be supplied and equidimensional.
senders	Character vector of cell names to be used as sender cells for interaction graph generation. By default, all cells in object are used as senders.
receivers	Character vector of cell naems to be used as receiver cells for interaction graph generation. By default, all cells in object are used as receivers.
weighted	logical. Weight the cell-cell interaction graph by ligands predicted to be active by NicheNet? Default: FALSE
nichenet_results	List or matrix. Predicted ligand activities using Scriabin's implementation of NicheNet in RankActiveLigands
pearson.cutoff	numeric. Threshold for determining which ligand activities are "active". Ligands below this threshold will be considered inactive and not used for weighting. Default: 0.075
scale.factors	numeric. Determines the magnitude of ligand and receptor expression values weighting by their predicted activities. Given scale factors of c(x,y), ligand-receptor pairs where the ligand's pearson = pearson.cutoff will be weighted by a factor of x, and the ligand with the highest pearson will be weighted by a factor of y. Default: c(1.5,3).
weight.method	One of "sum" or "product". Method to use for weighting CCIM by ligand activities. "Sum" sums receptor expression values with ligand activities passing the pearson.cutoff threshold. Thus, a zero value for receptor expression may still result in a positive value for the ligand-receptor pair mechanism. "Product" takes the product of the receptor expression value and the predicted ligand activities. A zero value in the receptor expression will always result in a zero value for that ligand-receptor pair.

Value

Returns a Seurat object with assay "CCIM" where columns are cell-cell pairs and rows are ligand-receptor pairs. "Expression" values are the geometric mean expression value between each pair of sender and receiver cells. By default, names ligand-receptor pairs and cell-cell pairs are separated by "="

References

Browaeys, et al. Nature Methods (2019)

BlishLab/scriabin documentation built on Sept. 16, 2024, 1:19 a.m.