liana_dotplot: Liana dotplot interactions by source and target cells
In saezlab/liana: LIANA: a LIgand-receptor ANalysis frAmework

liana_dotplot

R Documentation

Liana dotplot interactions by source and target cells

Description

Liana dotplot interactions by source and target cells

Usage

liana_dotplot(
  liana_res,
  source_groups = NULL,
  target_groups = NULL,
  ntop = NULL,
  specificity = "natmi.edge_specificity",
  magnitude = "sca.LRscore",
  y.label = "Interactions (Ligand -> Receptor)",
  size.label = "Interaction\nSpecificity",
  colour.label = "Expression\nMagnitude",
  show_complex = TRUE,
  size_range = c(2, 10),
  invert_specificity = FALSE,
  invert_magnitude = FALSE,
  invert_function = function(x) -log10(x + 1e-10)
)

Arguments

`liana_res`	aggregated 'liana_wrap' results from multiple methods, or alternatively results from running 'liana_wrap' with a single method. Should be filtered by some condition (e.g. preferential consesus ranking, specific interactions, etc).
`source_groups`	names of the source (sender) cell types (NULL = no filter)
`target_groups`	names of the target cell types (NULL = no filter)
`ntop`	number of interactions to return. Note that this assumes that the tibble is sorted in descending order of interaction importance!
`specificity`	column to represent the dot-size of the interaction (by default 'natmi.edge_specificity')
`magnitude`	column to represent interactions expression magnitude (by default 'sca.LRscore')
`y.label`	y label name
`size.label`	size (~specificty) label name
`colour.label`	colour (~magnitude) label name
`show_complex`	logical whether to show complexes (default - TRUE) or only the subunit with minimum expression.

Details

Here, we refer to 'specificity' as how specific this interaction is to a cell type pair regards to the rest of the cell type pairs ( e.g. CellPhoneDB's p-values, NATMI's specificity edges, Connectome's scaled weights, etc)

'magnitude' on the other hand is a direct measure of the expression alone, by default we use SingleCellSignalR's dataset indepent LRscore (bound between 0 and 1). Yet, one could also use CellChat's probabilities or CellPhoneDB's means, etc.