plotSignatures: Find enriched markers per identified cluster and visualise...
In kevinblighe/scToolkit: scDataviz: single cell dataviz and downstream analyses

Description Usage Arguments Details Value Author(s) Examples

Find enriched markers per identified cluster and visualise these as a custom corrplot.

plotSignatures(
  indata,
  assay = "scaled",
  clusterAssign = metadata(indata)[["Cluster"]],
  funcSummarise = function(x) mean(x, na.rm = TRUE),
  col = colorRampPalette(brewer.pal(9, "RdPu"))(100),
  labCex = 1,
  legendPosition = "right",
  legendCex = 1,
  labDegree = 90,
  verbose = TRUE
)

`indata`	A data-frame or matrix, or `SingleCellExperiment` object. If a data-frame or matrix, this should relate to expression data (cells as columns; genes as rows). If a `SingleCellExperiment` object, data will be extracted from an assay component named by `assay`.
`assay`	Name of the assay slot in `indata` from which data will be taken, assuming `indata` is a `SingleCellExperiment` object.
`clusterAssign`	A vector of cell-to-cluster assignments. This can be from any source but must align with your cells / variables. There is no check to ensure this when `indata` is not a `SingleCellExperiment` object.
`funcSummarise`	A mathematical function used to summarise expression per marker, per cluster.
`col`	colorRampPalette to be used for shading low-to-high expression.
`labCex`	cex (size) of the main plot labels.
`legendPosition`	position of legend. Can be one of 'top', 'right', 'bottom', 'left'
`legendCex`	cex (size) of the legend labels.
`labDegree`	Rotation angle of the main plot labels.
`verbose`	Boolean (TRUE / FALSE) to print messages to console or not.

Find enriched markers per identified cluster and visualise these as a custom corrplot. plotSignatures first collapses your input data's expression profiles from the level of cells to the level of clusters based on a mathematical function specified by funcSummarise. It then centers and scales the data range to be between -1 and +1 for visualisation purposes.

A corrplot object.

Kevin Blighe <kevin@clinicalbioinformatics.co.uk>

# create random data that follows a negative binomial
mat <- jitter(matrix(
  MASS::rnegbin(rexp(1000, rate=.1), theta = 4.5),
  ncol = 20))
colnames(mat) <- paste0('CD', 1:ncol(mat))
rownames(mat) <- paste0('cell', 1:nrow(mat))

u <- umap::umap(mat)$layout
colnames(u) <- c('UMAP1','UMAP2')
rownames(u) <- rownames(mat)
clus <- clusKNN(u)

plotSignatures(t(mat), clusterAssign = clus)