Skip to contents

Compute LISI scores on a Seurat object

Source: R/LISI.R
RunLISI.Rd

Compute per-cell Local Inverse Simpson's Index (LISI) scores from a dimensional reduction and store them in the meta.data and tools slots of a Seurat object.

Usage

RunLISI(
  srt,
  reductions = NULL,
  reduction = NULL,
  dims = NULL,
  label_colnames = NULL,
  prefix = NULL,
  tool_name = NULL,
  perplexity = 30,
  tol = 1e-05,
  max_iter = 50,
  overwrite = TRUE,
  verbose = TRUE
)

Arguments

srt

A Seurat object.

reductions

Character vector of dimensional reductions used to compute LISI. If NULL, DefaultReduction() is used.

reduction

Deprecated alias of reductions.

dims

Dimensions to use from the reduction. Default is NULL, which uses all available dimensions.

label_colnames

Character vector of metadata columns used for LISI. If NULL, RunLISI() will try to use srt@misc[["integration_batch"]].

prefix

Prefix used for the stored LISI metadata columns. If NULL, the reduction names are used.

tool_name

Name used to store detailed results in srt@tools. Default is "LISI" when multiple reductions are provided, otherwise paste0(prefix, "_LISI").

perplexity

Effective neighborhood size. Default is 30.

tol

Tolerance used in the binary search for the target perplexity. Default is 1e-5.

max_iter

Maximum number of binary-search iterations. Default is 50.

overwrite

Whether to overwrite existing metadata columns. Default is TRUE.

verbose

Whether to print the message. Default is TRUE.

Value

A modified Seurat object.

See also

thisutils::compute_lisi, LISIPlot

Examples

data(panc8_sub)
panc8_sub <- integration_scop(
  panc8_sub,
  batch = "tech",
  integration_method = "Harmony5"
)
names(panc8_sub@reductions)

panc8_sub <- RunLISI(
  panc8_sub,
  reductions = c("pcaUMAP2D", "Harmony5UMAP2D")
)
LISIPlot(
  panc8_sub,
  combine = TRUE
)