Compute local proximity scores — ComputeLPS • pixelatorR

Computes local proximity scores (LPS) for nodes in PNA cell graphs using local_proximity and stores the result on each CellGraph.

Computes local proximity scores for nodes in PNA cell graphs using local_proximity and stores the output on each CellGraph.

Usage

ComputeLPS(object, ...)

# S3 method for class 'CellGraph'
ComputeLPS(
  object,
  markers = NULL,
  method = c("analytical", "permutation"),
  mode = c("self-clustering", "all", "any"),
  iterations = 50L,
  k = 3L,
  A_k = NULL,
  seed = 123,
  name = "lps",
  ...
)

# S3 method for class 'CellGraphList'
ComputeLPS(
  object,
  markers = NULL,
  method = c("analytical", "permutation"),
  mode = c("self-clustering", "all", "any"),
  iterations = 50L,
  k = 3L,
  seed = 123,
  name = "lps",
  verbose = TRUE,
  cl = NULL,
  ...
)

# S3 method for class 'PNAAssay'
ComputeLPS(
  object,
  markers = NULL,
  method = c("analytical", "permutation"),
  mode = c("self-clustering", "all", "any"),
  iterations = 50L,
  k = 3L,
  seed = 123,
  name = "lps",
  verbose = TRUE,
  cl = NULL,
  ...
)

# S3 method for class 'PNAAssay5'
ComputeLPS(
  object,
  markers = NULL,
  method = c("analytical", "permutation"),
  mode = c("self-clustering", "all", "any"),
  iterations = 50L,
  k = 3L,
  seed = 123,
  name = "lps",
  verbose = TRUE,
  cl = NULL,
  ...
)

# S3 method for class 'Seurat'
ComputeLPS(
  object,
  assay = NULL,
  markers = NULL,
  method = c("analytical", "permutation"),
  mode = c("self-clustering", "all", "any"),
  iterations = 50L,
  k = 3L,
  seed = 123,
  name = "lps",
  verbose = TRUE,
  cl = NULL,
  ...
)

Arguments

object: An object
...: Additional parameters passed to other methods
markers: A character vector specifying the markers to use. If NULL, all markers in the count matrix of each CellGraph are used. Methods that iterate over multiple CellGraph objects keep the intersection with available markers. If a graph has no counts or none of the requested markers, a warning is emitted and that graph is left unmodified. The single-graph method still errors when counts are missing.
method: A character string specifying the method to use for computing the local proximity score. Options are "analytical" or "permutation".
mode: A character string specifying the mode of computation. See local_proximity for details. Default is "self-clustering".
iterations: An integer specifying the number of iterations to run when method = "permutation".
k: An integer specifying the neighborhood size to consider.
A_k: An optional pre-computed expanded adjacency matrix. Only used by the CellGraph method.
seed: An integer for random seed setting.
name: Name of the layer (when a matrix is returned) or metadata column (when a vector is returned) used to store scores. Default is "lps".
verbose: Print messages
cl: An integer to indicate number of child-processes (integer values are ignored on Windows) for parallel evaluations. See Details on performance in the documentation for pbapply. The default is NULL, which means that no parallelization is used.
assay: Name of PNAAssay containing the cell graphs to compute local proximity scores

Value

An object with local proximity scores stored on each CellGraph. Matrix results are stored as a layer; vector results are stored in node meta.data.

Details

The default mode is "self-clustering", which returns a node-by-marker matrix that is stored as a layer. Other modes return a single score per node, which is stored in meta.data.

Examples

library(pixelatorR)

se <- ReadPNA_Seurat(minimal_pna_pxl_file(), verbose = FALSE)
#> duckdb keeps downloaded extensions and secrets in a temporary directory:
#> ℹ /tmp/Rtmp3BjCtL/duckdb
#> This is removed when the R session ends.
#> • Extensions are re-downloaded each session.
#> • Secrets are lost.
#> ℹ Run duckdb(shared_home = TRUE) (or create ~/.duckdb) to keep them (suitable for most users).
#> ℹ Run duckdb(shared_home = FALSE) to accept the temporary directory (and silence this message).
#> ℹ See ?duckdb_storage for details and alternatives.
se <- LoadCellGraphs(se, cells = colnames(se)[1], verbose = FALSE)
#> duckdb keeps downloaded extensions and secrets in a temporary directory:
#> ℹ /tmp/Rtmp3BjCtL/duckdb
#> This is removed when the R session ends.
#> • Extensions are re-downloaded each session.
#> • Secrets are lost.
#> ℹ Run duckdb(shared_home = TRUE) (or create ~/.duckdb) to keep them (suitable for most users).
#> ℹ Run duckdb(shared_home = FALSE) to accept the temporary directory (and silence this message).
#> ℹ See ?duckdb_storage for details and alternatives.
cg <- CellGraphs(se)[[1]]

# Matrix result is stored as a layer
cg <- ComputeLPS(cg, markers = "B2M")
#> Warning: 'as(<dgCMatrix>, "ngCMatrix")' is deprecated.
#> Use 'as(., "nMatrix")' instead.
#> See help("Deprecated") and help("Matrix-deprecated").
SeuratObject::Layers(cg)
#> [1] "counts" "lps"   

# Vector result is stored in node metadata
cg <- ComputeLPS(cg, markers = "B2M", mode = "all", name = "lps_b2m")
head(CellGraphData(cg, slot = "meta.data"))
#>      lps_b2m
#> 1  0.3510601
#> 2  0.8001271
#> 3  1.5179853
#> 4  1.3778464
#> 5 -0.6555791
#> 6  1.5591532

# Compute LPS for loaded cell graphs in a PNAAssay
pna_assay <- ComputeLPS(se[["PNA"]], markers = "B2M")
#> ℹ Computing local proximity scores for 1 graph

# Seurat method (only loaded cell graphs are processed)
seur <- ComputeLPS(se, markers = "B2M")
#> ℹ Computing local proximity scores for 1 graph