Load CellGraphs — LoadCellGraphs • pixelatorR

Loads a list of CellGraph objects.

Usage

LoadCellGraphs(object, ...)

# S3 method for class 'FileSystemDataset'
LoadCellGraphs(
  object,
  cells,
  load_as = c("bipartite", "Anode", "linegraph"),
  add_marker_counts = TRUE,
  chunk_size = 10,
  verbose = TRUE,
  ...
)

# S3 method for class 'tbl_df'
LoadCellGraphs(
  object,
  cells,
  load_as = c("bipartite", "Anode", "linegraph"),
  add_marker_counts = TRUE,
  verbose = TRUE,
  ...
)

# S3 method for class 'MPXAssay'
LoadCellGraphs(
  object,
  cells = colnames(object),
  load_as = c("bipartite", "Anode", "linegraph"),
  add_marker_counts = TRUE,
  load_layouts = FALSE,
  force = FALSE,
  chunk_size = 10,
  cl = NULL,
  verbose = TRUE,
  ...
)

# S3 method for class 'CellGraphAssay'
LoadCellGraphs(
  object,
  cells = colnames(object),
  load_as = c("bipartite", "Anode", "linegraph"),
  add_marker_counts = TRUE,
  load_layouts = FALSE,
  force = FALSE,
  chunk_size = 10,
  cl = NULL,
  verbose = TRUE,
  ...
)

# S3 method for class 'CellGraphAssay5'
LoadCellGraphs(
  object,
  cells = colnames(object),
  load_as = c("bipartite", "Anode", "linegraph"),
  add_marker_counts = TRUE,
  load_layouts = FALSE,
  force = FALSE,
  chunk_size = 10,
  cl = NULL,
  verbose = TRUE,
  ...
)

# S3 method for class 'Seurat'
LoadCellGraphs(
  object,
  assay = NULL,
  cells = colnames(object),
  load_as = c("bipartite", "Anode", "linegraph"),
  add_marker_counts = TRUE,
  load_layouts = FALSE,
  force = FALSE,
  chunk_size = 10,
  cl = NULL,
  verbose = TRUE,
  ...
)

Arguments

object: A Seurat object or an CellGraphAssay object
...: Parameters passed to other methods
cells: A character vector of cell names to load CellGraphs for
load_as: Choose how the cell graph should be represented (see details below)
add_marker_counts: Should marker counts be added to the CellGraph objects?
chunk_size: Length of chunks used to load CellGraphs from edge list.
verbose: Print messages
load_layouts: Load layouts from the PXL file if available.
force: Force load graph(s) if they are already loaded
cl: A cluster object created by makeCluster, or 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: Assay name

Value

An object with a list of CellGraph objects

Details

Graphs can be loaded as one of 'bipartite', 'Anode' or 'linegraph'. See details about each of these graph representations in the sections below.

LoadCellGraphs will only work if the PXL file path(s) stored in the object are valid. You can check the PXL file paths stored in a Seurat, a CellGraphAssay or a CellGraphAssay5 object with the FSMap method. If the PXL file(s) are invalid, an error will be thrown. Visit RestorePaths for more information on how to update the PXL file paths.

Bi-partite graph

In the bi-partite graph, edges can only go from a upia to a upib. The bi-partite graph is first collapsed from a multigraph to a simple graph by aggregating marker counts for each upia/upib combination. For visualization of marker counts on the graph, it is often convenient to project values on the nodes; however, the marker counts are not available in the nodes in the bi-partite graph. To circumvent this issue, node counts are calculated by aggregating its edge counts. This means that the total marker count will be inflated.

A node projected-graph

In the A node projected-graph, pairs of upias that share a upib are connected with an edge. The (upia) node counts are calculated by aggregating counts across all edges associated with the A nodes. The number of nodes in the A node projected-graph is substantially lower than the bi-partite graph.

Line graph

Starting with a bi-partite graph, a node is placed on each edge. Then, an edge is drawn between each pair of adjacent nodes. The number of nodes in the linegraph is equivalent to the number of edges in the bi-partite graph. However, the number of edges is substantially larger. One major advantage with the linegraph is that the node counts represent the actual raw data, which is not the case for the bi-partite graph and the A node projected-graph. On the down side, linegraphs are much larger and tends to slow down layout computations and visualizations.