First layer wrapper function to build linear models measuring state changes

Builds linear models measuring marker based state changes in a cell type based of the proximity or abundance of another cell type. The function provides the option to build robust and mixed linear model variants

Usage

calcStateChanges(
  cells,
  marker = NULL,
  from = NULL,
  to = NULL,
  image = NULL,
  type = "distances",
  assay = 1,
  cellType = "cellType",
  imageID = "imageID",
  contamination = NULL,
  minCells = 20,
  verbose = FALSE,
  timeout = 10,
  nCores = 1
)

Arguments

cells

A dataframe with a imageID, cellType, and marker intensity column along with covariates (e.g. distance or abundance of the nearest cell type) to model cell state changes

marker

A vector of markers that proxy a cell's state. If NULL, all markers will be used.

from

A vector of cell types to use as the primary cells. If NULL, all cell types will be used.

to

A vector of cell types to use as the interacting cells. If NULL, all cell types will be used.

image

A vector of images to filter to. If null all images will be used.

type

What type of state change. This value should be in reduced dimensions.

assay

The assay in the SingleCellExperiment object that contains the marker expressions.

cellType

The column in colData that stores the cell types.

imageID

The column in colData that stores the image ids.

contamination

If TRUE, use the contamination scores that have previously been calculate. Otherwise a name of which reduced dimension contains the scores.

minCells

The minimum number of cells required to fit a model.

verbose

A logical indicating if messages should be printed

timeout

A maximum time allowed to build each model. Setting this may be important when building rlm mixed linear models

nCores

Number of cores for parallel processing

Examples

library(dplyr)
#> 
#> Attaching package: ‘dplyr’
#> The following objects are masked from ‘package:stats’:
#> 
#>     filter, lag
#> The following objects are masked from ‘package:base’:
#> 
#>     intersect, setdiff, setequal, union
data("kerenSCE")

kerenSCE <- kerenSCE[, kerenSCE$imageID %in% c(5,6)]

kerenSCE <- getDistances(kerenSCE,
  maxDist = 200,
)

imageModels <- calcStateChanges(
  cells = kerenSCE,
  from = "Macrophages",
  to = "Tumour"
)