Calculate the (normalised) mixing score for interested cell types

Source: R/mixing_score_summary.R

Produces a data.frame with mixing scores of input reference and target cells from a SpatialExperiment object. It calculates reference-target interactions and reference-reference interactions based on a radius. It derives the mixing score and the normalised mixing score. Function returns NA if the mixing score is being calculated between cells of the same type.

mixing_score_summary(
  spe_object,
  reference_celltype,
  target_celltype,
  radius = 20,
  feature_colname
)

Arguments

spe_object

SpatialExperiment object in the form of the output of format_image_to_spe.

reference_celltype

String Vector. Cell types of the reference cells.

target_celltype

String Vector. Cell types of the target cells.

radius

The maximum radius around a reference cell type for another cell to be considered an interaction.

feature_colname

String specifying the column with the desired cell type annotations.

Value

A data.frame of cell numbers, number of cell interactions, mixing scores, and normalised mixing scores. If there are no reference or target cells found in the image, or there are no reference cells found within the specified radius of any reference cells,the returned (normalised) mixing scores will be NA. If there are no target cells found within the radius of any refernece cells, the returned (normalised) mixing scores will be 0.

Details

The mixing score was originally defined as the number of immune-tumour interactions divided by the number of immune-immune interactions within a defined radius (Keren et al., 2018). The normalised mixing score normalises the immune-tumour interactions and immune-immune interactions within radius by the total number of immune-tumour and immune-immune interactions in the image, respectively. We have generalized this score to allow calculation of any two cell phenotypes defined by the user.

Examples

mixing_score_summary(SPIAT::defined_image, reference_celltype = "Tumour", target_celltype="Immune1",
radius = 50, feature_colname = "Cell.Type")
#>   Reference  Target Number_of_reference_cells Number_of_target_cells
#> 2    Tumour Immune1                       819                    338
#>   Reference_target_interaction Reference_reference_interaction Mixing_score
#> 2                           80                            2513   0.03183446
#>   Normalised_mixing_score
#> 2              0.03852158