ripleys_k: Calculate Ripley's K

In spatialTIME: Spatial Analysis of Vectra Immunoflourescent Data

Calculate Ripley's K

Description

ripleys_k() calculates the emperical Ripley's K measurement for the cell types specified by mnames in the mIF object. This is very useful when exploring the spatial clustering of single cell types on TMA cores or ROI spots following proccessing with a program such as HALO for cell phenotyping.

In the 'ripleys_k' function, there is the ability to perform permutations in order to assess whether the clustering of a cell type is significant, or the ability to derive the exact CSR and forgo permutations for much faster sample processing. Permutations can be helpful if the significance of clustering wasnts to be identified - run 1000 permutations and if observed is outside 95-percentile then significant clustering. We, however, recommend using the exact CSR estimate due to speed.

Some things to be aware of when computing the exact Ripley's K estimate, if your spatial file is greater than the ‘big' size, the edge correction will be converted to ’none' in order to save on resources and compute time. Due to the introduction of Whole Slide Imaging (WSI), this can easily be well over 1,000,000 cells, and calculating edge correction for these spatial files will not succeed when attempting to force an edge correction on it.

Usage

ripleys_k(
  mif,
  mnames,
  r_range = seq(0, 100, 1),
  num_permutations = 50,
  edge_correction = "translation",
  method = "K",
  permute = FALSE,
  keep_permutation_distribution = FALSE,
  workers = 1,
  overwrite = FALSE,
  xloc = NULL,
  yloc = NULL,
  big = 10000
)

Arguments

mif	object of class 'mif' created with 'create_mif'
mnames	cell phenotype markers to calculate Ripley's K for
r_range	radius range (including 0)
num_permutations	number of permutations to use to estimate CSR. If 'keep_perm_dis' is set to FALSE, this will be ignored
edge_correction	edge correction method to pass to 'Kest'. can take one of "translation", "isotropic", "none"
method	not used currently
permute	whether to use CSR estimate or use permutations to determine CSR
keep_permutation_distribution	whether to find mean of permutation distribution or each permutation calculation
workers	number of cores to use for calculations
overwrite	whether to overwrite the 'univariate_Count' slot within 'mif$derived'
xloc	the location of the center of cells. If left 'NULL', 'XMin', 'XMax', 'YMin', and 'YMax' must be present.
yloc	the location of the center of cells. If left 'NULL', 'XMin', 'XMax', 'YMin', and 'YMax' must be present.
big	the number of cells at which to flip from an edge correction method other than 'none' to 'none' due to size

Value

object of class 'mif'

Examples

x <- spatialTIME::create_mif(clinical_data =spatialTIME::example_clinical %>% 
  dplyr::mutate(deidentified_id = as.character(deidentified_id)),
  sample_data = spatialTIME::example_summary %>% 
  dplyr::mutate(deidentified_id = as.character(deidentified_id)),
  spatial_list = spatialTIME::example_spatial,
  patient_id = "deidentified_id", 
  sample_id = "deidentified_sample")
mnames = x$spatial[[1]] %>%
  colnames() %>%
  grep("Pos|CD", ., value =TRUE) %>%
  grep("Cyto|Nucle", ., value =TRUE, invert =TRUE)
x2 = ripleys_k(mif = x, 
  mnames = mnames[1], 
  r_range = seq(0, 100, 1), 
  num_permutations = 100,
  edge_correction = "translation", 
  method = "K", 
  permute = FALSE,
  keep_permutation_distribution =FALSE, 
  workers = 1, 
  overwrite =TRUE)

spatialTIME documentation built on April 1, 2023, 12:18 a.m.