Skip to contents

Generate the neighbourhoods akin to the miloR approach

rs_make_milor_nhoods.Rd

[Experimental]

Usage

rs_make_milor_nhoods(
  embd,
  knn_indices,
  sample_ids,
  n_samples,
  milor_params,
  seed,
  verbose
)

Arguments

embd

Numeric matrix. Represents the matrix used to generate the kNN graph and will be used to refine the neighbourhoods.

knn_indices

Integer matrix. Each row represents a given cell and the columns the neighbours. (0-indexed!)

sample_ids

Integer vector. 0-indexed(!) sample label per cell, in 0..n_samples. One entry per row of embd.

n_samples

Integer. Number of distinct samples.

milor_params

Named list. Contains the parameters for running the miloR approach.

seed

Integer. Seed for reproducibility.

verbose

Integer. 0L - quiet; 1L - normal verbosity; 2L - detailed verbosity.

Value

A list with the following elements:

  • index_cell - Integer. 0-indexed positions of the cells defining the neighbourhood.

  • nhoods_i - Integer. 0-indexed positions of the cells in the neighbourhood, the index cell included.

  • nhoods_j - Integer. 0-indexed neighbourhood the cell belongs to.

  • nhoods_x - Numeric. The x-value of the COO type matrix, always 1.0.

  • nrows - Integer. Number of cells in the matrix

  • ncols - Integer. Number of refined neighbourhoods.

  • kth_distances - Numeric. Distance of each index cell to its last kNN neighbour, for the spatial FDR.

  • sample_counts - Numeric matrix of neighbourhoods x samples. The cells of each sample found in each neighbourhood.

  • nhood_overlap - Numeric. Cells each neighbourhood shares with all the others, the "graph-overlap" weighting for the spatial FDR.