Computes samples silhouette score from method output — compute_samples

Calculates the samples silhouette width from the results of a dimension reduction method, according to some samples grouping from the samples metadata of a MultiDataSet object.

Usage

compute_samples_silhouette(
  method_output,
  mo_data,
  group_col,
  latent_dimensions = NULL,
  distance_metric = c("euclidean", "maximum", "manhattan", "canberra", "binary",
    "minkowski")
)

Arguments

method_output: method_output Integration method output generated via the get_output() function.
mo_data: A MultiDataSet-class object.
group_col: Character, name of the column in one of the samples metadata table from mo_data containing the samples grouping to be used.
latent_dimensions: Character vector, the latent dimensions to use for computing distance between samples. If NULL (default value), all latent dimensions will be used.
distance_metric: Character, name of the metric to use when computing the distance between samples from their coordinates in the latent dimensions. This will be passed to the stats::dist() function. Options include: "euclidean" (default value), "maximum", "manhattan", "canberra", "binary" or "minkowski".

Value

A list with two elements:

samples_silhouette: a tibble giving for each sample (sample_id column) the group to which it belongs (group column), its closest (other) group in the space spanned by the latent dimensions (neighbour_group), and its silhouette width (silhouette_width column).
groups_average_silhouette: a tibble giving for each samples group (group column) its average silhouette width (group_average_width column).

Details

The samples silhouette width and groups average width are calculated using the cluster::silhouette() function.