Previously: calculate_spatial_genes_python. This method computes a silhouette score per gene based on the
spatial distribution of two partitions of cells (expressed L1, and non-expressed L0).
Here, rather than L2 Euclidean norm, it uses a rank-transformed, exponentially weighted
function to represent the local physical distance between two cells.
New multi aggregator implementation can be found at silhouetteRankTest
silhouetteRank(
gobject,
expression_values = c("normalized", "scaled", "custom"),
metric = "euclidean",
subset_genes = NULL,
rbp_p = 0.95,
examine_top = 0.3,
python_path = NULL,
return_gobject = F
)
giotto object
expression values to use
distance metric to use
only run on this subset of genes
fractional binarization threshold
top fraction to evaluate with silhouette
specify specific path to python if required
data.table with spatial scores