detectSpatialCorFeats#
Last Updated: Jan 14, 2025
Description#
Detect features that are spatially correlated
Usage#
detectSpatialCorFeats(
gobject,
spat_unit = NULL,
feat_type = NULL,
spat_loc_name = "raw",
method = c("grid", "network"),
expression_values = c("normalized", "scaled", "custom"),
subset_feats = NULL,
spatial_network_name = "Delaunay_network",
network_smoothing = NULL,
spatial_grid_name = "spatial_grid",
min_cells_per_grid = 4,
cor_method = c("pearson", "kendall", "spearman")
)
Arguments#
|
giotto object |
|
spatial unit |
|
feature type |
|
name for spatial locations |
|
method to use for spatial averaging |
|
gene expression values to use |
|
subset of feats to use |
|
name of spatial network to use |
|
smoothing factor beteen 0 and 1 (default: automatic) |
|
name of spatial grid to use |
|
minimum number of cells to consider a grid |
|
correlation method |
Details#
For method = network, it expects a fully connected spatial network. You
can make sure to create a fully connected network by setting minimal_k >
0 in the createSpatialNetwork
function.
1. grid-averaging: average gene expression values within a predefined spatial grid
2. network-averaging: smoothens the gene expression matrix by averaging the expression within one cell by using the neighbours within the predefined spatial network. b is a smoothening factor that defaults to 1 - 1/k, where k is the median number of k-neighbors in the selected spatial network. Setting b = 0 means no smoothing and b = 1 means no contribution from its own expression.
The spatCorObject can be further explored with showSpatialCorFeats()
Value#
returns a spatial correlation object: “spatCorObject”
See Also#
showSpatialCorFeats