STgradient: STgradient: Tests of gene expression spatial gradients

In FridleyLab/spatialGE: Visualization and Analysis of Spatial Heterogeneity in Spatially-Resolved Gene Expression

STgradient: Tests of gene expression spatial gradients

Description

Calculates Spearman's coefficients to detect genes showing expression spatial gradients

Usage

STgradient(
  x = NULL,
  samples = NULL,
  topgenes = 2000,
  annot = NULL,
  ref = NULL,
  exclude = NULL,
  out_rm = F,
  limit = NULL,
  distsumm = "min",
  min_nb = 3,
  robust = T,
  nb_dist_thr = NULL,
  log_dist = F,
  cores = NULL
)

Arguments

x	an STlist with transformed gene expression
samples	the samples on which the test should be executed
topgenes	the number of high-variance genes to be tested. These genes are selected in descending order of variance as caclulated using Seurat's vst method
annot	the name of a column in ⁠@spatial_meta⁠ containing the tissue domain assignments for each spot or cell. These assignments can be generated using the STclust function
ref	one of the tissue domains in the column specified in annot, corresponding to the "reference" cluster or domain. Spearman's correlations will be calculated using spots assigned to domains other than this reference domain (or domains specified in exclude).
exclude	optional, a cluster/domain to exclude from the analysis
out_rm	logical (optional), remove gene expression outliers defined by the interquartile method. This option is only valid when robust=F
limit	limite the analysis to spots/cells with distances to ref shorther than the value specified here. Useful when gradients might occur at smaller scales or when the domain in ref is scattered through the tissue. Caution must be used due to difficult interpretation of imposed limits. It is suggested to run analysis without restricted distances in addition for comparison.
distsumm	the distance summary metric to use in correlations. One of min or avg
min_nb	the minimum number of immediate neighbors a spot or cell has to have in order to be included in the analysis. This parameter seeks to reduce the effect of isolated ref spots on the correlation
robust	logical, whether to use robust regression (MASS and sfsmisc packages)
nb_dist_thr	a numeric vector of length two indicating the tolerance interval to assign spots/cells to neighborhoods. The wider the range of the interval, the more likely distinct neighbors to be considered. If NULL, c(0.75, 1.25) and c(0.25, 3) is assigned for Visium and CosMx respectively.
log_dist	logical, whether to apply the natural logarithm to the spot/cell distances. It applies to all distances a constant (1e-200) to avoid log(0)
cores	the number of cores used during parallelization. If NULL (default), the number of cores is defined automatically

Details

The STgradient function fits linear models and calculates Spearman coefficients between the expression of a gene and the minimum or average distance of spots or cells to a reference tissue domain. In other wordsm the STgradient function can be used to investigate if a gene is expressed higher in spots/cells closer to a specific reference tissue domain, compared to spots/cells farther from the reference domain (or viceversa as indicated by the Spearman's cofficient).

Value

a list of data frames with the results of the test

FridleyLab/spatialGE documentation built on April 14, 2025, 9:37 a.m.