SpatialDM (Spatial Direct Messaging, or Spatial co-expressed ligand and receptor Detected by Moran's bivariant extension), a statistical model and toolbox to identify the spatial co-expression (i.e., spatial association) between a pair of ligand and receptor.
Uniquely, SpatialDM can distinguish co-expressed ligand and receptor pairs from spatially separating pairs, and identify the spots of interaction.
With the analytical testing method, SpatialDM is scalable to 1 million spots within 12 min with only one core.
It comprises two main steps: 1) global selection spatialdm_global to identify significantly interacting LR pairs; 2) local selection spatialdm_local to identify local spots for each interaction.
SpatialDM is available through PyPI.
To install, type the following command line and add -U for updates:
pip install -U SpatialDMAlternatively, you can install from this GitHub repository for latest (often development) version by the following command line:
pip install -U git+https://github.com/leeyoyohku/SpatialDMInstallation time: < 1 min
Using the build-in melanoma dataset as an example, the following Python script will compute the p-value indicating whether a certain Ligand-Receptor is spatially co-expressed.
import spatialdm as sdm
import spatialdm.plottings as pl
adata = sdm.datasets.melanoma()
raw = pd.DataFrame(adata.raw.X, index=adata.obs_names, columns=adata.var_names)
log = pd.DataFrame(adata.X, index=adata.obs_names, columns=adata.var_names)
spatialcoord = pd.DataFrame(adata.obsm['spatial'], index=adata.obs_names, columns=['x','y'])
# Preprocessing
my_sample = sdm.SpatialDM(log, raw, spatialcoord) # load spatial data with simply log, raw, spatial input
my_sample.extract_lr(species='human', min_cell=3)
my_sample.weight_matrix(l=1.2, cutoff=0.2, single_cell=False) # Not single-cell resolution
# Global selection of significant pairs
my_sample.spatialdm_global(1000, select_num=None, method='permutation') # complete in seconds
my_sample.sig_pairs(method='permutation', fdr=True, threshold=0.1) # select significant pairs
pl.global_plot(my_sample, pairs=['CSF1_CSF1R']) # Overview of global selection
# Local selection of significant spots
my_sample.spatialdm_local(n_perm=1000, method='both', select_num=None, nproc=1) # local spot selection complete in seconds
my_sample.sig_spots(method='permutation', fdr=False, threshold=0.1) # significant local spots
pl.plot_pairs(my_sample, ['CSF1_CSF1R'], marker='s') # visualize known melanoma pair(s)
The full manual is at https://spatialdm.readthedocs.io, including:
- Permutation-based SpatialDM (Recommended for small datasets, <10k spots).
- Analytical z-score-based SpatialDM.
- Differential analyses of whole interactome among varying conditions.
SpatialDM manuscript with more details is available on bioRxiv now and is currently under review.