Squidpy Algorithm

Purpose

Use the Squidpy toolkit for comprehensive spatial omics analysis, including cell interaction patterns (neighborhood enrichment), co-occurrence analysis, and evaluation of spatial distribution patterns of cell types (Ripley's statistics).

Usage

SDAS spatialRelate squidpy -i st.h5ad -o outdir --label_key anno_cell2location \
--spatial_coords_scale 0.5 \
--bin_size 50.0 \
--n_neighs 9 \
--coord_type grid \
--n_perms_enrichment 1000 \
--cooccurrence_interval '100,1000,10' \
--n_simulations_ripley 100 \
--ripley_modes 'F,G,L' \
--n_cpus 24 \
--seed 42

Input Parameter Description

Parameter	Required	Default	Description
-i / --input	Yes		Path to the input AnnData h5ad file. The file must contain spatial coordinates (e.g., in adata.obsm['spatial']) and cell type or region cluster annotation (e.g., in adata.obs).
-o / --output	Yes		Output directory for all analysis results and plots. Will be created if it does not exist.
--label_key	Yes		Column name in adata.obs used for cell type or region cluster annotation.
--spatial_coords_scale	No	0.5	Global scaling factor applied to spatial coordinates. For example, set to 0.5 to convert pixel coordinates to micrometers for stereo-seq data.
--min_cells_per_type	No	20	Minimum number of cells (or spots) required for each cell type/cluster to be included in the analysis. Too small a value may lead to unstable results.
--n_cpus	No	8	Number of CPU cores to use for parallel computation. Increasing this value can speed up analysis for large datasets.
--bin_size	No	50.0	Diameter of spots in the original coordinate units. This value, multiplied by spatial_coords_scale, determines the point size in spatial scatter plots.
--n_neighs	No	9	Number of neighbors used to construct the spatial graph (sq.gr.spatial_neighbors). Affects neighborhood enrichment and co-occurrence analysis.
--coord_type	No	grid	Coordinate type for Squidpy's spatial_neighbors. Use 'grid' for regular arrays (faster), 'generic' for irregular spatial data.
--n_perms_enrichment	No	1000	Number of permutations for neighborhood enrichment analysis. Higher values yield more robust statistics but increase computation time.
--cooccurrence_interval	No	"100,1000,10"	Distance interval for co-occurrence analysis, in the format "start,end,num_points". For example, "50,250,10" means analyze 10 evenly spaced distances from 50 to 250 units.
--n_simulations_ripley	No	100	Number of simulations for Ripley's statistics (F, G, L functions). Affects the accuracy of confidence intervals.
--ripley_modes	No	"F,G,L"	Comma-separated list of Ripley statistics modes to calculate. Supported values: "F", "G", "L".
--seed	No	42	Random seed for reproducibility. Must be a non-negative integer.

Notes:

• For sparse data or few cell types, consider increasing n_perms_enrichment and n_simulations_ripley for more stable results.

• Adjust cooccurrence_interval and bin_size based on cell size and spatial distribution for optimal visualization and analysis.

Output Results Display

Result File	Description
<input_name>_squidpy_spatial_scatter.pdf/png	Spatial distribution plot colored by --label_key.
<input_name>_squidpy_nhood_enrichment_zscores.csv	Z-score matrix of neighborhood enrichment analysis.
<input_name>_squidpy_nhood_enrichment.pdf/png	Heatmap visualization of neighborhood enrichment Z-score matrix.
<input_name>_squidpy_co_occurrence_scores_full.csv	Main quantitative result. This file is in long table format, recording the raw co-occurrence score for each cell type pair at each distance bin. Fields include from_celltype, to_celltype, distance, distance_interval_index, cooccurrence_score. Suitable for custom analysis and visualization.
<input_name>_squidpy_co_occurrence_all_types.pdf	Trends of co-occurrence probability for all cell type pairs as distance changes (multi-page).
<input_name>_squidpy_co_occurrence_plots_png/	Individual PNG plots for each cell type's co-occurrence trend.
<input_name>_squidpy_co_occurrence_<celltype>.png	Co-occurrence trend plot for a single cell type.
<input_name>_squidpy_ripley_<mode>_function.pdf/png	Ripley's statistics result plots for each specified mode.
<input_name>_squidpy_processed.h5ad	Final AnnData object after Squidpy analysis.

• Neighborhood enrichment Z-score matrix: <input_name>_squidpy_nhood_enrichment_zscores.csv shows the spatial proximity strength between cell types. Z-score indicates enrichment or depletion compared to random distribution.

	B	Endo	Epi	Fib	Mye	NK	T
B	0.0	-0.5	-1.2	0.8	1.5	2.1	-0.3
Endo	-0.5	0.0	0.7	-0.2	-1.1	0.4	0.9
Epi	-1.2	0.7	0.0	1.8	-0.6	-0.8	0.3
Fib	0.8	-0.2	1.8	0.0	0.5	-0.1	1.2
Mye	1.5	-1.1	-0.6	0.5	0.0	0.7	-0.4
NK	2.1	0.4	-0.8	-0.1	0.7	0.0	1.6
T	-0.3	0.9	0.3	1.2	-0.4	1.6	0.0

• Co-occurrence raw result table: <input_name>_squidpy_co_occurrence_scores_full.csv records the raw co-occurrence score for each cell type pair at each distance bin, suitable for custom analysis and visualization.

from_celltype	to_celltype	distance	distance_interval_index	cooccurrence_score
B	B	100.0	0	0.5327037
B	B	200.0	1	0.5388954
B	B	300.0	2	0.47285548
B	B	400.0	3	0.44997987
B	B	500.0	4	0.41206497
B	Endo	100.0	0	0.12482233
B	Endo	200.0	1	0.1646443
B	Endo	300.0	2	0.15779616
B	Endo	400.0	3	0.20279443
B	Epi	100.0	0	0.06660331

• Neighborhood enrichment heatmap: <input_name>_squidpy_nhood_enrichment.png visualizes the Z-score matrix, color indicates spatial proximity strength.

• Spatial scatter plot: <input_name>_squidpy_spatial_scatter.png shows spatial distribution of cell types colored by type.

• Co-occurrence trend plot: <input_name>_squidpy_co_occurrence_B.png shows co-occurrence probability of B cells with other types as distance changes.

• Ripley's F function analysis: <input_name>_squidpy_ripley_F_function.png shows spatial distribution pattern of cell types compared to random expectation.

Result Interpretation

• Neighborhood enrichment Z-score: Positive means enrichment (cell types tend to be close), negative means depletion (tend to be apart), larger absolute value means more significant.

• Co-occurrence analysis: Shows co-occurrence patterns as distance changes, helps understand spatial interactions.

• Ripley's statistics: Evaluates spatial distribution pattern; L function above random means aggregation, below means dispersion.

Parameter Tuning Suggestions

• If there are few cell types or data is sparse, adjust n_perms_enrichment and n_simulations_ripley for more stable results.

• n_perms_enrichment: Number of permutations for neighborhood enrichment, can be reduced for large datasets to save time.

• n_simulations_ripley: Number of simulations for Ripley's statistics, affects confidence interval accuracy.

• cooccurrence_interval: Distance interval for co-occurrence analysis, adjust based on cell size and spatial distribution.

• bin_size: Affects point size in scatter plot, adjust for data density and visualization needs.