Benchmark and processed datasets for LARIS: Ligand And Receptor Interaction in Spatial transcriptomics data

This repository contains benchmark and processed datasets for LARIS. It includes simulated spatial transcriptomics datasets for benchmarking, processed outputs from human tonsil Slide-tags data, and processed outputs from developing mouse cortex Stereo-seq data. For details, see the accompanying manuscript: https://www.biorxiv.org/content/10.1101/2025.11.26.690796v1

Simulated spatial transcriptomics benchmark datasets

brain_stem_adata_simulated_spatial_27_s4_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset based on a brainstem-inspired tissue structure. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance on simulated spatial transcriptomics data.

sccube_default_adata_simulated_spatial_29_s6_20000_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset generated using the default scCube-style synthetic tissue structure with 20,000 cells. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance and scalability.

hair_follicle_adata_simulated_spatial_30_s7_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset based on a hair follicle-inspired tissue structure. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance on simulated spatial transcriptomics data.

sccube_default_adata_simulated_spatial_29_s6_5000_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset generated using the default scCube-style synthetic tissue structure with 5,000 cells. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance and scalability.

lymph_node_adata_simulated_spatial_33_s10_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset based on a lymph node-inspired tissue structure. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance on simulated spatial transcriptomics data.

sccube_stripes_adata_simulated_spatial_25_s2_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset generated using a stripe-like synthetic tissue structure, where tissue regions are arranged in parallel with partial overlap in cell-type composition. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance on simulated spatial transcriptomics data.

mouse_lung_adata_simulated_spatial_32_s9_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset based on a mouse lung-inspired tissue structure. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance on simulated spatial transcriptomics data.

small_intestine_adata_simulated_spatial_31_s8_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset based on a small intestine-inspired tissue structure. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance on simulated spatial transcriptomics data.

sccube_default_adata_simulated_spatial_29_s6_10000_recligassigned.h5ad
Simulated spatial transcriptomics benchmark dataset generated using the default scCube-style synthetic tissue structure with 10,000 cells. The dataset contains simulated gene expression, spatial coordinates, and randomly assigned ligand and receptor gene identities. It was used for benchmarking LARIS performance and scalability.

Processed LARIS ligand-receptor interaction outputs for human tonsil Slide-tags data

adata_tonsil.h5ad
Processed AnnData object for the human tonsil Slide-tags dataset. The file contains gene count data only, with annotated cell types, including marginal reticular cells and follicular dendritic cells, for downstream analysis and comparison with LARIS ligand-receptor interaction scores.

lr_adata_combined_annotated_FDC_MRC_subset.h5ad
Processed LARIS output for human tonsil Slide-tags data, subset to follicular dendritic cells and marginal reticular cells. The file contains gene count data together with calculated LARIS ligand-receptor interaction scores for the FDC/MRC subset.

lr_adata_combined_tonsil.h5ad
Processed LARIS output for the full human tonsil Slide-tags dataset. The file contains gene count data together with calculated LARIS ligand-receptor interaction scores for the tonsil dataset.

Processed LARIS ligand-receptor interaction outputs for developing mouse cortex Stereo-seq data

adata_Cortex_Whole_Brain_E14.5_E16.5_P7_P14_P77_linear.h5ad
Processed AnnData object for developing mouse brain Stereo-seq data across E14.5, E16.5, P7, P14, and P77 time points. The file contains gene count data for cortex and whole-brain regions arranged in a linearized spatial coordinate layout, where developmental time points are placed next to each other in non-overlapping coordinate regions for joint visualization and analysis.

lr_only_cortex_annotated_E14.5_E16.5_P7_P14_P77.h5ad
Processed LARIS output for annotated developing mouse cortex Stereo-seq data across E14.5, E16.5, P7, P14, and P77 time points. The file contains calculated LARIS ligand-receptor interaction scores only, with cortical regions and layers annotated.

adata_only_cortex_annotated_E14.5_E16.5_P7_P14_P77.h5ad
Processed AnnData object for annotated developing mouse cortex Stereo-seq data across E14.5, E16.5, P7, P14, and P77. The file contains gene count data only, with cortical regions and layers annotated for downstream analysis and comparison with LARIS ligand-receptor interaction scores.

Files for ligand-receptor interaction database

mouse_lr_database_CellChatDB_formatted_v2.csv
Formatted mouse ligand-receptor interaction database derived from CellChatDB and used as input for LARIS. Multimeric ligand and receptor complexes were split into individual pairwise ligand-receptor entries so that each row represents a single ligand-receptor interaction. This database was used for the developing mouse brain and cortex Stereo-seq analyses.

human_lr_database_CellChatDB_formatted_v2.csv
Formatted human ligand-receptor interaction database derived from CellChatDB and used as input for LARIS. Multimeric ligand and receptor complexes were split into individual pairwise ligand-receptor entries so that each row represents a single ligand-receptor interaction. This database was used for the human tonsil Slide-tags analyses.