ClustSIGNAL_analysis2025

This repository contains code for data analysis and figures reported in the manuscript "ClustSIGNAL identifies cell types and subtypes using an adaptive smoothing approach for scalable spatial clustering". To access the method code and tutorials, see CLustSIGNAL repository.

Data analysis

• Simulation analysis - see Simulation_scripts folder.

  • Simulated_dataAnalysis: simulated data generation, parameter testing, benchmarking adaptive smoothing, stress testing - sparsity and segmentation errors, assessing clustering stability.

  • simulation_functions: functions supporting simulation analysis.

• Benchmarking analysis - see Benchmarking_scripts folder.

  • Benchmarking_dataAnalysis: benchmarking runs 4 methods (ClustSIGNAL, BANKSY, BASS, SpatialPCA) on 4 real-world datasets (SeqFISH mouse embryo, Xenium breast cancer, CosMx lung cancer, MERFISH mouse hypothalamus).

  • benchmarking_functions: wrapper functions for methods.

• Main figures - see Figures folder.

  • Fig2_ simulations: simulation figure panels.

  • Fig3_benchmarking: benchmarking figure panels.

  • Fig4_dataAnalysis: data analysis figure panels, and related analyses.

• Supplementary materials - see Supplementary folder.

  • Suppl_analysis: entropy analysis.

  • Suppl_figs: supplementary figures.