SimSpace: a comprehensive in-silico spatial omics data simulation framework

Tissue function is tightly linked to the spatial organization, and our knowledge of the relationship between tissue structure and function has increased due to the rapid technological development of spatial transcriptomics approaches. A bottle-neck is, however, the analysis of the complex and high-dimensional spatial transcriptomics data. Many computational tools have been proposed for spatial analysis but evaluating them remains challenging due to the scarcity of datasets with known ground truth. To address this, we developed SimSpace , a flexible simulation framework that can generate synthetic spatial cell maps with categorical cell type labels and biologically meaningful organization. Cell type spatial patterns are simulated using a Markov Random Field model, enabling the control of spatial autocorrelation and interaction between cell types. This approach captures a broad range of tissue architectures, from well-separated niches to spatially mixed environments. We evaluate the simulated data using spatial statistics metrics (eg. entropy, Moran’s I) and benchmark the results against real spatial transcriptomics and proteomics datasets. Our findings show that the simulations closely mirror key spatial properties observed in biological tissues. Additionally, we demonstrate the utility of our framework as a testbed for benchmarking spatial gene detection and cell type deconvolution methods. By providing reproducible, ground-truth-controlled datasets, SimSpace supports the development, validation, and comparison of computational methods in spatial omics.