Randomized Spatial PCA (RASP): A computationally efficient method for dimensionality reduction of high-resolution spatial transcriptomics data.

Spatial transcriptomics (ST) provides critical insights into the spatial organization of gene expression, enabling researchers to unravel the intricate relationship between cellular environments and biological function. Identifying spatial domains within tissues is key to understanding tissue architecture and mechanisms underlying development and disease progression. Here, we present Randomized Spatial PCA (RASP), a novel spatially-aware dimensionality reduction method for ST data. RASP is designed to be orders-of-magnitude faster than existing techniques, scale to datasets with 100,000+ locations, support flexible integration of non-transcriptomic covariates, and reconstruct de-noised, spatially-smoothed gene expression values. RASP itself is not a clustering or domain detection method; cell types and spatial regions are obtained by clustering the RASP PCs, and the effective cluster resolution depends on the K-nearest-neighbor (kNN) graph and a smoothing parameter β. It employs a randomized two-stage PCA framework and configurable spatial smoothing. RASP was compared to BASS, GraphST, SEDR, SpatialPCA, STAGATE, and CellCharter using diverse ST datasets (10x Visium, Stereo-Seq, MERFISH, 10x Xenium) on human and mouse tissues. In these benchmarks, RASP delivers comparable or superior accuracy in tissue-domain detection while achieving substantial improvements in computational speed. Its efficiency not only reduces runtime and resource requirements but also makes it practical to explore a broad range of spatial-smoothing parameters in a high-throughput fashion. By enabling rapid re-analysis under different parameter settings, RASP empowers users to fine-tune the balance between resolution and noise suppression on large, high-resolution subcellular datasets-a critical capability when investigating complex tissue architecture.