The paper introduces SpatialZoomer, a novel computational framework designed for the multi-scale analysis of single-cell resolution spatial transcriptomics (ST) data. This spectral graph-based method utilizes a set of low-pass filters to effectively extract spatial molecular features at various resolutions, including the single-cell, niche, and domain scales. A key innovation is the automatic identification of "critical" scales by partitioning a cross-scale similarity map using dynamic programming. The authors demonstrate SpatialZoomer's utility through applications such as detecting disease-progression signals in Alzheimer’s Disease and identifying spatially dependent cell subtypes and complex tissue architectures, like the stress-proliferation-EMT tri-layer organization in cancer. Overall, the tool offers a scalable, computationally efficient approach to understanding how molecular signals and spatial context shape cellular heterogeneity.

References:

• Li X, Fan Y, Han Y, et al. SpatialZoomer: multi-scale feature analysis of spatial transcriptomics[J]. bioRxiv, 2025: 2025.09. 08.674870.