Spatial transcriptomics (ST) technologies have provided us with unique opportunities for understanding how cell populations interact in situ and how their interactions may contribute towards disease progression. The proliferation of various ST platforms has generated a wealth of data with differing spatial resolutions and throughputs. However, a key challenge persists: the lack of a unified workflow for identifying cell states across different ST platforms, which hampers biological discoveries.
We introduce Φ-Space ST, a computationally efficient and platform-agnostic method for identifying continuous cell states in ST data using multiple scRNA-seq references. For ST data with super-cellular resolutions (e.g. 10x Visium, Slide-seqV2), Φ-Space ST provides reliable cell type deconvolution that matches the performance of state-of-the-art methods but with significantly faster computation; for ST data with sub-cellular resolution (e.g. NanoString CosMx, 10x Xenium), it enables cell segmentation-free annotation of cell states, leading to highly interpretable spatial niche identification. When a single reference dataset does not encompass all possible cell states in the query ST data, Φ-Space ST integrates multiple scRNA-seq references and harmonizes the annotation results. When the query ST data contains disease-altered cell states that are not defined in any healthy references, Φ-Space ST provides interpretable characterisation of disease cell states using healthy ones.
Φ-Space ST ensures that annotation outcomes are comparable across samples and platforms, facilitating multi-sample and multi-platform spatial biological discoveries. We validated Φ-Space ST in three case studies: non-small cell lung cancer (NSCLC) using CosMx and Visium platforms, and acute myeloid leukaemia (AML) using Stereo-seq. Our method revealed niche-specific enriched cell types and distinct cell type co-presence patterns that distinguish tumour from non-tumour tissue regions. These findings highlight the potential of Φ-Space ST as a robust and scalable tool for ST data analysis for understanding complex tissues and pathologies.