Advances in spatially resolved and single-cell technologies have transformed our ability to characterize tissues across multiple molecular layers, including transcriptomics, proteomics, imaging, and epigenomics. However, despite rapid experimental progress, most datasets remain inherently unpaired: proteins, transcripts, chromatin states, and spatial measurements are frequently generated in different cells, adjacent tissue sections, or independent experiments. As a result, the primary challenge facing the field is increasingly not data generation but the integration of heterogeneous, incomplete, and partially overlapping observations into coherent biological models.

This Topical Collection aims to highlight emerging computational, experimental, and analytical approaches that address the realities of modern multi-omics data integration. We welcome contributions spanning spatial transcriptomics, spatial proteomics, multimodal imaging, atlas construction, machine learning, data harmonization, uncertainty modeling, benchmarking, and open-source software development. Particular emphasis will be placed on methods and applications that enable integration across unpaired datasets, modalities, spatial scales, and biological systems.

By bringing together researchers developing technologies, computational methods, and biological applications, this collection will provide a forum for addressing one of the most important challenges in contemporary systems biology: connecting measurements across modalities and spatial contexts when direct correspondence is absent. The collection aims to accelerate the development of robust integration strategies that will enable more faithful biological interpretation and support the next generation of spatial and multimodal atlases.