Unifying physical and molecular coordinate systems across modalities in spatial biology

The scientific community has long grappled with the challenge of integrating fragmented multi-modal data into a unified physical and molecular coordinate system for spatial biology. In a pioneering effort, researchers introduce MAPS, a modality-agnostic platform designed to address this issue. MAPS comprises three key components: MAPS-alignment, MAPS-integration, and MAPS-Explorer.

MAPS-alignment enables ultrafast alignment of data from any modality, streamlining the process of combining diverse spatial biology datasets. MAPS-integration offers the capability for both anchored and unanchored integration across orthogonal modalities, facilitating the reconstruction of three-dimensional multi-modal representations. Lastly, MAPS-Explorer provides a powerful tool for large-scale interactive 3D analysis of spatial biology data.

The researchers rigorously tested MAPS against existing methods using extensive benchmarks involving 34 datasets, encompassing 16 technology platforms and 6 modalities. The results demonstrated MAPS' superior performance across all tested datasets. Furthermore, MAPS revealed intricate multi-modal tissue architectures across diverse biological systems, both in mouse and human models.

At the cross-consortium scale, MAPS integrated an impressive 434 slices containing 21 million cells from 18 atlases and 5 modalities. This integration culminated in the construction of the most comprehensive 3D multi-modal mouse brain atlas to date. At an individual laboratory scale, MAPS empowered researchers to transform routine 2D spatial assays into the reconstruction of continuous 3D multi-modal landscapes of human hepatocellular carcinoma.

This breakthrough not only illuminated the limitations of traditional 2D spatial relationships but also uncovered depth-dependent immune-state transitions within the tumor microenvironment.