Bridging the dimensional gap from planar spatial transcriptomics to 3D cell atlases

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Spatial transcriptomics (ST) advances tissue architecture understanding but struggles with 3D atlas construction due to sparse sampling and high costs.
• Existing 2D ST methods create data gaps, limiting comprehensive 3D organ organization insights.
• Bridging these gaps is crucial for high-resolution 3D cell atlases.

Purpose of the Study:

• To introduce SpatialZ, a computational framework for generating dense 3D cell atlases from planar ST data.
• To enable the creation of virtual slices between experimentally measured sections, filling data gaps.
• To provide a versatile tool for high-resolution 3D spatial molecular landscape analysis.

Main Methods:

• SpatialZ computationally generates virtual 2D slices from sparse experimental ST data.
• The framework operates at single-cell resolution and is independent of gene coverage limitations.
• Validation involved applying SpatialZ to BRAIN Initiative Cell Census Network data and imaging mass cytometry data.

Main Results:

• SpatialZ successfully constructed a 3D hemisphere atlas with over 38 million cells.
• The framework accurately preserves cell identities, gene expression, and spatial relationships.
• Demonstrated extensibility by analyzing 3D spatial gradients in human breast cancer using imaging mass cytometry data.

Conclusions:

• SpatialZ enables the construction of dense 3D cell atlases from planar ST data, overcoming current limitations.
• The generated atlases offer unprecedented 3D resolution for exploring spatial molecular architectures.
• SpatialZ facilitates novel analyses, including in silico sectioning and 3D mapping, advancing biological discovery.