STalign: Alignment of spatial transcriptomics data using diffeomorphic metric mapping

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Spatial transcriptomics (ST) technologies offer high-throughput gene expression analysis in tissue sections.
• Comparing ST data across different sections, samples, and technologies presents significant challenges due to variations and distortions.

Purpose of the Study:

• To develop a novel computational tool, STalign, for accurate alignment of ST datasets.
• To enable robust comparisons of gene expression data across diverse ST experiments and platforms.

Main Methods:

• Developed STalign using diffeomorphic metric mapping to handle non-linear distortions and partially matched tissue sections.
• Applied STalign for intra- and inter-technology dataset alignment.
• Integrated STalign with a 3D common coordinate framework for anatomical referencing.

Main Results:

• STalign demonstrated superior gene expression and cell-type correspondence compared to traditional affine alignment methods.
• Successful alignment of mouse brain ST datasets to the Allen Brain Atlas 3D common coordinate framework.
• Facilitated lifting of brain region annotations and analysis of compositional heterogeneity.

Conclusions:

• STalign provides a robust solution for aligning and integrating diverse spatial transcriptomics datasets.
• The tool enhances the ability to perform cross-platform comparisons and anatomical analyses.
• STalign is an open-source toolkit facilitating broader adoption in spatial biology research.