SCALPEL: A pipeline for processing large-scale spatial transcriptomics data

Area of Science:

• Genomics
• Bioinformatics
• Neuroscience

Background:

• Spatial transcriptomics offers detailed gene expression mapping within tissues.
• Traditional methods lack the resolution for complex tissue architecture analysis.
• Analyzing large-scale spatial datasets presents significant computational challenges.

Purpose of the Study:

• To present SCALPEL, an advanced workflow for processing large-scale spatial transcriptomics data.
• To enhance the accuracy and scope of spatial transcriptomics analysis.
• To provide a robust foundation for downstream spatial analyses and atlas-level studies.

Main Methods:

• Advanced 3D segmentation for dense and heterogeneous tissues.
• Transcriptome-based doublet detection and refined filtering criteria.
• Spatial domain detection, anatomical registration (Allen Mouse Brain CCFv3), and genome-wide expression imputation from scRNAseq.

Main Results:

• Substantial improvements in cell number and expression profile clarity compared to previous versions.
• Enhanced spatial registration and anatomical alignment.
• Demonstrated superior performance on a whole-mouse-brain dataset.

Conclusions:

• SCALPEL provides a robust and improved pipeline for large-scale spatial transcriptomics.
• The workflow sets a new standard for analyzing complex tissue architecture and gene expression.
• Enables deeper insights into cellular interactions and disease pathology.