Scaling up spatial transcriptomics for large-sized tissues: uncovering cellular-level tissue architecture beyond...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Spatial transcriptomics (ST) technologies offer insights into tissue gene expression with spatial context.
• Current ST platforms face limitations including high cost, low resolution, and small tissue capture areas, restricting their widespread use.
• Analyzing large tissue sections with cellular resolution remains a significant challenge in transcriptomic research.

Purpose of the Study:

• To introduce iSCALE, a novel computational method for predicting super-resolution gene expression.
• To enable automatic annotation of cellular-level tissue architecture in large tissue samples.
• To overcome the limitations of existing ST platforms for analyzing large-sized tissues.

Main Methods:

• iSCALE utilizes computational approaches to infer gene expression patterns at a higher resolution.
• The method integrates spatial information to predict gene activity across large tissue areas.
• Validation involved benchmarking, immunohistochemistry staining, and expert pathologist annotation.

Main Results:

• iSCALE successfully predicted super-resolution gene expression and annotated cellular architecture in large tissues.
• The method demonstrated accuracy through comprehensive validation experiments.
• Application to multiple sclerosis brain samples revealed lesion-associated cellular characteristics missed by conventional ST.

Conclusions:

• iSCALE significantly advances the analysis of large-sized tissues by providing automatic, unbiased annotation and cell type composition inference.
• The method enhances the discovery of biological features not easily discernible through traditional methods.
• iSCALE broadens the applicability of spatial transcriptomics for large-scale tissue studies.