SINFONIA

SINFONIA identifies spatially variable genes (SVGs) to characterize spatial domains within tissue samples from spatial transcriptomics datasets.

Key Features:

• Python implementation: Implemented in Python for integration into computational workflows.
• Scalability: Scales to datasets of varying sizes and complexities.
• Ensemble strategies: Employs ensemble strategies to improve identification of spatially variable genes.
• Robustness to SVG selection: Produces consistent results regardless of the number of SVGs chosen.
• Systematic benchmarking: Evaluated on 15 spatial transcriptomic datasets generated with varying protocols, sizes, dimensions, and qualities.
• Comparative performance: Demonstrated improvements relative to three baseline methods and two variants.
• Evaluation metrics: Assessed using 21 quantitative metrics including spatial clustering, domain resolution, latent representation, spatial visualization, and computational efficiency.

Scientific Applications:

• Spatial domain characterization: Delineates spatial domains within tissues by identifying SVGs.
• Developmental biology: Analyzes spatial gene expression patterns relevant to tissue development.
• Oncology: Resolves spatial heterogeneity in tumor microenvironments through SVG detection.
• Regenerative medicine: Investigates spatial organization relevant to tissue repair and regeneration.

Methodology:

SINFONIA employs ensemble strategies to identify SVGs and was benchmarked via systematic evaluation across 15 spatial transcriptomics datasets using 21 quantitative metrics with comparisons to three baseline methods and two variants.