MistyR

MistyR applies explainable machine learning to highly multiplexed spatial omics data to extract relationships and estimate structural and functional interactions across spatial contexts in tissues.

Key Features:

• Scalability: Handles spatial omics datasets ranging from dozens to thousands of measured markers.
• Explainability: Produces explainable machine learning outputs that expose the relationships extracted from spatial data.
• Multiview Analysis: Constructs multiple analytical views that focus on distinct spatial or functional contexts to dissect different effects.

Scientific Applications:

• Evaluation datasets: Evaluated on in silico datasets and real-world data from imaging mass cytometry and spatial transcriptomics.
• Breast cancer microenvironment: Applied to breast cancer datasets to estimate structural and functional interactions within the tissue microenvironment.
• Clinical associations: Relates estimated structural and functional interactions to clinical features.

Methodology:

Builds multiple analytical views that focus on distinct spatial or functional contexts and applies explainable machine learning to extract and relate structural and functional interactions to clinical features.