BarDensr

BarDensr demixes linearly multiplexed spatial transcriptomics images to estimate rolony densities and recover RNA signals in high-density or low-resolution imaging.

Key Features:

• Generative Model: Constructs a generative model representing the physical image formation process that produces the observed spatial transcriptomics data.
• Sparse Convex Optimization: Employs sparse convex optimization techniques to estimate underlying rolony densities from mixed voxel signals.
• Signal Recovery: Achieves high signal recovery performance in conditions with high rolony density or limited imaging resolution.
• Parallelization and Speed: Implements parallelizable computational routines to accelerate analysis of large datasets.

Scientific Applications:

• High-density Spatial Transcriptomics Mapping: Enables mapping of RNA expression when transcript density is high relative to imaging voxel size, facilitating interpretation of mixed signals.
• Profiling Fine Neuronal Processes and Complex Tissues: Supports localization of molecular targets within fine neuronal processes and other complex tissue structures by demixing overlapping rolony signals.

Methodology:

Constructs a generative model of the imaging process; applies sparse convex optimization to estimate rolony densities from mixed signals; and uses parallelizable algorithms to improve computational speed.