Proseg

Uses sampling-based models to assign transcripts with probabilities and provide uncertainty estimates, directly reducing misassignment and spurious co-expression as highlighted in the model arguments.

Supports major spatial transcriptomics platforms like Xenium, CosMx, MERSCOPE, and Visium HD through preset command-line arguments, simplifying data handling across different technologies.

Models cells across multiple z-layers with the --voxel-layers argument, enabling accurate segmentation for three-dimensional spatial data without additional tools.

Generates outputs in spatialdata zarr, GeoJSON, and matrix formats, facilitating seamless integration with analysis pipelines and tools like Xenium Explorer via the proseg-to-baysor command.