Cmd-line script for running Banksy in bioinformatics pipelines.

[vladimirkovacevic]

Community might benefit from this script for easier execution and testing of Babnksy, especially in integration in bioinformatics pipelines. It adds a self-contained R script, providing an end-to-end pipeline from .h5ad input to clustered results and visualizations.

Highlights

• CLI-driven, portable (base R argument parsing; Docker/HPC friendly)
• Robust .h5ad loading (sparse/dense matrices, flexible spatial coordinate detection)
• Adaptive QC and standard normalization
• Full BANKSY workflow: neighborhood features, PCA, optional UMAP, Leiden clustering
• Supports multiple lambda values for cell typing vs. domain segmentation
• Reproducible (fixed seed) with verbose logging
• Outputs: SpatialExperiment (.rds), cluster assignments (.csv), summary stats, spatial & UMAP plots