banksy

What problem does it solve?

Banksy provides a cohesive workflow to analyze spatial transcriptomics data by constructing spatial neighbourhood graphs, generating BANKSY matrices that integrate self and neighbour features, and performing clustering and visualization to reveal tissue structure.

Core Features & Use Cases

• Graph construction and weighting: build spatial graphs with fixed neighbours and azimuthal components, then assign distance-based weights.
• Matrix fusion and clustering: concatenate self- with neighbour- matrices into BANKSY representations and cluster with Leiden or MCLUST; compare spatial vs non-spatial results.
• Visualization and downstream analysis: generate spatial maps, UMAPs, PCA embeddings, and plot connectivity; convert results back to AnnData for downstream analyses.
• Use Case: apply BANKSY to a sci data object to identify tissue regions in a Slide-seq or 10x spatial dataset and compare against manual labels.

Quick Start

1. Load your spatial AnnData object with coordinates.
2. Run initialize_banksy to create banksy_dict and compute neighbour graphs.
3. Run generate_banksy_matrix to build the BANKSY matrix and perform clustering with run_Leiden_partition or run_mclust_partition; visualize results with plot_banksy.