Xen-Spat: An Analysis Pipeline for Imaging-Based Spatial Transcriptomic Data to Quantify Cellular Colocalization in the Tumor Microenvironment

Background/Objectives: Imaging-based spatial transcriptomic platforms measure gene expression in single cells within intact tissue, but cell type annotation and spatial analysis are usually performed with separate tools. We developed Xen-Spat, an analysis pipeline that carries one set of cell type labels through complementary measures of cellular colocalization in the tumor microenvironment. Methods: Cells were quality-filtered, annotated in two tiers within each sample and mapped to uniform cell type labels. Four spatial modules were applied: the Morisita index for regional colocalization, niche analysis with assignment of tumor niches by the Dice coefficient, Getis-Ord Gi* hotspot detection with quantification of hotspot overlap, and distance-based enrichment with pseudobulk differential expression. Results: In six HER2+ breast cancer biopsies profiled by 10x Xenium, one niche per sample was assigned to the tumor compartment (Dice coefficient 0.60 to 0.83, compared with 0.22 or lower for other niches). Macrophages were enriched within 25 µm of tumor cells in all six samples; after excluding genes affected by segmentation spillover, 50 genes differed between macrophages near tumor cells and at the margin. In an independent public Xenium dataset, the pipeline reproduced a reported difference in MMP11 expression between tissue regions. Conclusions: Xen-Spat provides a reproducible workflow for quantifying cellular colocalization in imaging-based spatial transcriptomic data.