Investigating spatial dynamics in spatial omics data with StarTrail

Abstract

Spatial omics technologies revolutionize our view of biological processes within tissues. However, existing methods fail to capture localized, sharp changes characteristic of critical events (e.g. tumor development). Here, we present StarTrail, a novel gradient based method that powerfully defines rapidly changing regions and detects "cliff genes", genes exhibiting drastic expression changes at highly localized or disjoint boundaries. StarTrail, the first to leverage spatial gradients for spatial omics data, also quantifies directional dynamics. Across multiple datasets, StarTrail accurately delineates boundaries (e.g., brain layers, tumor-immune boundaries), and detects cliff genes that may regulate molecular crosstalk at these biologically relevant boundaries but are missed by existing methods. For instance, StarTrail precisely pinpointed the cancer-immune interface in a HER2+ breast cancer dataset, unveiled key cliff genes including a potential prognostic biomarker IGSF3, highlighting NK-, B-cell mediated immunity, and B cell receptor signaling pathways missed by all spatial variable gene methods attempted. StarTrail, filling important gaps in current literature, enables deeper insights into tissue spatial architecture.