Loom: Multi-Region Analysis of Spatial Transcriptomics with Local Neighborhoods and Global Trajectories

Loom is a new visual computing system for spatial transcriptomics, aimed at making multi-region biological analysis more tractable. The core problem is familiar to anyone building data-heavy tools: once you combine spatial structure, cell-reference data, and pseudo-temporal behavior, the visualization and registration burden gets messy fast.

The system focuses on three things at once: local microenvironments, cross-sample or cross-region comparisons, and global trajectories that show how cells appear to transition over time. To support that, Loom pairs a novel glyph with a computational backbone designed to connect spatial enrichment, gene-expression dynamics, and temporal simulation data in one workflow.

The work was presented by Siyuan Zhao and collaborators and submitted to arXiv on 24 Jul 2026 under cs.GR, with a quantitative biology cross-listing. It was evaluated through two expert case studies with tissue pathology and oncology specialists, plus an external usability study, and the results point to better discovery of cellular transitions and spatiotemporal expression patterns.

For game developers, the interesting part is the interaction design problem: this is a reminder that glyph design, coordinated views, and careful data reduction can make or break complex analysis tools. Even outside biotech, the same principles apply to telemetry dashboards, simulation debugging, and any editor that has to reconcile local detail with system-wide structure.