Relational Organization and Directional Stability Across Mouse Organogenesis

Relational Organization and Directional Stability Across Mouse Organogenesis develops a relational framework for analyzing directional organization in cross-sectional spatial transcriptomic development. Using the Mouse Organogenesis Spatiotemporal Transcriptomic Atlas (MOSTA) from E9.5 through E16.5, the study analyzes 4.10 million spatial roots from 53 sections and 37 biological embryos. Spatial roots are represented through source-authored tissue identity, local neighborhood context, and discretized spatial descriptors, with developmental-stage labels excluded from relation construction. The analysis asks an upstream question in developmental trajectory inference: how does the definition of biological comparability affect the relational architecture and directional structure inferred from independently sampled developmental stages? The resulting comparison space exhibits extensive recurrence and many-to-many organization across mouse organogenesis. High direct-versus-composed relational agreement is shown to be largely reproduced by category- and degree-preserving architecture references. All seven adjacent developmental contrasts remain within same-stage biological variation under both embryo-pair and aggregation-matched pseudo-population calibration, while their accumulated chronological score retains a stable negative sign across biological resampling and neighboring estimator choices. Sensitivity analyses vary neighborhood scale, categorical comparison context, relation tolerance, metric weighting, transport kernel, descriptor discretization, relation support, and graph-randomization effort. These analyses distinguish three levels of stability: fine pairwise directional structure is representation-sensitive, local directional resolution remains unresolved beyond biological variation, and the aggregate directional sign persists across the tested estimator family. The study motivates a general robustness principle for cross-sectional developmental inference: biological sampling uncertainty and sensitivity to the comparison space should be evaluated as distinct components of directional analysis.