Abstract
<title>Abstract</title> <p>Temporal RNA-seq is widely used to study development and disease, yet many analytical approaches compare stage-matched samples, assuming temporal equivalence between conditions and limiting measurement of gene expression program reorganization. We introduce WATER (Windowed Assignment of Temporal Expression of RNA), a framework that reconstructs temporal gene expression trajectories independently within each condition before comparison, enabling measurement of gene expression mistiming, or molecular heterochrony. WATER integrates differential expression across pairwise temporal windows, silhouette-guided clustering, Pearson correlation confidence auditing, iterative refinement, and leave-one-out stability analysis. In a U11-null mouse forelimb time course, in which minor spliceosome inhibition causes severe limb shortening, WATER assigns 97.5% of differentially expressed genes to thirteen high-confidence trajectories, with 63.1% shifting trajectory class, 72.8% of which reflect cell-intrinsic transcriptional change rather than shifts in cell-type composition. Applied to cross-species limb evolution, Eed-dependent chromatin regulation, and 5xFAD Alzheimer's disease cortex, WATER detects molecular heterochrony across developmental and disease contexts.</p>