CellBRIDGE: Learning Cellular Trajectories via Interaction-Aware Alignment

Inferring dynamics from population snapshots is a fundamental challenge in machine learning and biology. In scRNA-sequencing (scRNA-seq), destructive measurements preclude direct tracking of individual cells across time, making trajectory inference underdetermined. Optimal Transport (OT) provides a principled framework for snapshot alignment, but a long-standing modeling question is which cost functions yield biologically meaningful couplings. Standard OT approaches rely on gene-expression distances, implicitly treating cells as independent points and neglecting structured cell-cell communication mediated by ligand-receptor signaling. We introduce CellBRIDGE (Cell-Based Regularized Interaction-Driven Gene Expression), which augments feature-based OT with a directed, typed interaction cost derived from ligand-receptor activity. By explicitly modeling cell-cell communication, CellBRIDGE improves cross-snapshot couplings and downstream trajectory estimates across synthetic and real scRNA-seq datasets relative to feature-only baselines. Notably, CellBRIDGE enables mechanistically interpretable in silico perturbations: on lung cancer data, silencing specific ligand-receptor pairs induces trajectory shifts that recapitulate expected effects of targeted pathway inhibition.

To understand disease and development, scientists often want to know how individual cells change over time. But many powerful single-cell experiments are destructive: they can read which genes are active in thousands of cells, but each cell is consumed in the process, so the same cell cannot be followed later. This leaves researchers with snapshots, and they must infer how cells moved from one state to another. We introduce CellBRIDGE, a machine-learning method that makes this reconstruction more reliable by using not only how similar cells look internally, but also how they appear to communicate with nearby cells through known signaling molecules. In tests on simulated and real biological data, this communication-aware matching produced better inferred cell paths than methods that only compare gene activity. Because the communication signals are explicit, researchers can also “turn off” particular signaling routes in the model and see how predicted cell behavior changes; in lung cancer data, these virtual interventions reflected known drug-target pathways. This could help biologists generate better hypotheses about disease progression and prioritize experiments before running costly lab studies.