PACE: Geometry-Aware Bridge Transport for Single-Cell Trajectory Inference

Researchers introduced PACE, a geometry-aware single-cell trajectory inference method for snapshot time-course data.

PACE combines anisotropic Riemannian transport, iterative cross-time coupling, and neural bridges to reconstruct continuous cell-state dynamics.

The method improves reconstruction scores and velocity alignment across multiple datasets, including embryoid body differentiation.

It helps address the identifiability challenge in inferring dynamics from destructive snapshots without cell pairing or lineage tracing.