Temporal ordering of migration increments carries directional memory under MYO10 depletion and collagen exposure

Cell migration is commonly summarized by speed, mean-squared displacement, or a single persistence time, although these descriptors discard the order of successive displacement increments. We reanalyzed 48,134 trajectories from 117 fields of view in a public two-by-two factorial experiment combining MYO10 depletion and collagen exposure, using equal-field and equal-repeat inference. Both perturbations suppressed motility, but their combination produced a positive buffering interaction in directional persistence. An analytical order-null that preserves each trajectorys increments, length, net displacement, and static polarity showed that most of the reproducible interaction depended on serial order. Exact decompositions localized the signal to directional organization and to both shared-field and cell-relative motion; leave-one-cell-out estimation excluded focal-cell self-inclusion as its source. We then froze the framework and evaluated 65 public movies from MDA-MB-231, HUVEC, and MDCK systems. HUVEC retained positive sequence excess through 120 min, MDA-MB-231 showed a shorter positive horizon, and MDCK regions transitioned from positive to negative sequence excess. A stationary linear active-memory model and a stationary angular hidden-state model failed to reproduce the complete lag-resolved hierarchy. Serial ordering therefore provides a transferable coordinate for distinguishing migration-memory regimes, whereas the specific MYO10-collagen interaction remains limited to the discovery dataset. Impact statementSerial-order analysis distinguishes sustained, short, and sign-reversing migration-memory regimes while separating transferable trajectory structure from dataset-specific MYO10-collagen effects.