moscot (multi-omics single-cell optimal transport) is an optimal-transport framework for mapping cells through time and space, scaling OT to millions of cells with multimodal support. As a headline demonstration it reconstructed a 1.7M-cell mouse-embryo developmental trajectory across 20 timepoints.

From Fabian Theis's group (joint→Helmholtz Munich + Hebrew University (Nitzan) + Apple ML Research); published in Nature (2025).

Library

Language Python
sciencebiologyframeworkresearch