Inferring cell trajectories of spatial transcriptomics via optimal transport analysis

Abstract The integration of cell transcriptomics and spatial coordinates to organize differentiation trajectories remains a challenge. Here we introduce spaTrack, a trajectory inference method using optimal transport to incorporate both transcriptomics and distance of spatial transcriptomics sequencing data into transition costs. spaTrack could construct fine spatial trajectories reflecting the true differentiation topology, as well as trace cell dynamics across multiple samples with temporal intervals. To capture the dynamic drivers, spaTrack models the cell fate as a function of expression profile along temporal intervals driven by transcription factors. Applying spaTrack, we successfully disentangle spatiotemporal trajectories of axolotl telencephalon regeneration and mouse midbrain development. Furthermore, we uncover diverse malignant lineages expanding in a primary tumor. One of the lineages with upregulated extracellular matrix organization implants to the metastatic site and subsequently colonizes to a secondary tumor. Overall, spaTrack greatly facilitates trajectory inference from spatial transcriptomics, providing insights in cell differentiation of broad areas.