The paper introduces GeneTrajectory, an innovative approach for inferring gene trajectories from single-cell RNA-sequencing data. Unlike traditional methods that focus on cell pseudotime and trajectory inference, GeneTrajectory identifies gene trajectories by calculating optimal transport distances between gene distributions across the cell-cell graph. This method effectively extracts gene programs and their pseudotemporal order, even in scenarios where multiple concurrent processes are occurring in the same cells. The authors demonstrate the effectiveness of GeneTrajectory through simulation experiments and real-world applications, including myeloid lineage maturation and dermal condensate differentiation in mouse skin. GeneTrajectory shows superior performance in recovering gene order along biological processes compared to other cell trajectory inference methods. The approach bypasses the need for constructing cell pseudotime, making it suitable for scenarios where cells do not form clear lineages. The paper also discusses potential extensions and applications of GeneTrajectory, highlighting its broad utility in understanding gene dynamics and biological processes.The paper introduces GeneTrajectory, an innovative approach for inferring gene trajectories from single-cell RNA-sequencing data. Unlike traditional methods that focus on cell pseudotime and trajectory inference, GeneTrajectory identifies gene trajectories by calculating optimal transport distances between gene distributions across the cell-cell graph. This method effectively extracts gene programs and their pseudotemporal order, even in scenarios where multiple concurrent processes are occurring in the same cells. The authors demonstrate the effectiveness of GeneTrajectory through simulation experiments and real-world applications, including myeloid lineage maturation and dermal condensate differentiation in mouse skin. GeneTrajectory shows superior performance in recovering gene order along biological processes compared to other cell trajectory inference methods. The approach bypasses the need for constructing cell pseudotime, making it suitable for scenarios where cells do not form clear lineages. The paper also discusses potential extensions and applications of GeneTrajectory, highlighting its broad utility in understanding gene dynamics and biological processes.