Fig. 2From: Integrating temporal single-cell gene expression modalities for trajectory inference and disease predictionIntegration improves inference of cell cycle and differentiation trajectories. Trajectory inference was performed with partition-based graph abstraction to assess the quality of inferred embryonic cell cycle, hematopoiesis differentiation (Nestorowa), NKT cell differentiation, and hematopoiesis differentiation (Olsson) trajectories from (A top panel) spliced and unspliced or (A bottom panel) moments of spliced and RNA velocity integrated features generated from ten integration methods. The boxplots represent trajectory inference correlation scores (\(\mathrm {TI}_{\text {corr}}\)) for ten random root cells. Asterisk indicates the method with the highest median \(\mathrm {TI}_{\text {corr}}\) score. B PAGA predicted trajectories and diffusion map embeddings representing the inferred biological trajectory for unintegrated data, as well as high ranking performers for unspliced and RNA velocity integrationBack to article page