omicverse-single-cell-rna-velocity

To run single-cell RNA velocity end to end, you need a velocity-ready AnnData object with spliced and unspliced layers. The workflow handles preprocessing, moment estimation, dynamics fitting, velocity estimation, graph construction, and final stream plots.

GraphVelo refinement improves RNA velocity embeddings by refining an existing velocity layer and projecting the result into shared embeddings. It validates expected output keys like X_pca, X_umap, velocity_S_graph, and velocity_gv_umap for downstream interpretation.

Yes, you can compare scvelo, dynamo, latentvelo, and graphvelo branches using the same input object. Branch-aware velocity analysis runs these paths while preserving the same input structure and workflow for accurate branch comparisons.

Before estimating RNA velocity with OmicVerse, you need a velocity-ready single-cell AnnData object containing spliced and unspliced layers. Preprocessing steps must be completed to ensure moment estimation and dynamics fitting can execute correctly.

After generating a velocity graph and UMAP embedding, the workflow validates specific keys including X_pca, X_umap, velocity_S_graph, and velocity_gv_umap. These keys ensure the velocity outputs are correctly structured for downstream interpretation.