OmicVerse Single-Cell — CellRank Fate Maps from RNA Velocity

Goal

Take a single-cell AnnData that already has RNA velocity computed (adata.layers['velocity'], adata.uns['velocity_graph'], etc.) and produce a fate map: terminal-state assignments, per-cell fate probabilities, and branch-aware marker dynamics. CellRank does the velocity-driven random-walk side; OmicVerse plotting + dynamic_features does the visualisation and gene-trend side.

This skill is the post-velocity skill: assumes velocity is already computed (use single-cell-rna-velocity for that). It pairs CellRank's VelocityKernel + ConnectivityKernel mix with the same dynamic_features GAM backend used by Monocle / VIA / etc., so plot helpers (branch_streamplot, dynamic_trends, dynamic_heatmap) are interoperable across all trajectory skills.

Quick Workflow