OmicVerse Single-Cell — VIA Trajectory (with optional RNA velocity)

Goal

Take a preprocessed annotated single-cell AnnData and run VIA (Stassen et al. 2021, Nature Communications) for: (a) topology construction, (b) pseudotime, (c) automated terminal-state detection, (d) temporal gene-trend visualisation along inferred lineages. VIA's signature feature vs. classical trajectory methods is automated terminal-state prediction — you don't have to specify endpoints.

Two modes:

1. Vanilla VIA (t_via) — gene-distance only; root cell required.
2. Velocity-guided VIA (t_via_velo) — pass velocity_matrix=adata.layers['velocity'], gene_matrix=adata.X.todense(), velo_weight=0.5. Velocity guides the random-walk transition probabilities; root cell can be left automatic (root_user=None).

pyVIA here is OmicVerse's wrapper — it improves the original VIA's colour scheme to default OmicVerse-friendly choices and lets the user pass AnnData directly instead of explicit numpy arrays.

Quick Workflow

Vanilla VIA (t_via)