scvelo - RNA Velocity Analysis with scVelo for Single-Cell RNA-seq
Estimate cell state transitions from unspliced/spliced mRNA dynamics, infer trajectory directions, compute latent time, and identify driver genes in single-cell RNA-seq data.
Tags
Updated: 2026-06-30Capabilities
Typical Inputs
Typical Outputs
What this skill does
- load spliced and unspliced RNA data
- filter and normalize count matrices
- estimate RNA velocity vectors
- compute velocity transition graph
- compute latent time from dynamics
- compute velocity pseudotime
- rank driver genes by velocity fit
- compute velocity confidence scores
- generate velocity embedding plots
- run PAGA trajectory graph
- plot gene phase portraits
Inputs
- AnnData object with spliced and unspliced layers
- Single-cell RNA-seq count matrices
- UMAP embeddings in obsm
- Cell cluster labels (e.g. Leiden)
Outputs
- RNA velocity vectors per gene per cell
- Latent time per cell
- Velocity pseudotime per cell
- Velocity confidence and speed scores per cell
- Ranked driver gene list
- Velocity embedding stream and arrow plots
- PAGA trajectory graph with transition confidences
- Gene-level kinetic parameters (alpha, beta, gamma)
Requirements
- Python environment with scvelo installed (pip install scvelo)
- Spliced and unspliced count matrices from velocyto, STARsolo, or kallisto|bustools
- Minimum 2,000 cells recommended
- Sufficient sequencing depth for intron coverage
- Pre-computed k-NN neighbor graph for moments calculation