scvelo.tl.latent_time

scvelo.tl.latent_time(data, vkey='velocity', min_likelihood=0.1, min_confidence=0.75, min_corr_diffusion=None, weight_diffusion=None, root_key=None, end_key=None, t_max=None, copy=False)

Computes a gene-shared latent time.

Gene-specific latent timepoints obtained from the dynamical model are coupled to a universal gene-shared latent time, which represents the cell’s internal clock and is based only on its transcriptional dynamics.

data: AnnData

Annotated data matrix

vkey: str (default: ‘velocity’)

Name of velocity estimates to be used.

min_likelihood: float between 0 and 1 or None (default: .1)

Minimal likelihood fitness for genes to be included to the weighting.

min_confidence: float between 0 and 1 (default: .75)

Parameter for local coherence selection.

min_corr_diffusion: float between 0 and 1 or None (default: None)

Only select genes that correlate with velocity pseudotime obtained from diffusion random walk on velocity graph.

weight_diffusion: float or None (default: None)

Weight applied to couple latent time with diffusion-based velocity pseudotime.

root_key: str or None (default: ‘root_cells’)

Key (.uns, .obs) of root cell to be used. If not set, it obtains root cells from velocity-inferred transition matrix.

end_key: str or None (default: None)

Key (.obs) of end points to be used.

t_max: float or None (default: None)

Overall duration of differentiation process. If not set, a overall transcriptional timescale of 20 hours is used as prior.

copy: bool (default: False)

Return a copy instead of writing to adata.

Returns

latent_time (.obs) – latent time from learned dynamics for each cell