Modes for transition matrix calculation. #756
-
Is there documentation on the modes available to calculate the transitions matrix? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
I am assuming you're referring to the
For the backward velocities (relevant to your use case), only
There are also other approaches that don't require RNA velocity:
pk = ~PseudotimeKernel(adata, time_key=...)
pk.compute_transition_matrix(...)
Lastly, if you're using the GPCCA._compute_initial_states
GPCCA._set_initial_states_from_macrostates |
Beta Was this translation helpful? Give feedback.
I am assuming you're referring to the
VelocityKernel
. The modes are explained in https://cellrank.readthedocs.io/en/stable/api/cellrank.tl.kernels.VelocityKernel.compute_transition_matrix.html#cellrank.tl.kernels.VelocityKernel.compute_transition_matrixBriefly:
deterministic
- similarity between RNA velocity and connectivities (based onscheme
, default is correlation, can be cosine, ...)sampling
- sample 1 vector from normal distribution with velocity mean/varmonte_carlo
- same as above, but with many samples that are averagedstochastic
- 2nd order approximation, …