scFates.tl.explore_sigma¶
- scFates.tl.explore_sigma(adata, Nodes, use_rep=None, ndims_rep=None, sigmas=[1000, 100, 10, 1, 0.1, 0.01], nsteps=1, metric='euclidean', seed=None, plot=False, second_round=False, **kwargs)¶
Explore varisou sigma parameters for best tree fitting. Given that high sigma tend to collapse the principal points into the middle of the whole data (meaning taking in account all the datapoints regardless their locality), it is possible to explore which sigma is best by detecting at which level the tree stops collapsing.
- Parameters
- adata
Annotated data matrix.
- Nodes
Number of nodes composing the principial tree, use a range of 10 to 100 for ElPiGraph approach and 100 to 2000 for PPT approach.
- use_rep
Choose the space to be learned by the principal tree.
- ndims_rep
Number of dimensions to use for the inference.
- sigmas
Range of sigma parameters to test.
- device
Run method on either cpu or on gpu.
- nsteps
Number of SimplePPT iteration, usually 1 is enough.
- metric
Distance metric to use.
- seed
A numpy random seed.
- plot
Plot the resulting tree.
- second_round
Perform a second exploration, on a restricted sigma parameters based on the first estimated sigma.
- **kwargs
Arguments passsed to
elpigraph.computeElasticPrincipalCircle()
- Returns
sigma – suggested sigma value
- Return type