API

class simpleppt.SimplePPT(F, R, B, L, d, score, lam, sigma, nsteps, metric, tips=None, forks=None, root=None, pp_info=None, pp_seg=None)

A python object containing the data used for dynamical tracks analysis.

Parameters:

  • F (array) – coordinates of principal points in the learned space.

  • R (array) – soft assignment of datapoints to principal points.

  • B (array) – adjacency matrix of the principal points.

  • L (array) – Laplacian matrix.

  • d (array) – Pairwise distance matrix of principal points.

  • score (float) – Score minimized during the tree learning.

  • tips (Optional[Iterable]) – Node IDs of the tree that have degree 1.

  • forks (Optional[Iterable]) – Node IDs of the tree that have a degree of more than 1.

  • root (Optional[int]) – Selected node ID as the root of the tree for distance calculations.

  • pp_info (Optional[DataFrame]) – Per node ID info of distance from the root, and segment assigment.

  • pp_seg (Optional[DataFrame]) – Per segment info with node ID extremities and distance.

__init__(F, R, B, L, d, score, lam, sigma, nsteps, metric, tips=None, forks=None, root=None, pp_info=None, pp_seg=None)
set_tips_forks()

Obtains the tips and forks of the tree.

Returns:

adds to SimplePPT object the following fields –

.tips

Node IDs of the tree that have degree 1..

.forks

Node IDs of the tree that have a degree of more than 1.

Return type:

simpleppt.SimplePPT

set_branches(root=None)

Assign branches/segments to nodes.

Returns:

adds to SimplePPT object the following fields –

.pp_info

Per node ID info of distance from the root, and segment assigment.

.pp_seg

Per segment info with node ID extremities and distance.

Return type:

simpleppt.SimplePPT

simpleppt.ppt(X, W=None, Nodes=None, init=None, sigma=0.1, lam=1, metric='euclidean', nsteps=50, err_cut=0.005, device='cpu', gpu_tbp=16, seed=None, progress=True)

Generate a principal tree.

Learn a simplified representation on any space, composed of nodes, approximating the position of the datapoints on a given space.

Parameters:

  • X – n-dimensionnal matrix to be learned.

  • W – weight matrix, having the same dimensions as X.

  • Nodes (Optional[int]) – Number of nodes composing the principial tree.

  • init (Optional[DataFrame]) – Initialise the point positions.

  • sigma (Union[float, int, None]) – Regularization parameter.

  • lam (Union[float, int, None]) – Penalty for the tree length.

  • metric (str) – The metric to use to compute distances in high dimensional space. For compatible metrics, check the documentation of sklearn.metrics.pairwise_distances if using cpu or cuml.metrics.pairwise_distances if using gpu.

  • nsteps (int) – Number of steps for the optimisation process.

  • err_cut (float) – Stop algorithm if proximity of principal points between iterations less than defined value.

  • gpu_tpb – Threads per block parameter for cuda computations.

  • seed (Optional[int]) – A numpy random seed.

  • progress (bool) – Show progressbar of the tree learning.

Returns:

SimplePPT object with the following fields –

.F

coordinates of principal points in the learned space.

.R

soft assignment of datapoints to principal points.

.B

adjacency matrix of the principal points.

.L

Laplacian matrix.

.d

Pairwise distance matrix of principal points.

.score

Score minimized during the tree learning.

Return type:

simpleppt.SimplePPT

simpleppt.project_ppt(SP, emb, size_nodes=None, plot_datapoints=True, alpha_seg=1, alpha_nodes=1, ax=None, show=None, **kwargs)

Project principal graph onto embedding.

Parameters:

  • SP – SimplePPT object.

  • emb – embedding to project the tree onto.

  • size_nodes (Optional[float]) – size of the projected prinicpal points.

  • alpha_seg – segment alpha

  • alpha_nodes – node alpha.

  • ax – Add plot to existing ax

  • show (Optional[bool]) – show the plot.

  • kwargs – arguments to pass to scanpy functions plt.scatter

Return type:

If show==False a Axes