Skip to contents

Run PHATE on a SingleCells/MetaCells object

phate_sc.Rd

Wrapper around manifoldsR::phate() for the SingleCells and MetaCells classes. PHATE (Potential of Heat-diffusion for Affinity-based Trajectory Embedding) produces a low-dimensional embedding that preserves both local and global structure by operating on a diffusion process over the data manifold. Unlike UMAP or t-SNE, PHATE is explicitly designed to reveal continuous progressions and branching structure, making it the preferred choice for data with developmental or trajectory-like organisation.

When use_knn = TRUE (the default), the kNN graph already stored on the object is reused; otherwise neighbours are computed from the chosen embedding. The algorithm then constructs a diffusion operator, raises it to a power (the diffusion time t, see manifoldsR::params_phate()) that denoises the manifold, and computes potential distances that are finally embedded via metric MDS.

Because PHATE inherently smooths over the kNN graph, it pairs naturally with MetaCells: the combination yields a particularly clean view of continuous biological processes on denoised data.

Usage

phate_sc(
  object,
  use_knn = TRUE,
  embd_to_use = "pca",
  slot_name = "phate",
  no_embd_to_use = NULL,
  modality = c("rna", "adt", "wnn"),
  n_dim = 2L,
  k = 5L,
  knn_method = c("kmknn", "hnsw", "balltree", "annoy", "nndescent", "exhaustive"),
  nn_params = manifoldsR::params_nn(),
  phate_params = manifoldsR::params_phate(),
  seed = 42L,
  .verbose = TRUE
)

Arguments

object

SingleCells, MetaCells class.

use_knn

Boolean. Use the kNN graph found in the object. Defaults to TRUE. If not available, will default to the embedding.

embd_to_use

String. The embedding to use for PHATE. Must be available in the object.

slot_name

String. The name of this embedding within the object. Defaults to "phate".

no_embd_to_use

Optional integer. Number of embedding dimensions to use. If NULL all will be used.

modality

String. On which modality to run the UMAP. One of c("rna", "adt", "wnn"). The two latter options are only available for multi-modal versions with the added data.

n_dim

Integer. Number of PHATE dimensions. Currently only 2L is supported. Defaults to 2L.

k

Integer. Number of nearest neighbours for graph construction. Defaults to 5L.

knn_method

String. Approximate nearest neighbour algorithm. One of "hnsw", "balltree", "annoy", "nndescent", or "exhaustive".

nn_params

Named list. See manifoldsR::params_nn().

phate_params

Named list. See manifoldsR::params_phate().

seed

Integer. For reproducibility.

.verbose

Boolean. Controls verbosity.

Value

The object with a "phate" embedding added.

Examples

# PHATE embedding off the cached kNN graph
sc <- demo_single_cells()
sc <- phate_sc(sc, .verbose = FALSE)
dim(get_embedding(sc, "phate"))
#> [1] 500   2

unlink(sc@dir_data, recursive = TRUE, force = TRUE)