Identify ligands responsible for trajectory dynamics using Monocle graph learning.

Runs a standard Monocle3 trajectory reconstruction pipeline, followed by Entrain to identify ligands responsible for trajectory dynamics.

Usage

get_traj_ligands_monocle(
  obj,
  sender_obj = NULL,
  cds = NULL,
  sender_cluster_key = NULL,
  sender_cluster_names = NULL,
  num_dim = 10,
  root_cells = NULL,
  root_pr_nodes = NULL,
  expressed_ligands = NULL,
  path_cell_names = NULL,
  path_nodes = NULL,
  minimal_branch_len = 5,
  ncenter = 50,
  use_partition = TRUE,
  precomputed_umap = NULL,
  prune_graph = T,
  close_loop = F,
  overwrite = FALSE,
  reduction_key = "MonocleUMAP_",
  sender_reduction_key = "umap",
  expression_proportion_cutoff = 0.01,
  covariance_cutoff = 0.05,
  metric = "Covariances",
  lr_network,
  ligand_target_matrix,
  export_cds = TRUE
)

Arguments

obj: A seurat object containing receiver cells undergoing a trajectory
sender_obj: A seurat obj containing sender cells expressing ligands. Set to NULL if your sender cells are contained in 'obj'.
cds: Optional. You can supply a monocle3 cds object here if you have run monocle3::order_cells() on it already.
sender_cluster_key: Column name in metadata that denotes where sender cells are specified. If NULL, this defaults to the default Seurat Idents given by levels(obj)
sender_cluster_names: Unique values in column sender_cluster_key that denote sender cell clusters.
num_dim: Number of dimensions for monocle reduce_dimension function.
root_cells: Optional. Character vector of root cell names that comprise the beginning of the trajectory (must be in colnames(obj)). If not given or NULL, will open a window to interactively select a root cell (Monocle's default behaviour).
root_pr_nodes: Argument for monocle3::order_cells(). Use if you want to specify root nodes programmatically.
expressed_ligands: Character vector of active ligands in dataset (must be in rownames(obj))
path_cell_names: Optional. Character vector of cell names that comprise the differentiation trajectory to be analyzed by Entrain (must be in colnames(obj)). If not given or NULL, will open a window to interactively select the trajectory path to analyze (Monocle's default behaviour).
path_nodes: Use if you want to specify the path without user interaction. e.g. path_nodes = c("Y_2", "Y_50") tells Entrain to analyze cells between the nodes Y_2 and Y_50
minimal_branch_len: Parameter for Monocle graph learning.
ncenter: Parameter for Monocle graph learning.
use_partition: Parameter for Monocle graph learning
precomputed_umap: If you have a precomputed UMAP on your receiver cells that you wish to build a monocle trajectory on, then set this to the name of your precomputed embedding e.g. precomputed_umap = 'umap'
prune_graph: Parameter for Monocle graph learning.
close_loop: Parameter for Monocle graph learning.
overwrite: If there is an existing Monocle trajectory in the obj (e.g. if you have run this function before), overwrite = TRUE will delete the old trajectory. Otherwise, the old trajectory will be used for Entrain.
reduction_key: Seurat reduction key for dimension reduction visualization.
sender_reduction_key: Only applicable if sender_obj is not NULL: The dimension reduction key of sender_obj that you wish to use,
expression_proportion_cutoff: Pct cutoff to threshold a ligand as active. A ligand is 'active' if it is expressed in more than expression_proportion_cutoff fraction of cells in the sender cluster.
covariance_cutoff: Remove the bottom covariance_cutoff fraction of covariances (e.g. 0.10 = bottom 10 percent of genes, ranked by covariance, removed from later analysis).
metric: One of "Covariances" or "Correlations". Default "Covariances". Determines whether to use pseudotime covariance or pseudotime correlation in the calculation of TRAINing genes.
lr_network: NicheNet ligand-receptor pairs data file.
ligand_target_matrix: NicheNet ligand-target data file.
export_cds: If TRUE, saves the Monocle cell_data_set trajectory object in obj@misc$monocle_graph. Set TRUE if you want to plot Entrain results afterwards.

Value

a Seurat object with ligand velocity results in obj$misc$entrain$velocity_result.