switchde: Switch-like model fitting and differential expression test

In switchde: Switch-like differential expression across single-cell trajectories

Description

Fit sigmoidal differential expression models to gene expression across pseudotime. Parameter estimates are returned along with a p-value for switch-like differential expression over a null model (constant expression).

Usage

switchde(object, pseudotime = NULL, zero_inflated = FALSE,
  lower_threshold = 0.01, maxiter = 1000, log_lik_tol = 0.01,
  verbose = FALSE, sce_assay = "exprs")

Arguments

object	Gene expression data that is either A vector of length number of cells for a single gene A matrix of dimension number of genes x number of cells An object of class SingleCellExperiment from package SingleCellExperiment
pseudotime	A pseudotime vector with a pseudotime corresponding to every cell. Can be NULL if object is of class SCESet and colData(sce)$pseudotime is defined.
zero_inflated	Logical. Should zero inflation be implemented? Default FALSE
lower_threshold	The minimum threshold below which to set expression to zero to avoid numerical issues. Default is 0.01
maxiter	Maximum number of iterations for EM algorithm if zero inflation enabled. Default 100
log_lik_tol	If the change in the log-likelihood falls below this for zero inflated EM the algorithm is assumed to have converged
verbose	Print convergence update for EM algorithm
sce_assay	The assay from the SingleCellExperiment to be used as expression, defaulting to "exprs"

Value

A matrix where each column corresponds to a gene, the first row is the p-value for that gene and subsequent rows are model parameters.

Examples

data(synth_gex)
data(ex_pseudotime)
sde <- switchde(synth_gex, ex_pseudotime)

switchde documentation built on Nov. 8, 2020, 5:19 p.m.