gen_zinb {SeqNet}

Generate ZINB counts from an underlying network

Description

The count data are generated based on the gene-gene associations of an udnerlying network. An association structure is imposed by first generating data from a multivariate Gaussian distribution, and counts are then obtained through the inverse tranformation method. To generate realistic counts, either a reference dataset or parameters for the ZINB model (size, mu, rho) can be provided. Parameters can be estimated from a reference using the est_params_from_reference function.

Usage

gen_zinb(
  n,
  network,
  reference = NULL,
  params = NULL,
  library_sizes = NULL,
  adjust_library_size = NULL,
  verbose = TRUE
)

Arguments

n

The number of samples to generate.

network

A 'network' object or list of 'network' objects.

reference

Either a vector or data.frame of counts from a reference gene expression profile. If a data.frame is provided, each column should correspond to a gene. If both reference and params are NULL, then parameters are estimated from the reference dataset.

params

A matrix of ZINB parameter values; each column should contain the size, mu, and rho parameters for a gene.

library_sizes

A vector of library sizes. Used only if reference is NULL.

adjust_library_size

A boolean value. If TRUE, the library size of generated counts are adjusted based on the reference library sizes. If both reference and library_sizes is NULL, then no adjustment is made. By default, this adjustment is made if the necessary information is provided.

verbose

Boolean indicator for message output.

Value

A list containing the generated counts and the ZINB parameters used to create them. If a list of networks were provided, then the results for each network are returned as a list.

Examples

nw <- random_network(10) # Create a random network with 10 nodes.
nw <- gen_partial_correlations(nw) # Add weights to connections in the network.
# If no reference is provided, ZINB data are generated using an internal reference.
x <- gen_zinb(20, nw)$x # Simulate 20 ZINB observations from the network.