Simulate a complete time course of metabolic labeling - nucleotide conversion RNA-seq data.

This function takes a vector of true synthesis rates and RNA half-lives, and then simulates data for multiple time points and replicates. Both synthesis rate and RNA half-lives are assumed to be constant, but the system might not be in steady-state.

Usage

SimulateTimeCourse(
  condition,
  gene.info,
  s,
  d,
  f0 = s/d,
  s.variation = 1,
  d.variation = 1,
  dispersion,
  num.reads = 1e+07,
  timepoints = c(0, 0, 0, 1, 1, 1, 2, 2, 2, 4, 4, 4),
  beta.approx = FALSE,
  conversion.reads = FALSE,
  verbose = TRUE,
  seed = NULL,
  ...
)

Arguments

condition: A user-defined condition name (which is placed into the Coldata of the final grandR object)
gene.info: either a data frame containing gene annotation or a vector of gene names
s: a vector of synthesis rates
d: a vector of degradation rates (to get a specific half-life HL, use d=log(2)/HL)
f0: the abundance at time t=0
s.variation: biological variability of s among all samples (see details)
d.variation: biological variability of d among all samples (see details)
dispersion: a vector of dispersion parameters (estimate from data using DESeq2, e.g. by the estimate.dispersion utility function)
num.reads: a vector representing the number of reads for each sample
timepoints: a vector representing the labeling duration (in h) for each sample
beta.approx: should the beta approximation of the NTR posterior be computed?
conversion.reads: also output the number of reads with conversion
verbose: Print status updates
seed: seed value for the random number generator (set to make it deterministic!)
...: provided to SimulateReadsForSample

Value

a grandR object containing the simulated data in its data slots and the true parameters in the gene annotation table

Details

If s.variation or d.variation are > 1, then for each gene a random gaussian is added to s (or d) such that 90 of the gaussian is log2(s.variation).