testPrior: Posterior check of the model

Description

This functions compares variance estimates obtained from the maximum a posterior estimate with a given prior to the data. The ratio between the predicted variance and the actual variance for a random subset of genes is computed across all pruned k nearest neighbourhoods.

Usage

testPrior(
  res,
  expData,
  gamma = c(0.2, 0.5, 1, 5, 1000),
  rseed = 12345,
  ngenes = 200,
  pvalue = 0.01,
  minN = 5,
  no_cores = NULL,
  x0 = 0,
  lower = 0,
  upper = 100
)

Value

List of three components:

pp.var.ratio: List of vectors for each gamma value of ratios between predicted and actual variances across all sampled genes and neighbourhoods.
noise: List of noise objects obtained from compTBNoise for each gamma value.
tc: Vector of total transcript counts for all cells

Arguments

res: List object with k nearest neighbour information returned by pruneKnn.
expData: Matrix of gene expression values with genes as rows and cells as columns. These values have to correspond to unique molecular identifier counts.
gamma: Vector of gamma-values to test for the Cauchy prior distribution. Default is c(0.2,0.5,1,5,1000). Large values correspond to weak priors (gamma=1000 corresponds to a maximum likelihood estimate).
rseed: Integer number. Random seed to enforce reproducible gene sampling. Default is 12345.
ngenes: Positive integer number. Randomly sampled number of genes (from rownames of expData) used for noise estimation. Genes are sampled uniformly across the entire expression range. Default is 200.
pvalue: Input parameter for compTBNoise. See help(compTBNoise).
minN: Input parameter for compTBNoise. See help(compTBNoise).
no_cores: Input parameter for compTBNoise. See help(compTBNoise).
x0: Input parameter for compTBNoise. See help(compTBNoise).
lower: Input parameter for compTBNoise. See help(compTBNoise).
upper: Input parameter for compTBNoise. See help(compTBNoise).