CNOGpro (version 1.1)

runBootstrap: Gene copy numbers with confidence intervals through bootstrapping

Description

For each genetic element in the gene table of a CNOGpro object, sample repeatedly from the read observations associated with that gene while replacing the observations.

Usage

runBootstrap(experiment, replicates = 1000, quantiles = c(0.025, 0.975))

Arguments

experiment

An object of class CNOGpro

replicates

The (integer) number of resampling experiments per genetic element. (That is, the number of bootstrap samples.)

quantiles

The quantiles of the distribution of mean copy numbers for which the mean is returned. (Given as c(lower, upper))

Value

An object of class CNOGpro, with proposed copy number (floating point) and the associated confidence interval for each genetic element in the table of genes.

Details

Implements the theory that since the read counts from each genetic element come from a (overdispersed) Poisson distribution with rate parameter lambda (=mean), this lambda can be inferred by simply taking the mean of the observations from that genetic element. A confidence interval is then constructed around the mean by random resampling with replacement. For each replicate and genetic element, the same number of samples is taken as there are read count observations associated with that particular genetic element. The observations of means are then sorted and the given lower and upper quantiles are returned.

Examples

data(carsonella)
carsonella_normalized <- normalizeGC(carsonella)
carsonella_bootstrapped <- runBootstrap(carsonella, replicates=500)
printCNOGpro(carsonella_bootstrapped)