goodnessOfFit: Binomial and poisson goodness of fit statistics for BSSeq...
In bsseq: Analyze, manage and store bisulfite sequencing data

Description Usage Arguments Details Value Author(s) See Also Examples

Binomial and poisson goodness of fit statistics for BSSeq objects, including plotting capability.

poissonGoodnessOfFit(BSseq, nQuantiles = 10^5)
binomialGoodnessOfFit(BSseq, method = c("MLE"), nQuantiles = 10^5)
## S3 method for class 'chisqGoodnessOfFit'
print(x, ...)
## S3 method for class 'chisqGoodnessOfFit'
plot(x, type = c("chisq", "pvalue"), plotCol = TRUE, qqline = TRUE,
  pch = 16, cex = 0.75, ...)

`BSseq`	An object of class `BSseq`.
`x`	A chisqGoodnessOfFit object (as produced by `poissonGoodnessOfFit` or `binomialGoodnessOfFit`).
`nQuantiles`	The number of (evenly-spaced) quantiles stored in the return object.
`method`	How is the parameter estimated.
`type`	Are the chisq or the p-values being plotted.
`plotCol`	Should the extreme quantiles be colored.
`qqline`	Add a `qqline`.
`pch, cex`	Plotting symbols and size.
`...`	Additional arguments being passed to `qqplot` (for `plot`) or ignored (for `print`).

These functions compute and plot goodness of fit statistics for BSseq objects. For each methylation loci, the Poisson goodness of fit statistic tests whether the coverage (at that loci) is independent and identically Poisson distributed across the samples. In a similar fashion, the Binomial goodness of fit statistic tests whether the number of reads supporting methylation are independent and identically binomial distributed across samples (with different size parameters given by the coverage vector).

These functions do not handle NA values.

The plotting method is invoked for its side effect. Both poissonGoodnessOfFit and binomialGoodnessOfFit returns an object of class chisqGoodnessOfFit which is a list with components

`chisq`	a vector of Chisq values.
`quantiles`	a vector of quantiles (of the chisq values).
`df`	degress of freedom

Kasper Daniel Hansen khansen@jhsph.edu

For the plotting method, see qqplot.

if(require(bsseqData)) {
  data(BS.cancer.ex)
  BS.cancer.ex <- updateObject(BS.cancer.ex)
  gof <- poissonGoodnessOfFit(BS.cancer.ex)
  plot(gof)

#-------------------------------------------------------------------------------
# An example using a HDF5Array-backed BSseq object
#

    library(HDF5Array)

    # See ?SummarizedExperiment::saveHDF5SummarizedExperiment for details
    hdf5_BS.cancer.ex <- saveHDF5SummarizedExperiment(x = BS.cancer.ex,
                                                      dir = tempfile())
    hdf5_gof <- poissonGoodnessOfFit(hdf5_BS.cancer.ex)
    plot(hdf5_gof)

}