diagplot.metaseqr: Diagnostic plots for the metaseqR package

In pmoulos/metaseqr: An R package for the analysis and result reporting of RNA-Seq data by combining multiple statistical algorithms.

Description

This is the main function for producing sructured quality control and informative graphs base on the results of the various steps of the metaseqR package. The graphs produced span a variety of issues like good sample reproducibility (Multi-Dimensional Scaling plot, biotype detection, heatmaps. diagplot.metaseqr, apart from implementing certain package-specific plots, is a wrapper around several diagnostic plots present in other RNA-Seq analysis packages such as EDASeq and NOISeq.

Usage

    diagplot.metaseqr(object, sample.list, annotation = NULL,
        contrast.list = NULL, p.list = NULL,
        thresholds = list(p = 0.05, f = 1),
        diagplot.type = c("mds", "biodetection", "countsbio", "saturation", 
        "readnoise", "rnacomp", "correl", "pairs", "boxplot", "gcbias", 
        "lengthbias", "meandiff", "meanvar", "deheatmap", "volcano", 
        "biodist", "filtered", "venn"),
        is.norm = FALSE, output = "x11", path = NULL, ...)

Arguments

object	a matrix or a data frame containing count data derived before or after the normalization procedure, filtered or not by the metaseqR's filters and/or p-value. The object can be fed to any of the diagplot.metaseqr plotting systems but not every plot is meaningful. For example, it's meaningless to create a "biodist" plot for a count matrix before normalization or statistical testing.
sample.list	the list containing condition names and the samples under each condition.
annotation	a data frame containing annotation elements for each row in object. Usually, a subset of the annotation obtained by get.annotation or a subset of possibly embedded annotation with the input counts table. This parameter is optional and required only when diagplot.type is any of "biodetection", "countsbio", "saturation", "rnacomp", "readnoise", "biodist", "gcbias", "lengthbias" or "filtered".
contrast.list	a named structured list of contrasts as returned by make.contrast.list or just the vector of contrasts as defined in the main help page of metaseqr. This parameter is optional and required only when diagplot.type is any of "deheatmap", "volcano" or "biodist".
p.list	a list of p-values for each contrast as obtained from any of the stat.* methods of the metaseqr package. This parameter is optional and required only when diagplot.type is any of "deheatmap", "volcano" or "biodist".
thresholds	a list with the elements "p" and "f" which are the p-value and the fold change cutoff when diagplot.type="volcano".
diagplot.type	one or more of the diagnostic plots supported in metaseqR package. Many of these plots require the presence of additional package, something that is checked while running the main metaseqr function. The supported plots are "mds", "biodetection", "countsbio", "saturation", "rnacomp", "boxplot", "gcbias", "lengthbias", "meandiff", "meanvar", "deheatmap", "volcano", "biodist", "filtered", "readnoise", "venn", "correl", "pairwise". For a brief description of these plots please see the main metaseqr help page.
is.norm	a logical indicating whether object contains raw or normalized data. It is not essential and it serves only plot annotation purposes.
output	one or more R plotting device to direct the plot result to. Supported mechanisms: "png", "jpg", "bmp", "pdf", "ps" or "json". The latter is currently available for the creation of interactive volcano plots only when reporting the output, through the highcharts javascript library. The default plotting ("x11") is not supported due to instability in certain devices.
path	the path to create output files.
...	further arguments to be passed to plot devices, such as parameter from par.

Value

A named list containing the file names of the produced plots. Each list member is names according to the selected plotting device and is also a named list, whose names are the plot types. The final contents are the file names in case the plots are written to a physical location (not meaningful for "x11").

Note

In order to make the best out of this function, you should generally provide the annotation argument as most and also the most informative plots depend on this. If you don't know what is inside your counts table or how many annotation elements you can provide by embedding it, it's always best to set the annotation parameter of the main metaseqr function to "download" to use predefined annotations that work better with the functions of the whole package.

Author(s)

Panagiotis Moulos

Examples

require(DESeq)
data.matrix <- counts(makeExampleCountDataSet())
sample.list <- list(A=c("A1","A2"),B=c("B1","B2","B3"))
contrast <- "A_vs_B"
diagplot.metaseqr(data.matrix,sample.list,diagplot.type=c("mds","boxplot"))

norm.args <- get.defaults("normalization","deseq")
object <- normalize.deseq(data.matrix,sample.list,norm.args)
diagplot.metaseqr(object,sample.list,diagplot.type="boxplot")

## More
#p <- stat.deseq(object,sample.list)
#diagplot.metaseqr(object,sample.list,contrast.list=contrast,p.list=p,
#        diagplot.type="volcano")

pmoulos/metaseqr documentation built on Dec. 29, 2020, 5:56 a.m.