SigFuge (version 1.4.0)

SFfigure: Plot expression as curves

Description

Function for producing various figures corresponding to the SigFuge functional data approach to studying RNA-seq data as expression curves along base positions. The primary input for the function is a read count matrix and GRanges. The default behavior is to identify clusters based on applying SFlabels to a normalized version of the data produced by SFnormalize. If specified, the function will compute a p-value for the significance of the labels by calling the SFpval function.

Usage

SFfigure(data, locusname, annot = c(), flip.fig = 1, label.exon = 1, print.n = 1, data.labels = 0, label.colors = c(), flag = 1, lplots = 2, log10 = 1, summary.type = "median", savestr = c(), titlestr = c(), pval = 1)

Arguments

data

a $d x n$ matrix or data.frame of read counts at $d$ base positions for $n$ samples.

locusname

a character string specifying gene or locus name to be used in figure title.

annot

a GRanges object or data.frame including annotation information for locus, including:

• start start of contiguous genomic regions
• end end of contiguous genomic regions
• seqname chromosome name for genomic region
• strand strandedness of sequence

flip.fig

an indicator whether to flip the plotting direction of the locus if strand == "-" when annotation information is provided.

label.exon

an indicator whether to print the exon boundaries to the figure.

print.n

an indicator whether to print cluster sizes.

data.labels

a $n x 1$ vector of class labels to use instead of calcuating SigFuge labels

label.colors

a $K x 3$ matrix of RBG colors specifying cluster colors for $K$ clusters. ggplot2 default colors are used if not specified. If using SigFuge default labels, $K=3$ even if no low expression samples are flagged.

flag

a $n x 1$ logical vector of samples flagged as low expression. If flag == 1, default low expression cutoffs are used. If flag == 0, no samples are flagged as low expression (equivalent to setting flag = rep(0, n)).

lplots

a specification of which figures to output

• 1: curves in single panel, random colors
• 2: curves in single panel, colored by cluster
• 3: curves in $K$ panels, separated and colored by cluster
• 4: curves in $n$ panels, colored by cluster (single sample per panel)
• 5: cluster medians in single panel, colored by cluster

log10

an indicator whether the y-axis (read depth) should be log10 transformed. Default is to plot on log-scale.

summary.type

a character string specifying which summary statistic should be used when plotting clusters in lplots == 2, 3, and 5. Options: "median" (default) or "mean".

savestr

a string specifying the file name for resulting figures. Extensions can also be specified in savestr. If no extension is specified figures will be saved as pdfs. If length(lplots) > 1, figures will be saved as paste0(savestr,"_x") for x in lplots with the appropirate extension. If no savestr is specified, function will return a list containing the created ggplot objects.

titlestr

a string specifying figure title. If unspecified, default is titlestr=paste(locusname," locus, SigFuge analysis").

pval

an indicator whether the SFpval should be computed. If pval == 1, the p-value is added to the title, i.e. (titlestr=paste0(titlestr, ", p-value = ", p)).

Value

SFfigure returns a figure that is saved to the current working directory if a savestr is specified. Else, a list containing the plots is returned.

Examples

# load data
data(geneAnnot)
data(geneDepth)

# only use first 50 samples
mdata <- geneDepth[,1:50]

# make plot
locusname <- "CDKN2A"
SFfigure(mdata, locusname, geneAnnot, flag=1,
 lplots=3, savestr=paste0(locusname,".pdf"), titlestr="CDKN2A locus, LUSC samples",
 pval=1)

mySFs <- SFfigure(mdata, locusname, geneAnnot, flag=1,
            lplots=1, savestr=c(), titlestr="CDKN2A locus, LUSC samples not saved",
            pval=0)
mySFs$plot1