sampleplot {globaltest} | R Documentation |
Produces a plot to show the influence of individual
samples on the test result produced by globaltest
.
sampleplot(gt, geneset = NULL, samplesubset = NULL, drawlabels = TRUE, labelsize = 0.6, ...)
gt |
The output of a call to globaltest . |
geneset |
The name or number of the geneset to be plotted (only necessary if multiple genesets were tested). |
samplesubset |
A vector of names or numbers of samples to be plotted (default: all samples) |
drawlabels |
Logical value to control drawing of the samplenames on the x-axis of the plot. |
labelsize |
Relative size of the labels on the x-axis. If it is NULL ,
the current value for par("cex.axis") is used |
... |
Any extra arguments will be forwarded to the plotting function. |
The sampleplot shows a bar and a reference line for each sample. The bar shows the influence of each gene on the test statistic. Samples with a positive influence carry evidence against the null hypothesis (in favour of a significant p-value), because they are are similar in expression profile to samples with a similar clinical outcome. Samples with a negative influence bar supply evidence in favour of the null hypothesis and of a non-significant p-value: they are relatively similar in expression profile to samples with a different clinical outcome.
The influence varies around zero if the tested geneset is not associated with the outcome. Marks on the bars show the standarddeviation of the influence under the null hypothesis for those samples which are more than one standard deviation away from zero.
The color of the bar indicates the sign of the residual of Y. In a logistic model the coloring this distinguishes the original groups.
The bottom margin is adjusted to allow enough space for the longest samplename to draw under the axis.
If samplenames were supplied in the call to
globaltest
, sampleplot returns the vector of
samplenames corresponding to the samplenumbers appearing in the
plot. Otherwise NULL
.
Jelle Goeman: j.j.goeman@lumc.nl; Jan Oosting
J. J. Goeman, S. A. van de Geer, F. de Kort and J. C. van Houwelingen, 2004, A global test for groups of genes: testing association with a clinical outcome, Bioinformatics 20 (1) 93–99. See also the How To Globaltest.pdf included with this package.
globaltest
, geneplot
,
regressionplot
, checkerboard
.
if (interactive()){ data(exampleX) # Expression data (40 samples; 1000 genes) data(exampleY) # Clinical outcome for the 40 samples pathway <- 1:25 # A pathway contains genes 1 to 25 gt <- globaltest(exampleX, exampleY, test.genes = pathway) gt sampleplot(gt) }