tTestPerRow.bootstrap.VolcanoBoxPlot: Volcano box plot from the result of multiple student tests...
In jvanheld/stats4bioinfo: Utilities for the book "Statistics for bioinformatics"
Description
Usage
Arguments
Value
Author(s)
Examples
Description
Draw a volcano box plot from the object returned by
stats4bioinfo::tTestPerRow.bootstrap(). This representation combines the features
of a volcano plot (significance versus effect size) with the box plots, to indicate
the dispersion (inter-quartille range) and extent of the bootstrapped values.
The qbcissa represents the mean difference between groups.
The ordinate the significance of the Welch test (sig=-log10(e-value)).
Each bootstrapped test is represented by a 2-dimensional box plot, where thick bars
denotes the median, the rectangle delimits the inter-quartile ranges, and the lines
indicate the extent of the bootstrapped values.
Usage
1
2
3
4
5
6
7
8
9
10
tTestPerRow.bootstrap.VolcanoBoxPlot(ttpr.result, control.type = "fdr",
support.quantile = ttpr.result$support.quantile, plot.sig.boxes = TRUE,
plot.effect.boxes = TRUE, plot.rectangles = TRUE, cex = 0.7,
xlab = "Effect size", ylab = paste(sep = "", "-log10(", control.type,
")"), density.colors = FALSE, col.boxplots = "#888888",
col.positive = "#4455DD", col.grid = "#BBBBBB", col.lines = "blue",
col.alpha = "darkred", lty.positive = "solid", lty.negative = "solid",
lty.alpha = "dashed", lwd.rectangles = 1, lwd.iqr = 3, lwd.extent = 1,
lwd.support.quantile = 1, support.ticks = 0.03, xlim = NULL,
ylim = NULL, legend.corner = "topleft", legend.cex = 0.8, ...)
Arguments
ttpr.result
Must be the data frame obtained with tTestPerRow.bootstrap()
cex
Point size, passed to plot() and lines()
...
Additional parameters are passed to plot()
control.type="fdr"
Type of control (supported: "fdr", "e.value", "p.value")
support.quantile=ttpr.result$support.quantile
Support quantile, i.e. the minimal fraction of significant bootstrap iterations required to declare a feature positive. By default, this value is read from the input tTestPerRow() result object. However, it can be over-written in order to estimate the impact of the threshold on support quantile on the result, without having to re-run the analysis.
plot.sig.boxes=TRUE
Draw boxplots for the significance (Y axis)
plot.effect.boxes=TRUE
Draw boxplots for the effect size (X axis)
plot.rectangles=TRUE
Draw inter-quartile rectangles, i.e. rectangles whose corners are defined by the first and third quartiles, on both axes.
density.colors=FALSE
Automatically set the colors according to feature status and local density in the Volcano space.
col.boxplots='#888888'
Color(s) for the boxplots. can be either a single value (same color for all boxplots), or a vector of the same length as the numer of multiple tests (rows of the result tables).
col.positive='#4455DD'
Color to highlight significant points (called positive).
col.lines='blue'
Color for the line highlighting the significance threshold
col.alpha="darkred"
Color for the line highlighting the significance threshold.
col.grid=#BBBBBB
Grid color
lty.positive=3
Line types for the positive tests.
lty.negative=1
Line types for the negative tests.
lty.alpha="dashed"
Line type for the horizontal line denoting the alpha theshold.
lwd.rectangles=1
Line width for the IQR rectangles. Set to 0 to avoid drawing rectangle.
lwd.iqr=3
Line width for the inter-quartile bars (thick bars from Q1 to Q"). Set to 0 to avoid drawing inter-quartile bars
lwd.extent=1
Line width for the extent lines Set to 0 to avoid drawing median bars
lwd.support.quantile=1
Line width for the ticks denoting support quantiles, i.e. the significance corresponding to the specified support quantile.
support.ticks=0.03
Width of the ticks denoting the support cutoff. Set to 0 to avoid drawing support cutoff.
xlab="Effect
size" Label for the X axis
ylab="-log10(p-value)"
Label for the Y axis
xlim=NULL
Range of the X axis. Automatically computed if NULL (default).
ylim=NULL
Range of the Y axis. Automatically computed if NULL (default).
legend.corner="bottomleft"
Corner for the legend. When NULL, no legend is displayed.
legend.cex=1
Font size for the legend.
Value
no return object
Author(s)
Jacques van Helden (Jacques.van-Helden@univ-amu.fr)
Examples
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
## Generate a random set with two samples from distinct
## populations A and B, characterized by and m rows (features),
## among which m1 are under H1 (mA != mB) and m0 under
## H0 (mA = mB).
m1 <- 500 ## Number of features under H_1
m0 <- 500 ## Number of features under H_0
m <- m0+m1
bootstrap.iterations <- 100
effect.size <- 1 ## Difference between means for the features under H1
sample.sizes <- c("A"=30,"B"=30)
sample.labels <- rep(names(sample.sizes), times=sample.sizes)
table(sample.labels)
row.means.g1 <- rep(0, m) ## All rows have a null mean for group 1
row.means.g2 <- c(rep(0+effect.size, m1), rep(0, m0)) ## group 2 mean differs depending on the row
rand.2grp <- data.frame(cbind(
matrix(nrow=m, ncol=sample.sizes[1], rnorm(m=row.means.g1,sd=1,n=sample.sizes[1]*m)),
matrix(nrow=m, ncol=sample.sizes[2], rnorm(m=row.means.g2,sd=1,n=sample.sizes[2]*m))))
## Apply Student t-test with classical bootstrap
## (draw samples of same size as original groups, with replacement).
## Set the support quantile to 0.9, in order to select only the features
## passing the alpha threshold in at least 90% of the iteratios.
student.bootstrap <- tTestPerRow.bootstrap(x = rand.2grp, cl=sample.labels,
iterations=bootstrap.iterations, var.equal=TRUE, support.quantile=0.9)
## Draw the volcano box plot with default parameters
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap)
## Draw the volcano box plot with max stringency on support:
## only consider as positive the rows declared positive in all bootstrap iterations
student.bootstrap.100pcsupport <- tTestPerRow.bootstrap(x = rand.2grp, cl=sample.labels,
iterations=bootstrap.iterations, var.equal=TRUE, support.quantile=1)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap.100pcsupport)
## Only draw extent lines and median bars (from quartile 1 to quartile 3) for effect size (horizontal bars and lines)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=3,
plot.effect.boxes=TRUE, plot.sig.boxes=FALSE)
## Only draw extent lines for effect size
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=0,
plot.effect.boxes=TRUE, plot.sig.boxes=FALSE)
## Only draw median bars, for both effect size (X axis) and significance (Y axis)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=0, lwd.rectangles=0, lwd.iqr=3)
## "Significance bootstrap volano plot": a volcano plot indicating the fluctuations of the significance across the bootstrap iterations
## Compare bootstrap volcanos with different control types
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=3, support.ticks=0.02,
control.type="p.value", plot.effect.boxes=FALSE, plot.sig.boxes=TRUE, plot.rectangles=FALSE, xlim=c(-0.7,1.7))
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=3, support.ticks=0.02,
control.type="fdr", plot.effect.boxes=FALSE, plot.sig.boxes=TRUE, plot.rectangles=FALSE, xlim=c(-0.7,1.7))
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=3, support.ticks=0.02,
control.type="e.value", plot.effect.boxes=FALSE, plot.sig.boxes=TRUE, plot.rectangles=FALSE, xlim=c(-0.7,1.7))
## Show the options to select subset of graphical elements (rectangles, box plots on each axis).
## For the sake of comparison, first plot Student t-test on the full dataset (not bootstrapped) + a Volcano CI plot (confidence intervals).
## which is included in the result.
x11(width=8, height=12)
par(mfrow=c(3,2))
par(mar=c(4,4,2,0.5))
tTestPerRow.plotVolcano(student.bootstrap$full.set.test, control.type="p.value", legend.corner="topleft", main="Volcano plot")
tTestPerRow.plotVolcano(student.bootstrap$full.set.test, control.type="p.value", legend.corner="topleft", plot.ci=TRUE, main="Confidence volcano")
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, control.type="p.value", legend.corner="topleft",
main="Effect size bootstrap volcano", plot.effect.boxes=TRUE, plot.sig.boxes=FALSE, plot.rectangles=FALSE)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, control.type="p.value", legend.corner="topleft",
main="Significance bootstrap volcano", plot.effect.boxes=FALSE, plot.sig.boxes=TRUE, plot.rectangles=FALSE)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, control.type="p.value", legend.corner="topleft",
main="Bootstrap volcano: quartile rectangles", plot.effect.boxes=FALSE, plot.sig.boxes=FALSE, plot.rectangles=TRUE)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, control.type="p.value", legend.corner="topleft",
main="Bootstrap volcano: box plots", plot.effect.boxes=TRUE, plot.sig.boxes=TRUE, plot.rectangles=TRUE)
par(mfrow=c(1,1))
jvanheld/stats4bioinfo documentation built on May 20, 2019, 5:16 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Description Usage Arguments Value Author(s) Examples
Description
Draw a volcano box plot from the object returned by stats4bioinfo::tTestPerRow.bootstrap(). This representation combines the features of a volcano plot (significance versus effect size) with the box plots, to indicate the dispersion (inter-quartille range) and extent of the bootstrapped values. The qbcissa represents the mean difference between groups. The ordinate the significance of the Welch test (sig=-log10(e-value)). Each bootstrapped test is represented by a 2-dimensional box plot, where thick bars denotes the median, the rectangle delimits the inter-quartile ranges, and the lines indicate the extent of the bootstrapped values.
Usage
1 2 3 4 5 6 7 8 9 10 | tTestPerRow.bootstrap.VolcanoBoxPlot(ttpr.result, control.type = "fdr",
support.quantile = ttpr.result$support.quantile, plot.sig.boxes = TRUE,
plot.effect.boxes = TRUE, plot.rectangles = TRUE, cex = 0.7,
xlab = "Effect size", ylab = paste(sep = "", "-log10(", control.type,
")"), density.colors = FALSE, col.boxplots = "#888888",
col.positive = "#4455DD", col.grid = "#BBBBBB", col.lines = "blue",
col.alpha = "darkred", lty.positive = "solid", lty.negative = "solid",
lty.alpha = "dashed", lwd.rectangles = 1, lwd.iqr = 3, lwd.extent = 1,
lwd.support.quantile = 1, support.ticks = 0.03, xlim = NULL,
ylim = NULL, legend.corner = "topleft", legend.cex = 0.8, ...)
|
Arguments
ttpr.result |
Must be the data frame obtained with tTestPerRow.bootstrap() |
cex |
Point size, passed to plot() and lines() |
... |
Additional parameters are passed to plot() |
control.type="fdr" |
Type of control (supported: "fdr", "e.value", "p.value") |
support.quantile=ttpr.result$support.quantile |
Support quantile, i.e. the minimal fraction of significant bootstrap iterations required to declare a feature positive. By default, this value is read from the input tTestPerRow() result object. However, it can be over-written in order to estimate the impact of the threshold on support quantile on the result, without having to re-run the analysis. |
plot.sig.boxes=TRUE |
Draw boxplots for the significance (Y axis) |
plot.effect.boxes=TRUE |
Draw boxplots for the effect size (X axis) |
plot.rectangles=TRUE |
Draw inter-quartile rectangles, i.e. rectangles whose corners are defined by the first and third quartiles, on both axes. |
density.colors=FALSE |
Automatically set the colors according to feature status and local density in the Volcano space. |
col.boxplots='#888888' |
Color(s) for the boxplots. can be either a single value (same color for all boxplots), or a vector of the same length as the numer of multiple tests (rows of the result tables). |
col.positive='#4455DD' |
Color to highlight significant points (called positive). |
col.lines='blue' |
Color for the line highlighting the significance threshold |
col.alpha="darkred" |
Color for the line highlighting the significance threshold. |
col.grid=#BBBBBB |
Grid color |
lty.positive=3 |
Line types for the positive tests. |
lty.negative=1 |
Line types for the negative tests. |
lty.alpha="dashed" |
Line type for the horizontal line denoting the alpha theshold. |
lwd.rectangles=1 |
Line width for the IQR rectangles. Set to 0 to avoid drawing rectangle. |
lwd.iqr=3 |
Line width for the inter-quartile bars (thick bars from Q1 to Q"). Set to 0 to avoid drawing inter-quartile bars |
lwd.extent=1 |
Line width for the extent lines Set to 0 to avoid drawing median bars |
lwd.support.quantile=1 |
Line width for the ticks denoting support quantiles, i.e. the significance corresponding to the specified support quantile. |
support.ticks=0.03 |
Width of the ticks denoting the support cutoff. Set to 0 to avoid drawing support cutoff. |
xlab="Effect |
size" Label for the X axis |
ylab="-log10(p-value)" |
Label for the Y axis |
xlim=NULL |
Range of the X axis. Automatically computed if NULL (default). |
ylim=NULL |
Range of the Y axis. Automatically computed if NULL (default). |
legend.corner="bottomleft" |
Corner for the legend. When NULL, no legend is displayed. |
legend.cex=1 |
Font size for the legend. |
Value
no return object
Author(s)
Jacques van Helden (Jacques.van-Helden@univ-amu.fr)
Examples
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 | ## Generate a random set with two samples from distinct
## populations A and B, characterized by and m rows (features),
## among which m1 are under H1 (mA != mB) and m0 under
## H0 (mA = mB).
m1 <- 500 ## Number of features under H_1
m0 <- 500 ## Number of features under H_0
m <- m0+m1
bootstrap.iterations <- 100
effect.size <- 1 ## Difference between means for the features under H1
sample.sizes <- c("A"=30,"B"=30)
sample.labels <- rep(names(sample.sizes), times=sample.sizes)
table(sample.labels)
row.means.g1 <- rep(0, m) ## All rows have a null mean for group 1
row.means.g2 <- c(rep(0+effect.size, m1), rep(0, m0)) ## group 2 mean differs depending on the row
rand.2grp <- data.frame(cbind(
matrix(nrow=m, ncol=sample.sizes[1], rnorm(m=row.means.g1,sd=1,n=sample.sizes[1]*m)),
matrix(nrow=m, ncol=sample.sizes[2], rnorm(m=row.means.g2,sd=1,n=sample.sizes[2]*m))))
## Apply Student t-test with classical bootstrap
## (draw samples of same size as original groups, with replacement).
## Set the support quantile to 0.9, in order to select only the features
## passing the alpha threshold in at least 90% of the iteratios.
student.bootstrap <- tTestPerRow.bootstrap(x = rand.2grp, cl=sample.labels,
iterations=bootstrap.iterations, var.equal=TRUE, support.quantile=0.9)
## Draw the volcano box plot with default parameters
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap)
## Draw the volcano box plot with max stringency on support:
## only consider as positive the rows declared positive in all bootstrap iterations
student.bootstrap.100pcsupport <- tTestPerRow.bootstrap(x = rand.2grp, cl=sample.labels,
iterations=bootstrap.iterations, var.equal=TRUE, support.quantile=1)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap.100pcsupport)
## Only draw extent lines and median bars (from quartile 1 to quartile 3) for effect size (horizontal bars and lines)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=3,
plot.effect.boxes=TRUE, plot.sig.boxes=FALSE)
## Only draw extent lines for effect size
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=0,
plot.effect.boxes=TRUE, plot.sig.boxes=FALSE)
## Only draw median bars, for both effect size (X axis) and significance (Y axis)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=0, lwd.rectangles=0, lwd.iqr=3)
## "Significance bootstrap volano plot": a volcano plot indicating the fluctuations of the significance across the bootstrap iterations
## Compare bootstrap volcanos with different control types
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=3, support.ticks=0.02,
control.type="p.value", plot.effect.boxes=FALSE, plot.sig.boxes=TRUE, plot.rectangles=FALSE, xlim=c(-0.7,1.7))
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=3, support.ticks=0.02,
control.type="fdr", plot.effect.boxes=FALSE, plot.sig.boxes=TRUE, plot.rectangles=FALSE, xlim=c(-0.7,1.7))
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, lwd.extent=1, lwd.rectangles=0, lwd.iqr=3, support.ticks=0.02,
control.type="e.value", plot.effect.boxes=FALSE, plot.sig.boxes=TRUE, plot.rectangles=FALSE, xlim=c(-0.7,1.7))
## Show the options to select subset of graphical elements (rectangles, box plots on each axis).
## For the sake of comparison, first plot Student t-test on the full dataset (not bootstrapped) + a Volcano CI plot (confidence intervals).
## which is included in the result.
x11(width=8, height=12)
par(mfrow=c(3,2))
par(mar=c(4,4,2,0.5))
tTestPerRow.plotVolcano(student.bootstrap$full.set.test, control.type="p.value", legend.corner="topleft", main="Volcano plot")
tTestPerRow.plotVolcano(student.bootstrap$full.set.test, control.type="p.value", legend.corner="topleft", plot.ci=TRUE, main="Confidence volcano")
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, control.type="p.value", legend.corner="topleft",
main="Effect size bootstrap volcano", plot.effect.boxes=TRUE, plot.sig.boxes=FALSE, plot.rectangles=FALSE)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, control.type="p.value", legend.corner="topleft",
main="Significance bootstrap volcano", plot.effect.boxes=FALSE, plot.sig.boxes=TRUE, plot.rectangles=FALSE)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, control.type="p.value", legend.corner="topleft",
main="Bootstrap volcano: quartile rectangles", plot.effect.boxes=FALSE, plot.sig.boxes=FALSE, plot.rectangles=TRUE)
tTestPerRow.bootstrap.VolcanoBoxPlot(student.bootstrap, control.type="p.value", legend.corner="topleft",
main="Bootstrap volcano: box plots", plot.effect.boxes=TRUE, plot.sig.boxes=TRUE, plot.rectangles=TRUE)
par(mfrow=c(1,1))
|
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.