I am plotting tons of graphs which essentially use the same type of formatting. Just wondering if it possible to store these layers in a variable and reuse them.
Approach 1 (does not work)
t <- layer1() + layer2()
ggplot(df,aes(x,y)) + t
Approach 2 (works but not very elegant)
t <- function(x) x + layer1() + layer2()
t(ggplot(df,aes(x,y))
Any suggestion along the lines of approach 1?
Thanks!
While I wait for some clarification, here are a few examples that demonstrate how to add previously created layers to an existing plot:
p <- ggplot(mtcars,aes(x = cyl,y = mpg)) +
geom_point()
new_layer <- geom_point(data = mtcars,aes(x = cyl,y = hp),colour = "red")
new_layer1 <- geom_point(data = mtcars,aes(x = cyl,y = wt),colour = "blue")
p + new_layer
p + list(new_layer,new_layer1)
Based on the Joran's answer, I now put my layers into a list, and add it in my plots. Works like a charm :
r = data.frame(
time=c(5,10,15,20),
mean=c(10,20,30,40),
sem=c(2,3,1,4),
param1=c("A", "A", "B", "B"),
param2=c("X", "Y", "X", "Y")
)
gglayers = list(
geom_point(size=3),
geom_errorbar(aes(ymin=mean-sem, ymax=mean+sem), width=.3),
scale_x_continuous(breaks = c(0, 30, 60, 90, 120, 180, 240)),
labs(
x = "Time(minutes)",
y = "Concentration"
)
)
ggplot(data=r, aes(x=time, y=mean, colour=param1, shape=param1)) +
gglayers +
labs(
color = "My param1\n",
shape = "My param1\n"
)
ggplot(data=r, aes(x=time, y=mean, colour=param2, shape=param2)) +
gglayers +
labs(
color = "My param2\n",
shape = "My param2\n"
)
I know this is old, but here is one that avoids the clunky t(ggplot(...)))
t<-function(...) ggplot(...) + layer1() + layer2()
t(df, aes(x, y))
Related
I would like to change the y axis label (or main title would also be fine) of a ggplot to reflect the column name being iterated over within an apply function.
Here is some sample data and my working apply function:
trial_df <- data.frame("Patient" = c(1,1,2,2,3,3,4,4),
"Outcome" = c("NED", "NED", "NED", "NED", "Relapse","Relapse","Relapse","Relapse"),
"Time_Point" = c("Baseline", "Week3", "Baseline", "Week3","Baseline", "Week3","Baseline", "Week3"),
"CD4_Param" = c(50.8,53.1,20.3,18.1,30.8,24.5,35.2,31.0),
"CD8_Param" = c(5.3,9.7,4.4,4.3,3.1,3.2,5.6,5.3),
"CD3_Param" = c(11.6,16.6,5.0,5.1,14.3,7.1,5.9,8.1))
apply(trial_df[,4:length(trial_df)], 2, function(i) ggplot(data = trial_df, aes_string(x = "Time_Point", y = i )) +
facet_wrap(~Outcome) +
geom_boxplot(alpha = 0.1) +
geom_point(aes(color = `Outcome`, fill = `Outcome`)) +
geom_path(aes(group = `Patient`, color = `Outcome`)) +
theme_minimal() +
ggpubr::stat_compare_means( method = "wilcox.test") +
scale_fill_manual(values=c("blue", "red")) +
scale_color_manual(values=c("blue", "red")))
Example plot output
This creates 3 graphs as expected, however the y axis just says "y". I would like this to display the column name for the column in that iteration. It would also be fine to add a main title with this information, as I just need to know which graph corresponds to which column.
Here are things I have already tried adding to the ggplot code above based on some similar questions I found, but all of them give me the error "non-numeric argument to binary operator":
ggtitle(paste(i))
labs(y = i)
labs(y = as.character(i))
Any help or resources I may have missed would be greatly appreciated, thanks!
So.....for the strangest of reasons I cannot figure out why. This gives what you want but for only one graph!!!
apply(trial_df[,4:length(trial_df)], 2, function(i) ggplot(data = trial_df, aes_string(x = "Time_Point", y = i )) +
facet_wrap(~Outcome) +
geom_boxplot(alpha = 0.1) +
geom_point(aes(color = `Outcome`, fill = `Outcome`)) +
geom_path(aes(group = `Patient`, color = `Outcome`)) +
theme_minimal() +
stat_compare_means( method = "wilcox.test") +
scale_fill_manual(values=c("blue", "red")) +
scale_color_manual(values=c("blue", "red"))+
labs(y=colnames(trial_df)[i]))
Gives these:
This is a minimal example that shows the plots I am trying to make.
Data looks like this:
plot1 = data.frame(
Factor1 = as.factor(rep('A', 4)),
Factor2 = as.factor(rep(c('C', 'D'), 2)),
Factor3 = as.factor(c( rep('E', 2), rep('F', 2))),
Y = c(0.225490, 0.121958, 0.218182, 0.269789)
)
plot2 = data.frame(
Factor1 = as.factor(rep('B', 4)),
Factor2 = as.factor(rep(c('C', 'D'), 2)),
Factor3 = as.factor(c( rep('E', 2), rep('F', 2))),
Y = c(-0.058585, -0.031686, 0.013141, 0.016249)
)
While the basic code for plotting looks like this:
require(ggplot2)
require(grid)
p1 <- ggplot(data=plot1, aes(x=Factor2, y=Y, fill=factor(Factor3))) +
ggtitle('Type: A') +
coord_cartesian(ylim = c(-0.10, 0.30)) +
geom_bar(position=position_dodge(.9), width=0.5, stat='identity') +
scale_x_discrete(name='Regime',
labels=c('C', 'D')) +
scale_y_continuous('Activations') +
scale_fill_brewer(palette='Dark2', name='Background:',
breaks=c('E','F'),
labels=c('E','F')) +
theme(axis.text=element_text(size=11),
axis.title.x=element_text(size=13, vjust=-0.75),
axis.title.y=element_text(size=13, vjust=0.75),
legend.text=element_blank(),
legend.title=element_blank(),
legend.position='none',
plot.title=element_text(hjust=0.5))
p2 <- ggplot(data=plot2, aes(x=Factor2, y=Y, fill=factor(Factor3))) +
ggtitle('Type: B') +
coord_cartesian(ylim = c(-0.10, 0.30)) +
geom_bar(position=position_dodge(.9), width=0.5, stat='identity') +
scale_x_discrete(name='Regime',
labels=c('C', 'D')) +
scale_y_continuous('Activations') +
scale_fill_brewer(palette='Dark2', name='Background:',
breaks=c('E','F'),
labels=c('E','F')) +
theme(axis.text=element_text(size=11),
axis.title.x=element_text(size=13, vjust=-0.75),
axis.title.y=element_blank(),
legend.text=element_text(size=11),
legend.title=element_text(size=13),
plot.title=element_text(hjust=0.5))
pushViewport(viewport(
layout=grid.layout(1, 2, heights=unit(4, 'null'),
widths=unit(c(1,1.17), 'null'))))
print(p1, vp=viewport(layout.pos.row=1, layout.pos.col=1))
print(p2, vp=viewport(layout.pos.row=1, layout.pos.col=2))
And the figure looks like this:
However, I would need something like this:
Thick black lines are the reference values. They are constant and the Figure presents that "reference situation". However, in other plots that I need to produce bars will change but the reference values should remain the same to make the comparisons straightforward and easy. I know I should be using geom_segment() but those lines in my attempts to make this work are just missing the bars.
Any help/advice? Thanks!
I was able to do this using geom_errorbarh. For instance, with the second figure:
p1 +
geom_errorbarh(
aes(xmin = as.numeric(Factor2)-.2,xmax = as.numeric(Factor2)+.2), #+/-.2 for width
position = position_dodge(0.9), size = 2, height = 0
)
OUTPUT:
And, if I understand the other plots you describe, you can specify the reference data in those, eg data = plot1
If your references are not going to be changed, you can create a second dataset and merge it to the dataset you are going to plot.
Here, I first add plot1 and plot2. Then, I create a new dataset that will be the reference dataset.
library(dplyr)
new_df = rbind(plot1, plot2)
ref_plot = new_df
ref_plot <- ref_plot %>% rename(Ref_value = Y)
Then, now you have the new_df which is the dataset to be plot and ref_plot that contains references values for each conditions.
Instead of using grid and create two different plot that I will merge after, I preferred to use facet_wrap which put all plots on the same figure. It is much more convenient and don't require to write twice the same thing.
As mentioned by #AHart few minutes before me, you can use geom_errorbar to define your reference values on the plot. The difference is I prefere to use geom_errorbar instead of geom_errobarh.
Here is for the plot:
library(ggplot2)
new_df %>% left_join(ref_plot) %>%
ggplot(aes(x = Factor2, y = Y, fill = Factor3))+
geom_bar(stat = "identity", position = position_dodge())+
geom_errorbar(aes(ymin = Ref_value-0.00001, ymax = Ref_value+0.0001, group = Factor3), position = position_dodge(.9),width = 0.2)+
facet_wrap(.~Factor1, labeller = labeller(Factor1 = c(A = "Type A", B = "Type B"))) +
scale_x_discrete(name='Regime',
labels=c('C', 'D')) +
scale_fill_brewer(palette='Dark2', name='Background:',
breaks=c('E','F'),
labels=c('E','F')) +
theme(axis.text=element_text(size=11),
axis.title.x=element_text(size=13, vjust=-0.75),
axis.title.y=element_blank(),
legend.text=element_text(size=11),
legend.title=element_text(size=13),
plot.title=element_text(hjust=0.5))
This code creates a nice plot but I would like to add a horizontal black line at y=50 AND have the legend show a black line with the text "cutoff" in the legend, but leave points in the legend for the sources. I can add the line with geom_line but cannot get the line in the legend.
library(ggplot2)
the.data <- read.table( header=TRUE, sep=",",
text="source,year,value
S1,1976,56.98
S1,1977,55.26
S1,1978,68.83
S1,1979,59.70
S1,1980,57.58
S1,1981,61.54
S1,1982,48.65
S1,1983,53.45
S1,1984,45.95
S1,1985,51.95
S1,1986,51.85
S1,1987,54.55
S1,1988,51.61
S1,1989,52.24
S1,1990,49.28
S1,1991,57.33
S1,1992,51.28
S1,1993,55.07
S1,1994,50.88
S2,1993,54.90
S2,1994,51.20
S2,1995,52.10
S2,1996,51.40
S3,2002,57.95
S3,2003,47.95
S3,2004,48.15
S3,2005,37.80
S3,2006,56.96
S3,2007,48.91
S3,2008,44.00
S3,2009,45.35
S3,2010,49.40
S3,2011,51.19")
ggplot(the.data, aes( x = year, y = value ) ) +
geom_point(aes(colour = source)) +
geom_smooth(aes(group = 1))
(1) Try this:
cutoff <- data.frame( x = c(-Inf, Inf), y = 50, cutoff = factor(50) )
ggplot(the.data, aes( year, value ) ) +
geom_point(aes( colour = source )) +
geom_smooth(aes( group = 1 )) +
geom_line(aes( x, y, linetype = cutoff ), cutoff)
(2) Regarding your comment, if you don't want the cutoff listed as a separate legend it would be easier to just label the cutoff line right on the plot:
ggplot(the.data, aes( year, value ) ) +
geom_point(aes( colour = source )) +
geom_smooth(aes( group = 1 )) +
geom_hline(yintercept = 50) +
annotate("text", min(the.data$year), 50, vjust = -1, label = "Cutoff")
Update
This seems even better and generalizes to mulitple lines as shown:
line.data <- data.frame(yintercept = c(50, 60), Lines = c("lower", "upper"))
ggplot(the.data, aes( year, value ) ) +
geom_point(aes( colour = source )) +
geom_smooth(aes( group = 1 )) +
geom_hline(aes(yintercept = yintercept, linetype = Lines), line.data)
Another solution :
gg <- ggplot(the.data, aes( x = year, y = value ) ) +
geom_point(aes(colour = source)) +
geom_smooth(aes(group = 1))
cutoff <- data.frame(yintercept=50, cutoff=factor(50))
gg +
geom_hline(aes(yintercept=yintercept, linetype=cutoff), data=cutoff, show_guide=TRUE)
This code generates exactly the same graphic as the one in point (1) of #G. Grothendieck. But it is more easy to adapt to graphics with several layers.
I am building charts that have two lines in the axis text. The first line contains the group name, the second line contains that group population. I build my axis labels as a single character string with the format "LINE1 \n LINE2". Is it possible to assign different font faces and sizes to LINE1 and LINE2, even though they are contained within a single character string? I would like LINE1 to be large and bolded, and LINE2 to be small and unbolded.
Here's some sample code:
Treatment <- rep(c('T','C'),each=2)
Gender <- rep(c('Male','Female'),2)
Response <- sample(1:100,4)
test_df <- data.frame(Treatment, Gender, Response)
xbreaks <- levels(test_df$Gender)
xlabels <- paste(xbreaks,'\n',c('POP1','POP2'))
hist <- ggplot(test_df, aes(x=Gender, y=Response, fill=Treatment, stat="identity"))
hist + geom_bar(position = "dodge") + scale_y_continuous(limits = c(0,
100), name = "") + scale_x_discrete(labels=xlabels, breaks = xbreaks) +
opts(
axis.text.x = theme_text(face='bold',size=12)
)
I tried this, but the result was one large, bolded entry, and one small, unbolded entry:
hist + geom_bar(position = "dodge") + scale_y_continuous(limits = c(0,
100), name = "") + scale_x_discrete(labels=xlabels, breaks = xbreaks) +
opts(
axis.text.x = theme_text(face=c('bold','plain'),size=c('15','10'))
)
Another possible solution is to create separate chart elements, but I don't think that ggplot2 has a 'sub-axis label' element available...
Any help would be very much appreciated.
Cheers,
Aaron
I also think that I could not to make the graph by only using ggplot2 features.
I would use grid.text and grid.gedit.
require(ggplot2)
Treatment <- rep(c('T','C'), each=2)
Gender <- rep(c('Male','Female'), 2)
Response <- sample(1:100, 4)
test_df <- data.frame(Treatment, Gender, Response)
xbreaks <- levels(test_df$Gender)
xlabels <- paste(xbreaks,'\n',c('',''))
hist <- ggplot(test_df, aes(x=Gender, y=Response, fill=Treatment,
stat="identity"))
hist + geom_bar(position = "dodge") +
scale_y_continuous(limits = c(0, 100), name = "") +
scale_x_discrete(labels=xlabels, breaks = xbreaks) +
opts(axis.text.x = theme_text(face='bold', size=12))
grid.text(label="POP1", x = 0.29, y = 0.06)
grid.text(label="POP2", x = 0.645, y = 0.06)
grid.gedit("GRID.text", gp=gpar(fontsize=8))
Please try to tune a code upon according to your environment (e.g. the position of sub-axis labels and the fontsize).
I found another simple solution below:
require(ggplot2)
Treatment <- rep(c('T','C'),each=2)
Gender <- rep(c('Male','Female'),2)
Response <- sample(1:100,4)
test_df <- data.frame(Treatment, Gender, Response)
xbreaks <- levels(test_df$Gender)
xlabels[1] <- expression(atop(bold(Female), scriptstyle("POP1")))
xlabels[2] <- expression(atop(bold(Male), scriptstyle("POP2")))
hist <- ggplot(test_df, aes(x=Gender, y=Response, fill=Treatment,
stat="identity"))
hist +
geom_bar(position = "dodge") +
scale_y_continuous(limits = c(0, 100), name = "") +
scale_x_discrete(label = xlabels, breaks = xbreaks) +
opts(
axis.text.x = theme_text(size = 12)
)
All,
Using Triad's cheat this is the closest I was able to get to solution on this one. Let me know if you have any questions:
library(ggplot2)
spacing <- 0 #We can adjust how much blank space we have beneath the chart here
labels1= paste('Group',c('A','B','C','D'))
labels2 = rep(paste(rep('\n',spacing),collapse=''),length(labels1))
labels <- paste(labels1,labels2)
qplot(1:4,1:4, geom="blank") +
scale_x_continuous(breaks=1:length(labels), labels=labels) + xlab("")+
opts(plot.margin = unit(c(1, 1, 3, 0.5), "lines"),
axis.text.x = theme_text(face='bold', size=14))
xseq <- seq(0.15,0.9,length.out=length(labels)) #Assume for now that 0.15 and 0.9 are constant plot boundaries
sample_df <- data.frame(group=rep(labels1,each=2),subgroup=rep(c('a','b'),4),pop=sample(1:10,8))
popLabs <- by(sample_df,sample_df$group,function(subData){
paste(paste(subData$subgroup,' [n = ', subData$pop,']',sep=''),collapse='\n')
})
gridText <- paste("grid.text(label='\n",popLabs,"',x=",xseq,',y=0.1)',sep='')
sapply(gridText, function(x){ #Evaluate parsed character string for each element of gridText
eval(parse(text=x))
})
grid.gedit("GRID.text", gp=gpar(fontsize=12))
Cheers,
Aaron
Is there a way to add labels to each point in a plot? I did this on an image editor just to convey the idea: 1.
The original one was generated with:
qplot(pcomments, gcomments , data = topbtw, colour = username)
To follow up on Andrie's excellent answer, I frequently employ two methods to add labels to a subset of points on a plot if I need to highlight specific data. Both are demonstrated below:
dat <- data.frame(x = rnorm(10), y = rnorm(10), label = letters[1:10])
#Create a subset of data that you want to label. Here we label points a - e
labeled.dat <- dat[dat$label %in% letters[1:5] ,]
ggplot(dat, aes(x,y)) + geom_point() +
geom_text(data = labeled.dat, aes(x,y, label = label), hjust = 2)
#Or add a separate layer for each point you want to label.
ggplot(dat, aes(x,y)) + geom_point() +
geom_text(data = dat[dat$label == "c" ,], aes(x,y, label = label), hjust = 2) +
geom_text(data = dat[dat$label == "g" ,], aes(x,y, label = label), hjust = 2)
Yes, use geom_text() to add text to your plot. Here is an example:
library(ggplot2)
qplot(mtcars$wt, mtcars$mpg, label=rownames(mtcars), geom="text")
ggplot(mtcars, aes(x=wt, y=mpg, label=rownames(mtcars))) + geom_text(size=3)
See the on-line documentation for more information: http://had.co.nz/ggplot2/geom_text.html