I'm using R to create a heatmap from a matrix using heatmap.2 - and i want to group these images into one big image - What i usually use to achieve this is layout() - but this doesn't work, as heatmap.2 uses layout, and apparently layout does not work recursively.
Does anyone have any suggestions on how to group together 2 images without layout, or how to make layout support recursive calls?
mat = matrix(nrow=3,nrow=3,1:9)
layout(matrix(nrow=2,ncol=1))
heatmap.2(mat) ## overrides the layout and produces only one plot that takes whole screen
heatmap.2(mat) ## still only one image
thanks.
What follows is a hack that is almost certainly not a perfect solution, but it may get you started.
Create your own version of the heatmap.2 function called hm3. In the code for hm3, comment out all the lines between:
if (missing(lhei) || is.null(lhei))
and the layout call:
layout(lmat, widths = lwid, heights = lhei, respect = FALSE)
it's a big chunk of code, maybe 30 lines. Now the following code produces two heat maps with dendrograms and keys side by side:
x <- as.matrix(mtcars)
lhei <- c(1.5, 4,1.5,4)
lwid <- c(1.5, 4,1.5,4)
layout(rbind(c(4,3,8,7),c(2,1,6,5)),
widths = lwid, heights = lhei, respect = FALSE)
hm3(x)
hm3(x)
Clearly, this will require considerable tweaks to make it look nice (and a larger plotting area; I've squished everything to be a reasonable size to post here).
This is entirely untested. It is likely that using any of the options in the hm3 function that control the appearance of the plot will cause things to go slightly haywire. But this may be a good starting point for your own experimentation to address those issues.
What are you planning on doing with the results?
If you just want to compare 2 heatmaps side by side on the screen then rather than combine them into one single plot you can open 2 plotting devices and arrange them side by side to compare (much simpler than creating a single graph):
heatmap.2(mat1)
dev.new()
heatmap.2(mat2)
Now drag one to the side of the other using your mouse.
If you want to include the combined graphic in a publication then it may be easiest to create the 2 plots and just set them side by side in whatever program you are using to create the article. If you need them in one file you can still save the 2 heatmaps (or other plots) as 2 files then use tools such as imagemagick, gimp, or inkscape to combine the 2 files into 1 with the graphs side by side.
Related
From an R function (cnetplot) I've obtained the following image which does not look very nice.
Therefore, I extracted the data from the R object and wrote a script to create an equivalent network data file that is readable by Cytoscape. The following equivalent plot from Cytoscape looks much better but the problem is that I am not able to add legends based on node size in Cytoscape as the R function did. I tried with Legend Creator app in cytoscspe but couldn't do it.
The original data and R code to reproduce the plots can be found in the following link.
ftp://ftp.lrz.de/transfer/Data_For_Plot/StackOverflow/
I looked into this Mapping nodes sizes and adding legends in cytoscape network, but in that case questioner already was able to load the node sizes as legends in cytoscape and moreover, he/she used a python package.
Any suggestions will highly be appreciated
Here's a little R script that will generate a min/max node size legend. You'll need to set the first variable to the name of the Visual Style in your network. This one works with the sample session file, "Yeast Perturbation.cys" if you want to test it there first.
If you are familiar with RCy3, then it should be self-explanatory. You can customize the positioning of the nodes, labels and label font size, etc. You can even adapt it to generate intermediate values (like in your example above) if you want.
NOTE: This adds nodes to your network. If you run a layout after adding these, then they will be moved! If you rely on node counts or connectivity measures, then these will affect those counts! Etc.
If you find this uesful, I might try to add it as helper function in the RCy3 package. Let me know if you have feedback or questions.
# https://bioconductor.org/packages/release/bioc/html/RCy3.html
library(RCy3)
# Set your current style name
style.name <- "galFiltered Style"
# Extract min and max node size
res<-cyrestGET(paste0("styles/",style.name,"/mappings/NODE_SIZE"))
size.col <- res$mappingColumn
min.size <- res$points[[1]]$equal
min.value <- res$points[[1]]$value
max.size <- res$points[[length(res$points)]]$equal
max.value <- res$points[[length(res$points)]]$value
# Prepare as data.frame
legend.df <-data.frame(c(min.size, max.size), c(min.value, max.value))
colnames(legend.df) <- c("legend.label",size.col)
rownames(legend.df) <- c("legend.size.min", "legend.size.max")
# Add legend nodes and data
addCyNodes(c("legend.size.min", "legend.size.max"))
loadTableData(legend.df)
# Style and position
setNodeColorBypass(c("legend.size.min", "legend.size.max"),"#000000")
setNodePropertyBypass(c("legend.size.min", "legend.size.max"),
c("E,W,l,5,0", "E,W,l,5,0"), # node_anchor, label_anchor, justification, x-offset, y-offset
"NODE_LABEL_POSITION")
setNodeLabelBypass(c("legend.size.min", "legend.size.max"), legend.df$legend.label)
setNodePropertyBypass("legend.size.max",
as.numeric(max.size)/2 + as.numeric(min.size)/2 + 10, # vertical spacing
"NODE_Y_LOCATION")
setNodeFontSizeBypass(c("legend.size.min", "legend.size.max"), c(20,20))
I am trying to create horizontal bar chart in in R using the plotly package. Due to the length of the legend items I would like for them to show horizontally at the top or bottom of the visual in 2 columns. Is it possible to dictate the number of columns for the legend?
I've been able to place the legend below the x axis successfully using Layout(legend = list(orientation='h')) however regardless of where I put the legend (using the x and y arguments) it is always just one long list. I've seen a github project for creating a multi column legend in js but not r.
Thanks,
This is not possible in a normal way. I think it has its own logic that determines how many place there it is and how many columns it will display then.
So I guess if you make your plot width smaller you could reach the goal that it will just display 2 column.
Also you can try to play around with the margin attribute (https://plot.ly/r/reference/#layout-margin) by setting r and l to 10 e.g.
An other idea could be to make the font-size in legend (https://plot.ly/r/reference/#layout-legend-font-size) bigger, so that it just uses two columns. Hope it helps.
I read the same github page and I thought that it is not possible, but seems to be! I only checked in Python, but I hope this will help in your endeavors in R as well as everyone in Python looking for information. Sadly, there is not a lot of information on Plotly here compared to other packages.
This solved my problem
Setting orientation='h' is not enough. You also have to put the legend items in different legendgroups, if you want them in different columns. Here is an example with legend labels:
fig = go.Figure([
go.Scatter(x=best_neurons_df['Test Size'],
y=best_neurons_df['Training Accuracy Max'],
# You can write anything as the group name, as long as it's different.
legendgroup="group2",
name='Training',
mode='markers',
go.Scatter(x=best_neurons_df['Test Size'],
y=best_neurons_df['Validation Accuracy Max'],
# You can write anything as the group name, as long as it's different.
legendgroup="group1",
layout=dict(title='Best Model Dependency on Validation Split',
xaxis=dict(title='Validation Set proportion'),
yaxis=dict(title='Accuracy'),
margin=dict(b=100, t=100, l=0, r=0),
legend=dict(x=1, y=1.01,xanchor='right', yanchor='bottom',
title='',
orientation='h', # Remember this as well.
bordercolor='black',
borderwidth=1
))
Example image
I have this pairs plot
I want to make this plot bigger, but I don't know how.
I've tried
window.options(width = 800, height = 800)
But nothing changes.
Why?
That thing's huge. I would send it to a pdf.
> pdf(file = "yourPlots.pdf")
> plot(...) # your plot
> dev.off() # important!
Also, there is an answer to the window sizing issue in this post.
If your goal is to explore the pairwise relationships between your variables, you could consider using the shiny interface from the pairsD3 R package, which provides a way to interact with (potentially large) scatter plot matrices by selecting a few variables at a time.
An example with the iris data set:
install.packages("pairsD3")
require("pairsD3")
shinypairs(iris)
More reference here
I had the same problem with the pairs() function. Unfortunately, I couldn't find a direct answer to your question.
However, something that could help you is to plot a selected number of variables only. For this, you can either subset the default plot. Refer to this answer I received on a different question.
Alternatively, you can use the pairs2 function which I came across through this post.
To make the plot bigger, write it to a file. I found that a PDF file works well for this. If you use "?pdf", you will see that it comes with height and width options. For something this big, I suggest 6000 (pixels) for both the height and width. For example:
pdf("pairs.pdf", height=6000, width=6000)
pairs(my_data, cex=0.05)
dev.off()
The "cex=0.05" is to handle a second issue here: The points in the array of scatter plots are way too big. This will make them small enough to show the arrangements in the embedded scatter plots.
The labels not fitting into the diagonal boxes is resolved by the increased plot size. It could also be handled by changing the font size.
I have a long time series of 10000 observations that I want to visualize. The problem is, if I just plot it normally the time-dimension will be squished and none of the fine detail of the time-series that I want to visualize will be apparent. For example:
plot((sin(1:10000/100)+rnorm(10000)/5),type='l')
What I would like is to somehow plot the following together side by side in one gigantically long plot without using par(mfrow=c(1,100)). I then want to export this very wide plot and simply scroll across to vizualise the whole series.
plot((sin(1:10000/100)+rnorm(10000)/5)[1:100],type='l')
plot((sin(1:10000/100)+rnorm(10000)/5)[101:200],type='l')
plot((sin(1:10000/100)+rnorm(10000)/5)[201:300],type='l')
.....
Eventually I would like to have 3 or 4 of these gigantically wide plots on top of each other with a par(mfrow=c(4,1)).
I know that the answer has something to do with the pin setting in par, but I keep getting Error in plot.new() : plot region too large. I'm guessing this has something to do with the interaction of pin with the other par parameters
Bonus points are awarded if we can get the pixel height and width exactly right. It is preferable that the plot doesn't skip random pixels due to the export sizing being imperfect.
Further bonus points if the image can be encoded in a .html. and viewed this way
An alternative that you might consider is svg, which will produce something of better quality than png/jpeg in any case.
Something like
svg(width = 2000, height = 7)
par(mfrow=c(4,1), mar = c(4, 4, 0, 2))
for (i in 1:4){
plot((sin(1:10000/100)+rnorm(10000)/5),type='l',
bty = "l", xaxs = "i")
}
dev.off()
will produce a very wide plot, just over 1MB in size, which renders quite nicely in Chrome.
Note the width and height are in inches here.
P.S. svg also offers the potential for interactive graphics. Just seen a nice example allowing the user to select a region of a long time series to zoom in on, see Figure 22 in Dynamic and Interactive R Graphics for the Web: The gridSVG Package, a draft paper by Paul Murrell and Simon Potter.
It could be a Cairo-specific problem, or it could be a lack of RAM on your machine. The following code works fine for me on a Windows 7 machine with 8GB RAM.
png("wide.png", width = 1e5, height = 500)
plot((sin(1:10000/100)+rnorm(10000)/5),type='l')
dev.off()
If I change the width to 1e6 pixels, then R successfully creates the file (it took about a minute), but no image viewing software that I have available can display an image that large.
I would go on some alternative route. First of all, what exactly is the point of viewing the entire plot at hi-res? If you're searching for some sort of anomalies or irregularities, well, that's what data processing is for :-) . Think about something like finding allx > 3sigma, or doing an FFT, etc.
Next, if you really want to examine the whole thing by eye, how about writing some R-TclTK code or using dynamicGraph or iplots or zoom to produce an interactive graph that you can scroll thru "live."
ETA: IIRC RStudio has tools for interactive graph scrolling and zoom as well.
If I create a multi-plot window with par(mfrow=...), is it possible to send data to a specific plot (i.e. "the one in the lower left corner") or is the plotting always necessarily sequential? Is there a package for R that does something like this?
For those that are interested, this problem arises out of the fact that R is a single-threaded application and is not ideal for real-time visualization. I have multiple real-time data streams coming into R from an outside source that produces the data asynchronously (and therefore the data streams don't always come in the same order). This results in R flipping around the order of the data visualization plots every time it updates.
You could use split.screen():
par(bg = "white") # erase.screen() will appear not to work
# if the background color is transparent
# (as it is by default on most devices).
split.screen(c(2,1)) # split display into two screens
split.screen(c(1,3), screen = 2) # now split the bottom half into 3
screen(1) # prepare screen 1 for output
plot(10:1)
screen(4) # prepare screen 4 for output
plot(10:1)
Have a look at help(layout). This allows you to specify the what, where and in which sizes.
Once plotted, I don't think you re-plot just partially. But you you can use dev.set() et al to switch between different 'plot devices' (ie windows); see help(dev.list).
Note that the suggested answer here is to use split.screen(). It may work, but according to the split.screen help file: "The recommended way to use these functions is to completely draw a plot and all additions (i.e. points and lines) to the base plot, prior to selecting and plotting on another screen. The behavior associated with returning to a screen to add to an existing plot is unpredictable and may result in problems that are not readily visible."
In an answer to my question, there is a more useful solution, using the par(mfg) option:
Change plot panel in multipanel plot in R
Another option is that of implementing a little GUI e.g. with RGtk2 or RTclTk.
I generally do this for graphs that I want to change in realtime and it works great.
For instance, with RGtk2 and cairoDevice you could just do something like (I assume you have a Glade interface)
# Helper function to get a widget from the Glade interface
getWidget <- function(name)
{
return (interface$getWidget(name))
}
interface <- gladeXMLNew("interface.glade", root="mainWindow")
# Our cairo devices (to draw graphics).
# plot1, plot2, and plot3 are GtkDrawingArea widgets
asCairoDevice(getWidget("plot1"))
# dev.cur() will give the device number of the last device we created
# You'll use this to switch device when you draw in different plots
# Storing the device number is important because you may have other
# devices open from other unrelated plots
# (so never assume they'll just start from 1 and be sequential!!!)
plot1.dev <- as.integer(dev.cur())
asCairoDevice(getWidget("plot2"))
plot2.dev <- as.integer(dev.cur())
asCairoDevice(getWidget("plot3"))
plot3.dev <- as.integer(dev.cur())
# To draw in a specific plot you just do
dev.set(plot2.dev)
plot(....)
This has many other advantages, like that of being able to positions the graphs easily where you want (using Glade Interface Designer) and having the possibility of user interaction through specific buttons (e.g. you may have a "pause acquisition" button).