5

I am currently trying to generate an automated script to create a labeled line graph. This line graphs will have 1 or more maxima that if they are above a certain threshold must be labeled. Currently I can label the maxima just fine.

https://drive.google.com/file/d/0B50V8cCfhJDQWF94SG85NXVZY1k/view?usp=sharing

For that I am using textxy() in the calibrate package

However when multiple peaks occur things get more complicated:

https://drive.google.com/file/d/0B50V8cCfhJDQUmF5VERhUV8zNmc/view?usp=sharing

It quickly becomes difficult to see where the label is assigned to. So what I would like is a line from the label to the top (or just below) of the peak.

I've looked around all day today with no luck. I tried everything in Intelligent point label placement in R

wordcloud, but that unfortunately doesn't allow you to offset the labels, and fails if you have only one label.

identify, is much too slow. I need to be able to automate this to do thousands of images a day.

pointLabel, thigmophobe.labels both didn't work as they don't draw a line, and I am not dealing with lots of labels anyway.

I also tried manually drawing an arrow between the label and the point, but that got very time consuming.

Does anyone know of any package or easy way to do this? Is this not possible to automate?

Thanks! Cameron

Community
  • 1
  • 1
Cameron
  • 63
  • 8

1 Answers1

3

You can try the labelPeaks function provided by the MALDIquant package. Its algorithm to place the peak labels is taken from Ian Fellows' wordcloud::wordlayout.

library("MALDIquant")

data("fiedler2009subset")

## a simplified preprocessing to provide some example data
s <- trim(fiedler2009subset[[1]], c(3000, 3500))
r <- removeBaseline(s)
p <- detectPeaks(r)

## plot the peaks and label them
plot(p, ylim=c(0, 30000))
labelPeaks(p, avoidOverlap=TRUE, underline=FALSE, digits=1)

enter image description here

## rotate labels by 90 degree
plot(p, ylim=c(0, 30000))
labelPeaks(p, underline=FALSE, srt=90, adj=c(0, 0.5), col="red")

enter image description here

## label peaks above 5000
plot(p, ylim=c(0, 30000))
labelPeaks(p, index=intensity(p) > 5000)

enter image description here

Please see ?labelPeaks for more details.

sgibb
  • 25,396
  • 3
  • 68
  • 74
  • That sounds promising although I'm not a huge fan of the labels naturally being on top of the peaks. I'll give it a try later today. – Cameron May 18 '15 at 17:00
  • @Cameron Any suggestions are welcome. I am the maintainer of MALDIquant, please don't hesitate to ask for improvements of the `labelPeaks` function. – sgibb May 18 '15 at 18:33
  • Unfortunatly this doesn't work. While you have correctly identified the fact that I am graphing Mass spec data, the data files I am using are not compatible with the MALDIquant or MALDIquantForeign package. As such I cannot create peak objects for the labelPeaks function. – Cameron May 18 '15 at 18:41
  • @Cameron: Which file format to you using? Via `createMassPeaks(mass=1:10, intensity=1:10)` you could create `MassPeaks` manually (should be easy if your data are already in R); maybe you want to send me an email with an example dataset? – sgibb May 18 '15 at 19:03
  • I am using a .RPT file generated by Waters/Micromass TOF's. It is typically opened in MassLynx. – Cameron May 19 '15 at 01:58
  • @Cameron: I am like to add RPT support into MALDIquantForeign. Unfortunately I have no access to this file format. I would be very happy if you could send me an email with an example dataset. Please find my email address in the *Maintainer* field on http://cran.r-project.org/web/packages/MALDIquant/index.html – sgibb May 19 '15 at 16:51