Colocalization in 3 channels

Hello everyone,

This is quite similar to a previous post, however I’m pretty new to ImageJ so I was hoping to get some guidance on the best way to do colocalization with three fluorescent markers in an object-based manner.

Basically, I have two yeast proteins (1 and 2) tagged with RFP and GFP. I know that 1-RFP recruits 2-GFP creating larger GFP puncta. But I’d like to see if these puncta then co-localize with the mitochondria, which I have stained using DAPI. Unfortunately, all puncta are pretty dim so I lose a lot of information if I just threshold the image. But I can easily pick out the puncta using Find Maxima. Ideally the question I’m trying to answer is: when protein 2 colocalizes with 1, is it at the mitochondria?

For reference, my previous experience with colocalization has been using the Metamorph software and combining the granularity app (for identification of dots) with the colocalization app (to look a colocalization of granule area in the two channels).

Sorry if this is a super simple question! Any help would be greatly appreciated!! :slight_smile:

This is an old question, so maybe you found a way already, but I’ll give it a try.

The colocalization functions I know of are written based on 2 labels, but you could use them in a 2-step process to find your answer.

For example:

  1. Find out “when protein 2 colocalizes with 1” (GFP puncta within a certain threshold distance from RFP puncta). The JaCoP plugin, discussed on this forum and in the documentation wiki, does object-based and distance-based colocalization. There is also a tutorial in the free e-book “Bioimage Data Analyisis.”

  2. Take that subset of puncta (a binary image or list of points) and find out how many are inside DAPI-stained mitochondria. There’s a macro function to find out if a point (one of your puncta) is inside a selection (that you can make from a thresholded mitochondria image).

(I’m assuming DAPI stains the whole mitochondrion in your system – if it’s patchy, a label that fills the matrix or membrane would be better.)

Hope this helps!


Thanks so much for the reply!! I’ve been using the logical AND function to combine binary images of my GFP and RFP channels then analyzing colocalization of resulting dots with DAPI but my results haven’t been great. This method sounds much better and I’ll give it try!! Thanks again :smile:

1 Like


We started working on something similar for a user who needed to check for cells with multiple markers. The markers being inhomogeneous, we opted for the following approach

For each channel

  1. Laplacian of Gaussian (LoG) of the feature the user was interested in
  2. Find maxima on the LoG image to detect puncta of interest.

After each channel was detected

  1. Clustering of very nearby objects to consider them as the same cell.
  2. Output of table containing each detection and which channels they belonged to.

I can provide you with an example output if you can provide an image to test it out.



Thank you for getting back to me and I apologize for the delay - I got sidetracked on another project since I was having issues with the quantification. I’ve uploaded the GFP, RFP, and DAPI exposures separately for one image. An example output would be great!

Thanks again!