Using ilastik for determining protein subcellular localization

I have a IF pictures of cells stained for my protein of interest which are also stained with DAPI for nuclei. I was wondering if how to calculate the amount of my protein of interest that is found in the nucleus or cytoplasm. I guess I would need to determine the pixel density/intensity found 1) within the segmented nuclei or 2) outside of the segmented nuclei region but not in the background either (i.e. the cytoplasm).

I saw in another post @VolkerH commented:

Does anyone know how this can be done? My experience with these types of image analyses is almost 0.

Also, I don’t necessarily need the information on an individual cell basis, just more so for any given picture (that i have 50-100 cells) the percentage of protein X that is nuclear and the percentage that is cytoplasmic.

Thank you!

Hi @mroberto93,

that problem sounds interesting. Without an image to comment on, I will try to guess how it looks :wink:

Problem statement: measure stained protein in the nucleus, in the cytoplasm.

So the actual measurement you’ll have to figure out yourself (how to quantify your signal in general). With the assignment to nucleus and cytoplasm, ilastik can help!
Again, it would’ve been helpful to see an image, but let’s do it like this:

  1. Train an ilastik pixel classification project with three classes to segment just the dapi, cytoplasm and background area in each cell

  2. with the exported data you can probe every pixel you do a measurement in (i suppose in your protein staining channel) (with your own method) whether it is in the nucleus, cytoplasm or background class.

About that, so you really don’t care that in extreme cases you might be reporting statistics, assuming you have 50-100 cells that might actually only be derived by e.g. one or two cells? You’re also not interested how the distribution looks like across the cells?
In any case, you could also get these kinds of statistics, but it would be more involved.



This is very much possible with Cellprofiler. You might have to check the example tutorials, but specifically you could look at translocation assay pipeline and for basic pipeline human cells.
But it would be great to suggest with the look at image as mentioned above.

Fujifilm Wako Automation (Consultant)
Join us at the High Content Analysis Symposium Oct 21-22, London

Society for Laboratory Automation and Screening

SLAS 2019 Advanced 3D Human Models and High-Content Analysis Symposium

Read more on our site.

Yokogawa CV8000 - The Ultimate in Confocal HCS