Quantify the Percentage of GFP-Expressing Nuclei without a Fluorescent Biomarker for DNA

Hello, I am new to CellProfiler but have a couple of questions regarding scoring fluorescent objects.
I have followed the example on the CP website to find the percentage of positive nuclei vs total number of nuclei using the CalculateMath Module. However, this example assumes that a DNA stain (i.e. DAPI) was used to help segment individual cells.

  1. Is there a way in which I can determine the percentage of nuclei expressing GFP in a population of cells without using a channel for a DNA stain (since I did not use one)?
    I would like nuclei to be filtered according to the property of having GFP (positive) or not having GFP (negativeobject) related to them.

  2. Also, can this be done using bright field images?

Thank you in advance!

It would probably help to have a sample image, but to clarify, you have a 2 channel image, where one channel is brightfield and one is GFP? Or, otherwise, how are you determining your total cell count. It seems unlikely that you would be able to with a mostly black image and a few GFP spots.

Thank you for the quick response. Unfortunately I do not yet have a sample image, but yes that is correct. I plan on having a two channel image in which one channel is brightfield and one is GFP.
This is a preliminary experiment meant to determine the relative percentage of cells that express GFP following the addition of varying amounts of GFP control transduction particles in a 96-well plate. (I was planning on approximating the total cell count based on confluence of each well.)
I was hoping to make a simple adjustment to the example pipeline for which a DNA stain is not present, but I’m not quite sure if this can be done.

Generally you want some kind of marker for all cells, and then a subset will have the second marker. Without the primary marker it becomes much more difficult, as brightfield images are harder to segment. Not impossible, but much, much harder, and the focus being even slightly off can make things very difficult for any algorithm.

That makes sense, thank you. At least it’s possible! Any particular suggestions on how to identify my primary object (nuclei) without a primary marker?

Nope. That will depend too much on image quality, focus, whether it is phase contrast, etc. I don’t really recommend it. It would be far easier to have a marker like Hoechst. You might check the forum for brightfield segmentation or phase contrast segmentation topics for an idea of what you are getting into.