Problem with clumped cells and low signal/background ratio

cellprofiler

#1

Dear all,
I’m having problems with setting a workable pipeline with CP.
The aims seems pretty straightforward though. I want to be able to count the total cell number (using DAPI) and then characterize the percent of stained cells with one or more different antidodies.

I have a few questions:

  1. I have .tif files with 4 channel (DAPI, 488, 560 and 641). Can I upload this to CP and ask him to split channel? I only found this function with the colortogray function (but only as RGB files).
  2. I started with the pipeline from CP website (percentpositive) which seemed to do what I want to do. I tried to twick it a bit to work, but it doesn’t. The software doesn’t distinguish well enough my primaryobject as being the DAPI nuclei. I tried a lot of different things but can’t make it work :frowning:
  3. For some of my staining, I seem to have a “high” background but we are still able to see positive cells. I was wondering what parameter I can change so CP can distinguish between the background and a real staining?
    Is it possible to adapt that parameter for different staining (which would have different background level)…

Thank you for your help!

any idea anyone please?
@Minh @mbray @bcimini
Kamal

PercentPositive.cppipe (20.5 KB)
DIFF8A_d5_GFP_Sox2_Olig2_1_MMStack_Pos0.ome-1.tif.zip (9.8 MB)


#2

Hi,

I have .tif files with 4 channel (DAPI, 488, 560 and 641). Can I upload this to CP and ask him to split channel? I only found this function with the colortogray function (but only as RGB files).

This can be done in two different ways in CP- by calling the 4 channel image a “ColorImage” in NamesAndTypes and splitting it up in ColorToGray (see the first pipeline below). The other (in the second pipeline) is to extract the number of channels in the Metadata module and use the extracted metadata in NamesAndTypes; just make sure to hit the top “Update button” first before hitting the bottom one (see below). I find the first less cumbersome but YMMV.


For some of my staining, I seem to have a “high” background but we are still able to see positive cells. I was wondering what parameter I can change so CP can distinguish between the background and a real staining?

I’ve added some background subtraction to your pipeline; I’d recommend closely reading the help of the “CorrectIlluminationCalculate” and “CorrectIlluminationApply” modules to learn more.


I started with the pipeline from CP website (percentpositive) which seemed to do what I want to do. I tried to twick it a bit to work, but it doesn’t. The software doesn’t distinguish well enough my primaryobject as being the DAPI nuclei.

I didn’t actually have time to look at your segmentation (perhaps someone else can build on my edits to this point) so I don’t know what “doesn’t distinguish well enough” means, but I’d recommend reading the help for IdentifyPrimaryObjects and really thinking about how each setting might affect the output. The other thing to consider is that your nuclei are SUPER clumpy, so an imperfect segmentation or estimation may be all that can be done here; without knowing the biology I don’t know if it’s possible to change your plating conditions or something to make it less clumped, but it’s worth pursuing if you can.

Good luck!

4ColorSplit.cppipe (4.4 KB)
PercentPositive_edits.cppipe (21.8 KB)