Nuclear translocation help

hi. Im new at cell profiler. I am trying to see nuclear translocation of my protein of interest. The problem i am having is to measure the nuclear and cytoplasmic intensities (which are in the red channel) from only the green cells. The images contain non green as well as green cells. I want to measure the red signal from only the cells that are transfected i.e green. I have been able to get the nuclear vs cytoplamic intensities ratio to work when all cells are taken into consideration. But as soon as i try to select only the green cells everything goes haywire. Any help will be appreciated. Thanks.

You would probably want to do something like identify cells->measure their intensity in the green channel-> filter down to only the green ones -> perform your nuclear vs cytoplasmic measurements. It would be helpful to know what you’ve already tried, and what you specifically mean by “everything goes haywire”. See more about what we suggest including here.

pipeline nuclear cyt final final.cppipe (17.3 KB)

hi. thanks for the reply. This is the pipeline that i have built up till now. The idprimary searches for the nuclei in the dapi channel. The id primary (again) searches for the green cells in the GFP channel. The idsecondary identifies the Cells using the nuclei in the red channel (red=protien of interest). What i want to do now is to obtain the intensity from only the green cells for the red channel. And obtain the nuclear signal from only the green cells for the red channel. I.e i want the intensity of the protien of interest from the nucleus and the cytoplasm from only the green cells. The green channel of shows the over-expression in cells and is only there for the selection of the transfected cells. If you could help me with this it would be great.

Rather than using ID Primary to find green cells I’d find all red cells in the way you describe (Primary followed by Secondary), do a MeasureObjectIntensity on the green channel, then use FilterObjects to only keep cells that are above a certain level of green- this is slightly cleaner because you can use the “Relabel additional objects to match the filtered object” option to filter your nuclei and cytoplasm all together in one go. Then once you have your green-positive nuclei and cells, do MeasureObjectIntensity in the red channel and you’re good to go!

If for some reason you don’t want to do this, you could use two separate MaskObjects modules using your green cells as a mask on nuclei and cells but I definitely recommend option 1.

Finally, if you truly want to measure only the cytoplasm after your ID Secondary you should include an ID Tertiary to pull out a separate cytoplasmic region- otherwise you can just do the subtraction yourself after.

Good luck!

hi. thanks for the reply. i made the changes to the pipeline but in both the ways you mentioned it still missed out on some on the nuclei which then gives me wrong values for some of the measurements. I have attached the new pipeline here and am also sharing some of the images. If you could look at what is causing the quantification to miss out on some of the nuclei it would be great. It either misses out on some nuclei of counts nuclei that are not a part of that cell. You’ve been a great help and it would be awesome if you could check this.corrected nuc cyto.cpproj (683.0 KB)

here is the link to the images:

I looked at one image set briefly, here are my thoughts-

-I don’t think you really need any of those ApplyThreshold modules, they aren’t really doing anything helpful.
-You should spend some time tinkering in your IdentifyPrimaryObjects for the nuclei- if you look at what it’s selecting the declumping isn’t going very well, which affects everything downstream. You can potentially avoid this in the future by plating less densely.
-IntegratedIntensity isn’t a good metric to use for your filter because it also includes measures of size- try Median or Upper Quartile instead.
-If you’re going to use IdentifyTertiary and create a cytoplasm object, make sure to actually filter it alongside the nuclei and use it in your measurements; otherwise just omit that step.

Finally, it’s clear there’s room for improvement here but you may never find settings that identify every cell 100% perfectly; generally 95%+ is good enough but if 100% is critical you’re going to need to include a module like EditObjectsManually so that you can hand curate which cells are good enough and which are not.