Plastid DNA masking

First of all, thanks for making this incredible piece of software free for public use! It has saved me days of time, and enabled significant reduction of human bias in my measurements–not to mention prevented much frustration over licensing fees. Keep up the good work.

Secondly, help! I am having some trouble using a mask to remove plastid DNA from consideration during measurement of nuclei.

The samples: Plant nuclei isolated as though for flow cytometry, and stained with DAPI. Generally less than 20 nuclei within an entire field, very bright, well-defined, plenty of empty background and very few artifacts.

The problem: Chloroplasts which made it through the filter mesh also appear(in the DAPI image) as nuclei because of their compartmentalized plastid DNA. I took an additional image of each field at a red wavelength to capture chloroplast autofluorescence.

Attempted solution:

  1. Identify Primary Objects - Nuclei (ID’d in blue image, includes false positives from chloroplasts)
  2. Identify Secondary Objects - Chloroplasts (ID’d in red image, shows outline of plastid)
  3. Mask Objects - Mask/remove any “nuclei” identified in the DAPI image that are enclosed within the outline of a plastid from the red image (mask object nuclei using object chloroplast)
  4. Measure Objects - measurements of the remaining “authentic” nuclei

Unfortunately, the mask does not seem to be carrying over to the rest of the pipeline after step 3. I tried:

  1. Inverting the mask
  2. Making sure I had requested removal of any overlapping objects and re-numbering of objects
  3. A second “Identify Primary Objects” immediately after the masking step (with and without inversion in masking step), using the same algorithm and thresholds as my original “Identify Primary Objects”
  4. Testing to see if the mask was filtering properly–the pipeline produces a grey-background image with large black “holes” where the plastids were (or vice versa), but then proceeds to identify or measure the enclosed plastid DNA along with true nuclei in the following step (both when inversion is checked and unchecked).
  5. Making sure the nuclei and chloroplast objects in step 1 & 2 of the pipeline were being identified and outlined accurately.
  6. Consulting the documentation, but couldn’t identify where I went wrong.

I feel like I must be missing something very obvious. Might the problem be that I am using a red image (the chloroplasts) to create the mask, and then asking the program to measure the DAPI image? The steps after masking don’t have any additional image selection options (apart from my original red and blue), so I assumed the mask was applied automatically–although the “plastid DNA” object (nuclei enclosed by chloroplasts) becomes available after the masking step.

I’d appreciate any advice you can provide.

Hi Elspeth,

Without seeing the images, I think the issue you are seeing is that IdentifySecondary identifies a secondary object for every primary object; if there is no secondary staining (for the non-chloroplasts in the red channel in your case), the secondary objects are still identified, just that they share the same contour as the nuclei. Masking won’t do the job in this case.

For this instance, you can follow the basic workflow in the percent-positive pipeline from our Examples page. Essentially, you would use IdentifyPrimary to identify the red-stained chloroplasts, then use RelateObjects to establish the parent-child relationship between the DAPI-stained objects and red-stained objects, then FilterObjects to exclude all objects which have both stains, i.e., a child count greater than 1.