RDI Calculator: An Analysis Tool to Assess RNA Distributions in Cells

Localization of RNAs to various subcellular destinations has emerged as a widely used mechanism that regulates a large proportion of transcripts in polarized cells. A number of methodologies have been developed that allow detection and imaging of RNAs at single-molecule resolution. However, methodologies to quantitatively describe RNA distributions are limited. Such approaches usually rely on the identification of cytoplasmic and nuclear boundaries which are used as reference points. Here, we describe an automated, interactive image analysis program that facilitates the accurate generation of cellular outlines from single cells and the subsequent calculation of metrics that quantify how a population of RNA molecules is distributed in the cell cytoplasm. We apply this analysis to mRNAs in mouse and human cells to demonstrate how these metrics can highlight differences in the distribution patterns of distinct RNA species. We further discuss considerations for the practical use of this tool. This program provides a way to facilitate and expedite the analysis of subcellular RNA localization for mechanistic and functional studies.


6)
Next, the program asks the user if they would like to run the RDI analysis. Generally, the user will want to select 'Yes', unless the only desire is to produce saved masks for the nuclear and cytoplasmic signal.

7)
A reminder to the user to save images 8) Next, the program will ask the user if they would like to use a set of techniques to batchclean up their images in the cell mask channel if, for instance there is a lot of background noise. These same options are available per-image, and this is simply the batch option.
For images with good signal-to-noise 'No' can be selected. The test images provided can be run through without any further processing (proceed to step 9A below). For more noisy images, select 'Yes' (proceed to step 9B below).

9A)
For each image, the program displays an overlaid boundary for the cell mask with the RNA channels (to make sure that all RNA spots are included in the generated cell mask), the signal from the cell mask channel with an overlaid boundary of the generated cell mask and the signal from the nuclear channel with an overlaid boundary of the generated nuclear mask. Note that the program will always pick the largest (by area) object in the image to be the cell of interest, and the largest object within the cell to be the nucleus. A pop-up window is shown, giving the option to the user to first save the cell mask and then the nuclear mask. (If any one of the generated masks is not acceptable, 'No' can be selected (proceed to step 9C below)).
The program iterates through each image in the original directory. If the masks are accepted all relevant data are stored in the original image folder, including the masks, Z-projected images with user-determined subtraction, and a .csv file with all relevant analysis data. The analysis data (PDI, PI, DI, Cell area, and average intensity) can also be quickly obtained in the variable 'A' within Matlab.

Options for cleaning up the mask channels or manually altering masks.
For noisy images, the default parameters might not successfully generate acceptable cell masks.
In such a case, the user can specify in step 8 above that the 'images are sub-optimal' and attempt to alter the parameters 9B) If the user chooses the 'my images are sub-optimal' option, a small information box is presented Four options are available to the user to try to clean up the cell mask channel. These methods can be performed multiple times. The first option is the application of the Wienerfilter, for noise removal. The second will sharpen the image, boosting contrast between signal and background. The third will dilate the cell mask by one pixel in all directions. The fourth will modify the threshold for the sobel edge detection.
The program will use the new parameters to generate masks and, as in step 9A, it will iterate through the images, displaying the generated masks and asking the user to save or not the images.
As an example, below is a particularly poor cell mask. The brightness has been increased significantly for visibility.