We have a pipeline that is working reasonably well for Cell Profiler analysis of 3 RNAScope transcripts and co-localization, which we are excited to share as soon as we confirm it is working okay! My question is, I know some groups use individual puncta as a threshold for whether they include a cell (>3 dots for example). Our stain did not yield puncta that can be separated out individually in all cases, so we decided to go with a “% of cell covered by staining” approach to see relative levels of RNA. BUT I am still getting some cells that are 0.001% covered (as an example) that I think I should probably exclude from analysis as cells with too little transcript to include. Has anyone else decided on a lower limit for this process that they feel makes biological sense? Or perhaps a statistical method for determining what lower cut-off to use?
Any and all ideas appreciated!