In this repository we provide code to accompany the paper:
The code is based on both python and R and includes Jupyter notebooks (https://jupyter.org/) for setting the thresholds, as discussed below. The code has been tested on macOS version 12.1, python version 3.7.6 and R version 3.6.2. Code for analysis of the PLA data has also been tested on Windows 10 with R version 4.0.2.
Python libraries required: pandas, numpy, matplotlib.
R libraries required: readr, ggplot2, dplyr, ggpubr, rstatix, psych.
The code is split by 'cells', 'tissues' and 'treatment'. For all three cases the RCPs in the cells or tissues (termed 'blobs' in the code) are thresholded as belonging to either the Mouse, Rabbit or both antibodies. This thresholding is done based on inspecting the plots of the intensities of the stains, with Mouse (TexRed) on the x-axis and Rabbit (Atto647) on the y-axis. Also used to inform this thresholding is a histogram of the angles for the blobs in the x-y plots, with an angle of zero for a blob parallel with the x-axis and an angle of 90 for a blob parallel with the y-axis. See the Jupyter notebooks (.ipynb files) for examples for each of the three cases. For some cases a new orgin was defined for thresholding of the intensities (as shown here in MB_thresholding_treatments.ipynb).
Once the blobs have been so categorised a csv file is produced. This csv file is run through the R scripts to produce the final plots as shown in the manuscript: boxplots for the 'cells', piecharts for the 'tissues' and boxplots with signficance added for the 'treatments'.
The examples given include EMD-LMNB1 (Fig. 6a and Supp. Fig. 6a in manuscript) for the 'cells', Ecad-bcat in prostate (Fig. 5b and Supp. Fig. 5c) for the 'tissues' and TGFb1 treated Ecad-bcat (Fig. 5a and Supp. Fig. 5a) for the 'treatments'.
Reproducing the results as shown in the Jupyter notebooks and the manuscript plots requires creating csv files from the three CellProfiler output files (with the suffices 'Image', 'Both_Cells' and 'BlobsInCells') from the source data shared with the manuscript.
Running the code for the thresholding and plotting in each case takes only a couple of minutes.
The code is split by 'cells' and 'treatment'. For all cases RCP in the cells (termed blobs in the code) are available from the csv output file from CellProfiler with the suffix '_Cells'. The CellProfiler output file with the suffix '_Image' contains information on content of the image via the image file name.
The CellProfiler csv file '_Cells' is run directly through the R script to produce boxplots as shown in the manuscript for both 'cells' and 'treatments'. The examples given include EMD-LMNB1 for 'cells' and Ecad-bcat for 'treatments'.
Reproducing the results requires two CellProfiler output files with the suffixes '_Cells' and '_Image' from the source data shared with the manuscript.
Running the code for plotting takes less than a minute.