This repository contains the scripts that are necessary to perform drug simulations with PhysiBoSS-Drugs (https://github.com/anmeert/PhysiBoSSv2) and the data analysis of my master thesis "PhysiBoSS-Drugs: A customizable drug simulation framework that enables predicting drug synergies in cell populations”.
The raw GDSC data (http://ftp.sanger.ac.uk/pub/project/cancerrxgene/releases/current_release/GDSC2_public_raw_data_25Feb20.csv) needs to be fitted to a multi-level mixed effect model (Vis et al, 2016). This can be done with the help of the R package gdscIC50 (https://github.com/CancerRxGene/gdscIC50). The script fit_GDSC_data.R contains the necessary code to obtain dose response curves and the CSV file that is necessary to integrate into PhysiBoSS-Drugs. The raw GDSC data has to be downloaded beforehand and stored to the data folder.
The script PhysiBoSS-Drugs_to_csv.py which is based on the script plot_time_course.py from the Github repository: https://github.com/migp11/tools4physicell enables the conversion of single-, double- and untreated drug simulation data that was obtained with PhysiBoSS-Drugs into a CSV file. The output CSV file will be stored in the output folder of tools4PhysiBoSS-Drugs. The CSV file is needed to perform the following data analysis and to obtain corresponding plots.
The CSV file used for the analysis in my master thesis can be found in the data folder with the name LNCaP_simulation_data.csv.
PhysiBoSS-Drugs_to_csv.py --single data/single_data_folder --double data/double_data_folder --untreated data/untreated_data_folder
For plotting the pair growth and synergy heatmaps (Fig 2 and 3 of the Main text) the two scripts growth_pair_heatmap.py and synergy_pair_plot.py can be used. The scripts require the specification of the path that leads to the CSV file containing the simulation data. The data used in my master thesis can be found in "/data/LNCaP_simulation_data.csv.
For plotting the complete growth and synergy heatmaps (SuppMat, Fig 13, Fig 14, Fig 15) the two scripts full_growth_heatmap.py and full_synergy_plot.py can be used. The same path specifications as above have to be done. Furthermore, the two drugs that should be analyzed in the heatmaps have to be specified within the script.
To test which of the drug combinations is significant a Kruskal-Wallis rank sum test can be performed with the script kruskal_wallis.py. The path to the simulation data and the used drugs have to be specified inside the script.
The experimental data used in my thesis is stored in data/LNCaP_experimental_data.csv Plots of the experimental data can be obtained with the script plot_experimental_data.py. The output will be saved in the output folder of the repository.
To calculate the experimental AUCs the script calc_experimental_AUC.py can be used.
To validate the results additional drug simulations were performed for Luminespib, Pictilisib and Selumetinib with matching drug concentrations. The corresponding CSV files can be found in data/validation_Luminespib_3_3uM.csv,data/validation_Pictilisib_10uM.csv and data/validation_Selumetinib_30uM.csv. The growth indices for those simulations can be calculated with the script validation_simulation_AUC.py.
The script to plot the timecourse of drug simulations as in SuppMat, Fig 12 is timecourse_SuppMat_section7_fig12.py which is based on the plot_time_course.py script from https://github.com/migp11/tools4physicell.