Scripts used to build E2up and P4up QSAR models and to use them to predict E2up and P4up properties on a set mammary carcinogens

All scripts are included in the folder sources and are separated by type of scripts

• R scripts, mostly used to draw figures
• Python scripts used for data analysis

Dependencies

By coding languages

• Python3.9

  • RDKIT (2020) in environement conda rdkit-env
  • CompDesc ($pip install -i https://test.pypi.org/simple/ CompDesc)
  • scipy ($pip install scipy)
  • molvs ($pip install molvs)
  • bitarray ($pip install bitarray)
  • openpyxl ($pip install openpyxl)
  • tensorflow ($pip install tensorflow) use keras from tensorflow package
  • sklearn ($pip install sklearn)

• R 4.0

  • Toolbox from github (https://github.com/ABorrel/Toolbox_packageR)
  • Clone QSAR R scripts available at https://github.com/ABorrel/QSAR-QSPR

• set up the path for a internal run of the R script in the python script source/py/runExternal.py

Prediction - QSAR models

• genotoxicity: TEST (https://www.epa.gov/chemical-research/toxicity-estimation-software-tool-test#install)
• ER-AR prediction: OPERA2.7 (https://github.com/kmansouri/OPERA)
  
  
  

Input data

The input folder included input file to reproduce the analysis:

• list of exposure "BCRelExposureSources_P65_051221.csv"
• list of chemicals considered "cross_lists_for_analysis_042522.xlsx"
• list of hormones "hormones.csv"
• QSAR models "QSARs/"
  
  
  

Run the script

A Jupiter notebook was developed to run all analysis partially or completely and apply QSAR models.
Jupiter notebook