Probing the origin of estrogen receptor alpha inhibition via large-scale QSAR study

This repository is comprised of the following folders and files that constitutes the entire workflow used in this study for the construction of QSAR models for predicting the pIC₅₀ value of estrogen receptor inhibition.

Files

File names	Description
01_ER_alpha_preparation.ipynb	Retrieves bioactivity data from ChEMBL database, curates and pre-process the data
02_ER_alpha_RO5.ipynb	Performs Lipinski's rule-of-five analysis
03_Fingerprint_gen.ipynb	Calculates fingerprint descriptors
04_Regression.ipynb	Constructs the initial QSAR models via random forest to obtain the set of top 20 descriptors
05_Regression_select_importance.ipynb	Constructs QSAR models using the top 20 descriptors
06_ER_alpha_preparation-test.ipynb	Prepares the input CSV file of the external set (< and > symbols in the bioactivity label)
07_External_test.ipynb	Applies the constructed QSAR model on the external set from 06_ER_alpha_preparation-test.ipynb.
08_Applicability_domain.ipynb	Performs applicability domain analysis via PCA bounding box approach
environment.yml	The conda environment that allows the replication of the Python environment (specific versions of installed packages) used in this study

Folders

Folder names	Description
applicability_domain	Contain CSV files and output PDF files generated via 07_Applicability_domain.ipynb
Fingerprint	Contain CSV files of fingerprint descriptors calculated by the PaDEL software
model	Contain CSV files of bioactivity data obtained programmatically from the ChEMBL database
PaDEL-Descriptor	Contain PaDEL JAR file along with fingerprint XML files
QSAR	Contain CSV files of fingerprint descriptors along with bioactivity data of all compounds used for QSAR model building
QSAR_select	Contain CSV files of the top 20 descriptors (from feature selection) used for building the final QSAR model
Result	Contain all results data
second_external_set	Contain XLSX files of the second external set where bioactivity label contains the < and > symbols
smiles	Contain SMILES data of all compounds used in this project
SubFiles	contain raw data files used in constructing plots
Train_Fp_normalized	contain fingerprint descriptors after normalized process

Citing this work

If you use these codes and data, please cite the following paper:

Citing us
Suvannang N, Preeyanon L, Malik AA, Schaduangrat N, Shoombuatong W, Worachartcheewan A, Tantimongcolwat T, Nantasenamat C. Probing the origin of estrogen receptor alpha inhibition via large-scale QSAR study. RSC Advances 8 (2018) 11344-11356.