Author: Tarapong Srisongkram

This is a simple command-line tool to load and use a pre-trained EnRAQSAR model on your local computer. With this tool, you can make skin irritation predictions directly from the terminal without the need for a GUI or web interface. This software is based on the paper: Ensemble Read-Across Quantitative Structure-Activity Relationship Algorithm for Predicting Skin Cytotoxicity

1. Python: Ensure you have Python 3.9 or newer installed on your machine. We recommend you install Python via Anaconda package management.

2. Required Libraries: Before you start, make sure to install the required Python libraries by running (in terminal):

First locate to the path of this software

    cd path of the software

Then create Conda environment name raqsar and then install the requirements.txt file (you can change environment name raqsar as you want).

    conda create --name raqsar --file requirements.txt

Alternatively, you can install requirements.txt file via pip python package management

    pip install -r requirements.txt

3. Input File: This software accepts input as an Excel file. The format of Excel should include LigandID as index column, canonical_smiles as the SMILES column. for example.

The test of the compound is unlimited. You can put the compound with SMILES as much as you want. Then, place the input file in the folder input

Activate raqsar Conda environment

conda activate raqsar

Analyze the chemicals in the INPUT_FILE_NAME in the input folder and save in the output folder as CSV file.

python main.py --input INPUT_FILE_NAME

--input : This option is available for specifies the input file name INPUT_FILE_NAME, default is example.xlsx.

Default will return the output from the example.xlsx file

python main.py

Specify the INPUT_FILE_NAME

python main.py --input INPUT_FILE_NAME

--output: This option is available for specifies the output file name OUTPUT_FILE_NAME. The default is output.csv, and this file will save in the output folder

python main.py --input INPUT_FILE_NAME --output OUTPUT_FILE_NAME

1. Predict skin irritation of Ciprofloxacin and save in output.csv:

python main.py --input ciplofoxacin.xlsx

2. (Option) Predict skin irritation of Ciprofloxacin and save in result.csv:

python main.py --input ciplofoxacin.xlsx --ouput result.csv

The example of result is shown below:

LigandID : Chemical Name
ECFP : Extended circular fingerprint (ECFP)-based Read-Across predictive result
APF : Atom Pairs 2D fingerprint (APF)-based Read-Across predictive result
RF : Physicochemical properties-based RF-QSAR predictive result
Lasso : Physicochemical properties-based Lasso-QSAR predictive result
Predicted pIC50 : Stacked ensemble of Lasso-QSAR, RF-QSAR, ECFP-RA, APF-RA based prediction (final pIC50 result). High value (pIC50 > 5) indicates skin irritation chemical
Predicted IC50 : IC50 values calculated from the Predicted pIC50 by using pIC50 = -logIC50 equation.
Similarity : Tanimoto-based similarity to the training data set (excluded outliers). A higher similarity value indicates better prediction performance.

If you'd like to contribute to this project, please fork the repository, make your changes, and submit a pull request.

This project is licensed under the MIT License.

If you have feedback or run into any issues, please submit an issue on the project's GitHub page.