sleep-model 0.1.0

Your package description

Machine Learning-Guided Mapping Sleep-Promoting Volatiles in Aromatic Plants

Project Description

Project Description: This repository presents an advanced machine learning pipeline for identifying sleep-promoting volatile organic compounds (VOCs) from aromatic plants.

Installation

Dependency

The code has been tested in the following environment:

Package	Version
Python	3.8.16
conda	23.5.0
RDKit	2023.3.1

How to Use

Installation

# Use the yaml environment file on the GitHub homepage to directly copy the current environment
conda env create -f environment.yaml -n sleep_model
conda activate sleep_model
conda install jupyter notebook
conda install ipykernel
python -m ipykernel install --user --name sleep_model --display-name   "sleep_model"
jupyter notebook

File Structure

├── data/                   # Input data files
├── data_analysis/          # Data processing and analysis
├── models/                 # Pretrained base model files for Stacking model training
│   ├── RF/
│   │   ├── rf_MACCSkeys_random_0.ipynb
│   │   ├── rf_RDkit_random_0.ipynb
│   ├── SVM/
│   │   ├── svm_MACCSkeys_random_3.ipynb
│   ├── XGB/
│   │   ├── xgb_MACCSkeys_random_0.ipynb
│   │── stacking_predict.ipynb
├── predict_smiles.py 
└── README.md

These four models are the base models that we use to train the final stacking model.

Predicting

python predict_smiles.py --smiles "CC(=O)OC1=CC=CC=C1C(=O)O"

Batch prediction from CSV

Predict for a CSV file containing a SMILES column (default column name: SMILES):

python predict_smiles.py --csv path/to/input.csv --out path/to/output.csv

Customize the SMILES column name and file encoding if needed:

python predict_smiles.py \
  --csv path/to/input.csv \
  --out path/to/output.csv \
  --smiles-column SMILES \
  --input-encoding utf-8

The output CSV includes: smiles, prediction, probability, and base-model probabilities (rf_maccs_p0/p1, rf_rdkit_p0/p1, svm_maccs_p0/p1, xgb_maccs_p0/p1).