/Polymer2Vec

Embedding representation of polymer derived from PI1M dataset

Primary LanguagePythonMIT LicenseMIT

Polymer2Vec

Embedding representation of polymer derived from PI1M dataset

Installation Requirements

  • Python 3.x
  • pandas
  • numpy
  • rdkit
  • LightGBM
  • Optuna
  • scikit-learn
  • gensim
  • mol2vec

Model Training with quick_qspr.py

Use the following command to quickly train a QSPR model:

python quick_qspr.py -in your_data.csv -x smi -y tg -o your_model

Command-Line Arguments

  • -in: The name of the input CSV file.

    • Example: train_test.csv

  • -x: The name of the SMILES column.

    • Example: smi

  • -y: The name of the target column.

    • Example: tg

  • -o: The name of trained model will be saved.

    • Example: your_model

References

  1. Machine learning discovery of high-temperature polymers

  2. Mol2vec: Unsupervised Machine Learning Approach with Chemical Intuition