About Project
Predicting the solubility of large, chameleonic molecules—compounds that alter their shape, influencing their solubility. Understanding solubility is crucial in drug development, as it determines the administration route (oral ingestion vs. intravenous) and impacts dosage considerations that can affect other organs.
Link to Project
Prediction of Molecule Solubility
Category
Machine Learning for chemical compounds
Key Highlights
- Trained a Random Forest Regressor using solubility data from small molecules, due to the scarcity of solubility data for large molecules.
- This model predicts the solubility of larger molecules by evaluating key properties: Number of Rotational Bonds, Hydrogen Bond Donors, Molecular Weight, Topological Polar Surface Area.
Application displays the 3D image of the molecule, calculated properties and the predicted solubility level.
LIBRARIES AND TOOLS
- Flask
- RDKit
- PyMol
- Scikit-learn
- Random Forest Regressor
- HTML