fgqsar_gen

FG-guided QSAR-based molecule generation for AIDD.

Features

• RF + MACCS QSAR prediction
• SHAP-derived functional group guidance
• Scaffold-based generation using dummy atoms
• Returns top hits with design explanations

Usage

from fgqsar_gen.pipeline import design_molecules

fgs = {
    "Me": "[*]C",
    "Cl": "[*]Cl",
    "NH2": "[*]N",
    "OMe": "[*]OC"
}

scaffold = "c1ccc([*])cc1[*]"
df = design_molecules(scaffold, fgs, "models/rf_model.pkl")

print(df)