SED

Key Features:

• Lasso-Based Feature Selection on Long ECFPs: Applies Lasso regression to long extended-connectivity fingerprints to select informative substructural features associated with GPCR bioactivity.
• Deep Neural Network Regression: Utilizes selected ECFP features in a deep neural network model to predict ligand bioactivities.

Scientific Applications:

• GPCR Ligand Bioactivity Modeling: Predicts and analyzes bioactivities across multiple human GPCR subfamilies to support virtual screening and drug discovery.

Methodology:

SED represents ligands using long extended-connectivity fingerprints, performs feature selection via Lasso regression to identify key substructures, and trains a deep neural network regression model to estimate GPCR ligand bioactivities.