ChemMORT

ChemMORT optimizes absorption, distribution, metabolism, elimination, and toxicity (ADMET) properties of small molecules to refine drug candidates while preserving bioactivity.

Key Features:

• SMILES Encoder: Simplified Molecular Input Line Entry System (SMILES) Encoder generates a 512-dimensional vector representation of molecular structures.
• Descriptor Decoder: Decoder translates the high-dimensional vector back into a corresponding molecular structure with high accuracy.
• Molecular Optimizer: Utilizes reversible molecular representation together with particle swarm optimization to modify molecular structures.
• Deep Learning: Leverages deep learning techniques for molecular representation and translation between encoder and decoder spaces.
• Multi-objective Particle Swarm Optimization: Implements multi-objective and constrained particle swarm optimization to balance ADMET properties and bioactivity.
• Inverse QSAR Modeling: Enables inverse Quantitative Structure-Activity Relationship (QSAR) style adjustments for targeted molecular design.

Scientific Applications:

• ADMET optimization in drug discovery: Optimizes absorption, distribution, metabolism, elimination, and toxicity profiles of drug candidates.
• Bioactivity-preserving optimization: Improves ADMET properties while maintaining compound bioactivity.
• Inverse QSAR and targeted design: Facilitates targeted molecular modifications using inverse QSAR approaches.
• PARP-1 inhibitor optimization: Applied to poly(ADP-ribose) polymerase-1 (PARP-1) inhibitor optimization as a case study.

Methodology:

SMILES encoding to a 512-dimensional vector, decoding vectors back to molecular structures, reversible molecular representation, deep learning techniques, and multi-objective particle swarm optimization (constrained multi-objective optimization) enabling inverse QSAR-style optimization.