MolSearch

MolSearch performs search-based multi-objective optimization to generate small molecules that satisfy multiple property requirements for drug discovery and chemical research.

Key Features:

• Search-based multi-objective optimization: MolSearch uses a search-based framework for multi-objective optimization to balance multiple molecular property requirements.
• Two-stage iterative search: It starts from existing molecules and applies a systematic two-stage search strategy to iteratively transform them into new compounds.
• Transformation rules from compound libraries: Molecular modifications are guided by transformation rules derived exhaustively from large compound libraries.
• Computational efficiency versus deep learning: MolSearch achieves performance comparable to deep learning approaches while requiring substantially less computational resources.
• Benchmark evaluation: The method has been evaluated across multiple benchmark generation settings for producing molecules that satisfy diverse property requirements.

Scientific Applications:

• Drug discovery: Generating and optimizing small molecules to meet multiple property requirements relevant to drug discovery.
• Chemical research: Exploration of chemical space and design of compounds for chemical research applications.
• Multi-objective molecular generation and optimization: Producing molecules that satisfy diverse, simultaneous property constraints in molecular design tasks.

Methodology:

MolSearch implements a search-based framework for multi-objective optimization that starts from existing molecules and applies a systematic two-stage search strategy to iteratively transform them using transformation rules derived exhaustively from large compound libraries.