CRUSH-Cleavage Rules Using SMIRKS Heuristics: an enhanced molecular fragmentation algorithm.

Area of Science:

• Computational Chemistry
• Chemoinformatics
• Drug Discovery

Background:

• Molecular fragmentation is crucial for drug discovery, virtual screening, and de novo design.
• Existing methods are limited by classical retrosynthetic rules, restricting chemical space exploration.
• There is a need for advanced fragmentation techniques that offer greater flexibility and broader applicability.

Purpose of the Study:

• To introduce CRUSH (Cleavage Rules Using SMIRKS Heuristics), an enhanced, chemistry-aware molecular fragmentation algorithm.
• To extend traditional retrosynthetic strategies for fine-grained molecular deconstruction.
• To enable systematic exploration of multiple disconnection pathways for broader chemical space coverage.

Main Methods:

• CRUSH employs a flexible bond cleavage strategy using a curated set of SMIRKS-based heuristics.
• Heuristics are exhaustively applied across eligible bonds at each fragmentation step.
• A unified benchmarking framework was used across diverse datasets (drugs, natural products, peptides, macrocycles, alimentary compounds).

Main Results:

• CRUSH consistently increases fragment yield compared to established methods.
• It substantially expands chemical space coverage, accessing unexplored regions.
• CRUSH generates smaller, less complex fragments, advantageous for fragment-based and diversity-driven workflows.
• Results are consistent across low- and high-resolution molecular fingerprint representations.

Conclusions:

• CRUSH is a robust and versatile molecular fragmentation framework.
• It complements and, in exploratory contexts, surpasses current methods.
• CRUSH provides a strong foundation for fragment-based chemoinformatics applications across multiple chemistry domains.