STNGS: a deep scaffold learning-driven generation and screening framework for discovering potential novel...

Area of Science:

• Forensic Chemistry
• Computational Chemistry
• Pharmacology

Background:

• Global challenge in regulating novel psychoactive substances (NPSs).
• Existing methods rely on structural matching, which can be circumvented by minor chemical modifications.
• Inaccuracy and delays in current NPS supervision methods hinder effective control.

Purpose of the Study:

• To develop a systematic framework for identifying and evaluating potential NPSs.
• To overcome limitations of existing methods in NPS regulation.
• To enable proactive rather than reactive control of emerging NPSs.

Main Methods:

• Proposed a scaffold and transformer-based NPS generation and Screening (STNGS) framework.
• Utilized a scaffold-based generative model for molecule design and optimization.
• Implemented a four-part rank function including synthetic accessibility, frequency, confidence, and affinity scores.
• Integrated molecular docking and a G protein-coupled receptor (GPCR) activation-based sensor (GRAB) for evaluation.

Main Results:

• The generative model demonstrated strong performance in designing and optimizing NPS-like molecules.
• The rank function precisely positioned potential NPSs based on multiple scoring criteria.
• Successfully identified three novel synthetic cannabinoids with demonstrated activity.
• Generated a diverse and novel database of NPS-like molecules.

Conclusions:

• The STNGS framework offers a robust approach for the systematic identification and evaluation of potential NPSs.
• This method enhances the diversity and novelty of generated NPS-like molecules, aiding in proactive regulation.
• The framework assists in constraining chemical space for effective ex-ante NPS regulation.