Chemoenzymatic Synthesis Planning Guided by Reaction Type Score.

Area of Science:

• Computational Chemistry
• Organic Synthesis
• Biotechnology

Background:

• Computer-aided synthesis planning (CASP) tools are prevalent, yet specialized tools for multistep chemoenzymatic reactions remain scarce.
• Enzymatic reactions are increasingly vital in modern chemical synthesis, necessitating advanced planning strategies.

Purpose of the Study:

• To introduce a novel strategy, RTscore-guided chemoenzymatic synthesis planning (RTS-CESP), for optimizing chemoenzymatic synthesis routes.
• To enhance the efficiency and prediction accuracy of identifying suitable enzymatic steps within complex synthesis pathways.

Main Methods:

• Developed RTscore using a text-based convolutional neural network (TextCNN) to evaluate reaction efficiency and differentiate synthesis from decomposition.
• Integrated RTscore with retrosynthesis tools to rank potential chemical routes and identify optimal enzymatic replacements.
• Applied RTS-CESP to known chemoenzymatic syntheses and a large database of molecules.

Main Results:

• RTS-CESP successfully predicted known chemoenzymatic synthesis routes for 10 diverse molecules, with six identified as top-ranked.
• The strategy predicted chemoenzymatic routes for 554 out of 1000 molecules from a boutique database, surpassing the performance of the ASKCOS tool.
• A novel, experimentally validated chemoenzymatic route for Alclofenac was designed, proving shorter than existing literature methods.

Conclusions:

• RTS-CESP offers a significant advancement in chemoenzymatic synthesis planning, outperforming existing state-of-the-art tools.
• The strategy demonstrates high accuracy in predicting and designing efficient chemoenzymatic synthesis routes for complex molecules.
• RTS-CESP has the potential to accelerate drug discovery and development by optimizing synthetic pathways.