In a 2020 Nature paper, researchers highlighted significant advancements in computer-aided synthetic design, demonstrating that computational planning for complex natural products is feasible. Prior to this, no algorithm could design plausible routes for such intricate compounds, which require multistep planning. The study showed that by augmenting a program's knowledge of organic chemistry with data-based AI and causal relationships, it is possible to 'strategize' over multiple synthetic steps. The routes designed were validated in the lab and were largely indistinguishable from those created by human experts.

This research underscores the capabilities of SYNTHIA^® Retrosynthesis Software, a hybrid expert-AI system developed over the last decade. Unlike earlier programs that could only 'think' one step at a time, SYNTHIA^® incorporates over 120,000 expert-coded rules, including stereoselective and scaffold-directed transformations. These high-quality rules, refined by filters using machine-learning or quantum-chemistry methods, enable SYNTHIA^® to efficiently plan syntheses of complex targets.
By leveraging SYNTHIA^®, chemists can explore innovative strategies and optimize the production of valuable compounds, enhancing the capabilities of modern synthetic chemistry

Read the full article


Mikulak-Klucznik, B., Gołębiowska, P., Bayly, A.A. et al.  
Computational planning of the synthesis of complex natural products. 
Nature 588, 83–88 (2020).  https://doi.org/10.1038/s41586-020-2855-y