Integrating artificial intelligence and quantum chemistry for sustainable pharmaceutical retrosynthesis

In this context, artificial intelligence (AI) has brought about major changes to computer-assisted retrosynthetic planning; nevertheless, currently developed approaches are mainly oriented towards the prediction of reactions rather than taking both mechanistic plausibility and sustainability of such processes into account. In the current work, we introduce a clear-cut multi-criteria computational approach that combines transformer-based retrosynthetic prediction with density functional theory (DFT) computations and green chemistry considerations in one decision-making approach for sustainable drug design and synthesis planning. The novelty of the approach is based on the quantitative combination of these three aspects. Justification for the selection of ibuprofen as the primary focus for case studies is provided. When used along with quantum chemistry calculations of activation energy, HOMO-LUMO gap and Fukui indices, AI-assisted filtration reduced the number of potential reactions. To evaluate the environmental performance of this pathway, atom economy, process mass index (PMI), and E-factor were all taken into account. The optimized pathway had a score of 0.94, atom economy of 87.3%, PMI of 16.8, and E-factor of 12.6 in total. The ranking of the pathways did not change with 10% variations of the weights of the corresponding criteria, which was confirmed by sensitivity analysis. Rather than simply proposing a novel retrosynthesis method, this work presents an established methodology that includes AI prediction, mechanistic validation and sustainability impact assessment in synthetic pathway selection.