Do Chemformers Dream of Organic Matter? Evaluating a Transformer Model for Multistep Retrosynthesis

Abstract

Synthesis planning of new pharmaceutical compounds is a wellknown bottleneck in modern drug design. Template-free methods, such as transformers, have recently been proposed as an alternative to template-based methods for single-step retrosynthetic predictions. Here, we trained and evaluated a transformer model, called the Chemformer, for retrosynthesis predictions within drug discovery. The proprietary data set used for training comprised ∼18 M reactions from literature, patents, and electronic lab notebooks. Chemformer was evaluated for the purpose of both single-step and multistep retrosynthesis. We found that the single-step performance of Chemformer was especially good on reaction classes common in drug discovery, with most reaction classes showing a top-10 round-trip accuracy above 0.97. Moreover, Chemformer reached a higher round-trip accuracy compared to that of a template-based model. By analyzing multistep retrosynthesis experiments, we observed that Chemformer found synthetic routes, leading to commercial starting materials for 95% of the target compounds, an increase of more than 20% compared to the templatebased model on a proprietary compound data set. In addition to this, we discovered that Chemformer suggested novel disconnections corresponding to reaction templates, which are not included in the template-based model. These findings were further supported by a publicly available ChEMBL compound data set. The conclusions drawn from this work allow for the design of a synthesis planning tool where template-based and template-free models work in harmony to optimize retrosynthetic recommendations.

AiZynthFinder 4.0: developments based on learnings from 3 years of industrial application

“…This feature offers the possibility of applying the most suitable contemporary single-step retrosynthesis model complementing the multi-step retrosynthetic process. The ModelZoo currently supports models such as the Chemformer [ 12 , 13 ], MHNreact [ 45 ] and LocalRetro [ 46 ]. Furthermore, we have implemented the functionality of incorporating multiple expansion strategies simultaneously.…”

“…These improvements include a mechanism to prevent the formation of cycles when expanding the search tree. We have also implemented features that do not change the underlying algorithm but make the utilization of expensive models such as Chemformer more effective, including sibling node-expansion and model caching [ 12 ]. Moreover, we have expanded the search capabilities by incorporating additional search algorithms like the Breadth-First Search, Depth First Proof Number Search [ 53 ], and Retro* [ 50 ], within the search sub-package.…”