Revisiting the Pfizer Case Study: Key Takeaways

In the world of synthetic organic chemistry, the process of planning how to create new materials is crucial, yet traditionally it relies heavily on human intellect and creativity. This process, known as retrosynthetic analysis, involves working backwards from a desired compound to determine the best synthetic route. While chemists have long employed this technique through collaborative discussions and iterative brainstorming, it remains a complex task fraught with challenges. The limitations of human capacity to manage vast amounts of data can introduce biases and inefficiencies in selecting the optimal synthesis path.

Recent advancements in digital technology, particularly in the pharmaceutical sector, are transforming this traditional approach. The development of innovative digital tools is streamlining the synthesis planning process, offering a more collaborative and unbiased platform for decision-making. This is exemplified by the case of Pfizer and their use of Lotiglipron, a drug currently under development. The integration of digital solutions into synthesis planning aims to capture and utilise the collective knowledge of scientists more effectively, reducing reliance on individual human effort and mitigating unconscious biases.

The process begins with capturing initial ideas from chemists using user-friendly digital interfaces. These ideas are then enriched and transformed through sophisticated algorithms before being stored in a centralised system. The system allows for the continuous update of data, as new ideas are added and connected, ensuring that the knowledge base remains current and comprehensive. For Pfizer, this approach meant that contributions from different scientists could be integrated and built upon, facilitating more effective and innovative synthesis strategies.

Figure 1: Research article published by Digital Discovery (RSC 2024), available here.

One of the key digital tools in this endeavour is ASKCOS, a machine learning-based software developed by the Machine Learning for Pharmaceutical Discovery and Synthesis (MLPDS) consortium. ASKCOS helps generate potential synthesis routes by processing a wide array of chemical data. Although this tool provides valuable insights, it is not without its limitations. For complex molecules like Lotiglipron, which involves intricate transformations and specific chemical conditions, the software’s predictions often require manual refinement to ensure practical feasibility. The challenge lies in filtering out less relevant suggestions and focusing on the most promising routes.

In practice, integrating human insights with algorithm-generated data is essential. For Pfizer’s Lotiglipron, the team combined manually curated routes with those suggested by ASKCOS, resulting in a richer pool of ideas. This hybrid approach aims to harness the strengths of both human creativity and computational power. However, the effectiveness of these algorithms depends significantly on the quality and breadth of the data they are trained on. Current data sources, such as literature and patent repositories, often contain biases and incomplete information, which can affect the reliability of the predictions.

Ultimately, the goal is to refine the synthesis planning process by blending traditional expertise with modern digital tools. As the field advances, the continued integration of such technologies promises to enhance the efficiency and accuracy of developing new compounds. For Pfizer and similar organisations, this means more robust and innovative approaches to pharmaceutical development, paving the way for more effective treatments and breakthroughs in medicine.

Optimising Cost Efficiency, Environmental Impact, and Collaboration

The strategy outlined in the study presents a ground-breaking approach to optimising synthetic routes by balancing multiple, often conflicting priorities. This method promises to revolutionise the pharmaceutical industry by facilitating more cost-effective production and speeding up the development of new medicines. By integrating comprehensive impurity control measures into the decision-making process, the proposed approach addresses one of the industry’s stringent regulatory requirements, ensuring that undesired compounds are kept in check throughout the synthesis. Moreover, this holistic strategy aligns with environmental goals by incorporating considerations such as chemical hazards, waste minimisation, and energy consumption, potentially moving processes towards net zero emissions and reducing overall production costs.

The approach also offers significant improvements in cost estimation and operational efficiency. By embedding real-time pricing data for raw materials and pre-GMP intermediates into the synthesis planning process, predictive algorithms can provide more accurate cost estimates, enabling better financial planning and optimisation of production routes. This data-driven insight can reveal cost-effective alternatives that might otherwise be overlooked. Additionally, the approach supports the modularisation of pharmaceutical operations, allowing companies to efficiently collaborate with contract research organisations, universities, and other partners by sharing controlled portions of data. This method enhances data transfer and integration across different entities, bridging gaps in knowledge and improving the overall efficiency of chemical process development.

Our Platform Can Transform the Way You Think Chemistry

Our platform embodies the principles highlighted in the Pfizer case study, offering a comprehensive digital solution that transforms synthetic route planning into a more efficient, collaborative, and unbiased process. By leveraging advanced machine learning algorithms, similar to ASKCOS, our platform processes vast amounts of chemical data to suggest optimal synthesis routes. This helps to reduce the burden on individual chemists, mitigating biases and inefficiencies from the start of the ideation process.

We provide a user-friendly interface for initial idea capture, enriched by sophisticated algorithms, ensuring that every contribution is stored in a centralised, continuously updated system. This allows for seamless integration and collaboration among scientists, fostering innovative synthesis strategies. Our platform’s ability to incorporate real-time data, including cost and environmental impact considerations, aligns with the industry’s regulatory and sustainability goals. By modularising operations and facilitating data sharing with partners, our platform enhances overall efficiency and supports ground-breaking chemical development. Moreover, you do not need to modify your digital infrastructure to accommodate for this. Our platform can be deployed on-premise, connecting to all your systems invisibly.