Modern agrochemicals must strike the right balance across a large panel of target properties from biological efficacy, environmental impact, resistance management, and cost of goods. This is arguably one of the most complex optimization tasks in the chemical industry. Recent breakthroughs in inverse design and generative chemistry enable to rethink this optimization approach.^1,2 Successful adoption of inverse design as research strategy, requires high quality data to build accurate models for relevant target properties. Most importantly, compounds need to be designed that can be readily synthesized. To address these challenges, Syngenta Crop Protection Research Chemistry has initiated an ambitious program to overhaul the whole software infrastructure that supports chemical synthesis from idea to physical sample.

In this presentation we will describe the main concepts and philosophy that went into the design of the platform and how it enables to integrate recent cutting-edge technology in a production environment that will ultimately serve hundreds of chemists worldwide. We will highlight the underlying modeling of chemical information and incorporation of large-scale reaction data for reaction prediction and mapping of synthesis targets and routes against the network of known organic reactions. Several challenges that are subject to current research will be touched.

[1] Vanhaelen, Q.; Lin, Y.-C.; Zhavoronkov, A. The Advent of Generative Chemistry. ACS Med. Chem. Lett. 2020, 11 (8), 1496–1505. https://doi.org/10.1021/acsmedchemlett.0c00088
[2] Sanchez-Lengeling, B.; Aspuru-Guzik, A. Inverse Molecular Design Using Machine Learning: Generative Models for Matter Engineering. Science (80). 2018, 361 (6400), 360–365. https://doi.org/10.1126/science.aat2663