Application of automated network generation for retrosynthetic planning of potential corrosion inhibitors

Retrosynthesis is the process of designing chemical pathways from a set of reactants to a set of desired products. However, when both the pools of potential reactants and products grow to a substantial size, this becomes infeasible without the aid of computational tools. This work uses Pickaxe, an automated network generation tool, to perform computational retrosynthesis on a pool of 297 bioprivileged candidate molecules as reactants and 44 003 potential corrosion inhibitors that were generated by a variational autoencoder. Unlike typical approaches in computational synthesis planning, the use of automated network generation allows flexibility in pathways and starting material beyond those that are documented. This work starts by replicating known pathways to corrosion inhibitors from a single bioprivileged candidate molecule and applying the constituent reaction families to the entirety of the reactant pool and concludes by generating networks with a more extensive reaction family list and two sets of co-reactants, or "helper molecules". Network size, both from the perspective of total reactions enacted and total products formed, was analyzed. This work uses automated network generation, specifically the Python-based tool Pickaxe, for retrosynthetic planning towards making potential corrosion inhibitors from a pool of candidate bioprivileged molecules.