Over the past decades, synthetic molecule pharmaceuticals have continuously expanded and advanced, resulting in discovery market size expected to hit more than 100 billion USD by 2030. Along with the growth, synthetic molecule drugs became dramatically complex and heavier in molecule weight. Therefore, demands to obtain properties of APIs in an accelerated manner have increased, while difficulty of predicting properties have also increased.
To accommodate needs and accelerated pharmaceutical development at Takeda’s pipeline drugs, we developed predictive solubility workflows, leveraging in-silico models and automated high-throughput experimentation (HTE) platforms. Starting from generations of feasible solvents and anti-solvents lists with in-silico models, automated HTE measurement is followed to determine a broad range of solvents, quantifying precise solubility by high-performance liquid chromatography. The data collected is fed to refine the model prediction and expand prediction to produce solubility curves. With continuous feedback between model and automated HTE platform, broader parameter space including various temperatures and binary/tertiary solvent sets are explored, providing information about thermal and synergistic solvation effect. Then, we validate the solubility with bespoke experimentation, ensuring accurate measurement for further development in scale up and formulation. The proposed workflow shown in the study demonstrates an efficient and systematic approach toward development of accurate and validated solubility models, accelerating development of synthetic molecule drugs.