Something for nothing: improved solvation free energy prediction with Δ-learning

• Published in: Theoretical chemistry accounts 2023-10, Vol.142 (10)
• Main Authors: Meng, Fanwang, Zhang, Hanwen, Collins Ramirez, Juan Samuel, Ayers, Paul W.
• Format: Article
• Language: English
• Subjects: Atomic/Molecular Structure and Spectra; Bioavailability; Chemistry; Chemistry and Materials Science; Decision trees; Free energy; Gaussian process; Inorganic Chemistry; Learning; Organic Chemistry; Physical Chemistry; Quantum chemistry; Regression models; Solubility; Solvation; Theoretical and Computational Chemistry
• ISSN: 1432-881X; 1432-2234
• DOI: 10.1007/s00214-023-03047-z

Molecular solubility is among the key properties that determine the clinical performance of a drug candidate because poor molecular solubility often indicates inadequate bioavailability. Using the CombiSolv-Exp database, we test several models (Gaussian process regression, decision trees, k-nearest neighbors) for hydration free energies by integrating Δ -learning and a universal quantum-chemistry continuum solvation model, SMD. The optimal model is Gaussian process regression with MAE of 0.63 kcal/mol. The reported models improve the accuracy of SMD, but have negligible additional computational cost.