Hypoxia enhances the malignant nature of bladder cancer cells and concomitantly antagonizes protein O -glycosylation extension Hypoxia enhances the malignant nature of bladder cancer cells and concomitantly antagonizes protein O-glycosylation extension

Many challenges persist in developing accurate computational models for predicting solvation free energy (ΔG_sol). Despite recent developments in Machine Learning (ML) methodologies that outperformed traditional quantum mechanical models, several issues remain concerning explanatory insights for broad chemical predictions with an acceptable speed-accuracy trade-off. To overcome this, we present a novel supervised ML model to predict the ΔG_sol for an array of solvent-solute pairs. Using two different ensemble regressor algorithms, we made fast and accurate property predictions using open-source chemical features, encoding complex electronic, structural, and surface area descriptors for every solvent and solute. By integrating molecular properties and chemical interaction features, we have analyzed individual descriptor importance and optimized our model though explanatory information form feature groups. On aqueous and organic solvent databases, ML models revealed the predictive relevance of solutes with increasing polar surface area and decreasing polarizability, yielding better results than state-of-the-art benchmark Neural Network methods (without complex quantum mechanical or molecular dynamic simulations). Both algorithms successfully outperformed previous ΔG_sol predictions methods, with a maximum absolute error of 0.22 ± 0.02 kcal mol^-1, further validated in an external benchmark database and with solvent hold-out tests. With these explanatory and statistical insights, they allow a thoughtful application of this method for predicting other thermodynamic properties, stressing the relevance of ML modeling for further complex computational chemistry problems.

A linear least-squares methodology was used to determine the vibrational scaling factors for the X3LYP density functional. Uncertainties for these scaling factors were calculated according to the method devised by Irikura et al. [J. Phys. Chem. A 109, 8430 (2005)]. The calibration set was systematically partitioned according to several of its descriptors and the scaling factors for X3LYP were recalculated for each subset. The results show that the scaling factors are only significant up to the second digit, irrespective of the calibration set used. Furthermore, multivariate statistical analysis allowed us to conclude that the scaling factors and the associated uncertainties are independent of the size of the calibration set and strongly suggest the practical impossibility of obtaining vibrational scaling factors with more than two significant digits.