Machine learning in prediction of intrinsic aqueous solubility of drug‐like compounds: Generalization, complexity, or predictive ability?. (7th May 2021)

• Record Type: Journal Article
• Authors: Lovrić, Mario; Pavlović, Kristina; Žuvela, Petar; Spataru, Adrian; Lučić, Bono; Kern, Roman; Wong, Ming Wah
• Is Part Of: Journal of chemometrics. Volume 35:Number 7/8(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-05-07
• Subjects: consensus modeling -- LASSO -- LightGBM -- PCA -- permutation importance -- QSAR -- random forests; Chemistry -- Mathematics -- Periodicals; Chemistry -- Statistical methods -- Periodicals; 542.85
• DOI: 10.1002/cem.3349
• Languages: English
• ISSNs: 0886-9383

Abstract

Abstract: We present a collection of publicly available intrinsic aqueous solubility data of 829 drug‐like compounds. Four different machine learning algorithms (random forests [RF], LightGBM, partial least squares, and least absolute shrinkage and selection operator [LASSO]) coupled with multistage permutation importance for feature selection and Bayesian hyperparameter optimization were used for the prediction of solubility based on chemical structural information. Our results show that LASSO yielded the best predictive ability on an external test set with a root mean square error (RMSE) (test) of 0.70 log points, an R 2 (test) of 0.80, and 105 features. Taking into account the number of descriptors as well, an RF model achieves the best balance between complexity and predictive ability with an RMSE(test) of 0.72 log points, an R 2 (test) of 0.78, and with only 17 features. On a more aggressive test set (principal component analysis [PCA]‐based split), better generalization was observed for the RF model. We propose a ranking score for choosing the best model, as test set performance is only one of the factors in creating an applicable model. The ranking score is a weighted combination of generalization, number of features, and test performance. Out of the two best learners, a consensus model was built exhibiting the best predictive ability and generalization with RMSE(test) of 0.67 log points and a R 2 (test) of 0.81. Abstract : This work focuses on balancing the trade‐offs in machine learning that one has to deal with in QSAR, such as limiting the number of the utilized features, which can affect the models' complexity, the choice of algorithm, and how these factors affect their predictive ability. The challenges were evaluated on a created collection of intrinsic solubility data.