MultiDK: A Multiple Descriptor Multiple Kernel Approach for Molecular Discovery and Its Application to The Discovery of Organic Flow Battery Electrolytes

TL;DR

MultiDK is a machine learning approach that combines multiple molecular descriptors and kernels to improve the speed and accuracy of discovering electrolyte molecules for flow batteries.

Contribution

The paper introduces MultiDK, a novel multiple descriptor multiple kernel method that enhances molecular property prediction by integrating diverse features and kernels.

Findings

Achieved r^2 = 0.92 for solubility prediction.

Improved prediction accuracy over linear regression.

Successfully predicted pH-dependent solubility of quinone molecules.

Abstract

We propose a multiple descriptor multiple kernel (MultiDK) method for efficient molecular discovery using machine learning. We show that the MultiDK method improves both the speed and the accuracy of molecular property prediction. We apply the method to the discovery of electrolyte molecules for aqueous redox flow batteries. Using \emph{multiple-type - as opposed to single-type - descriptors}, more relevant features for machine learning can be obtained. Following the principle of the 'wisdom of the crowds', the combination of multiple-type descriptors significantly boosts prediction performance. Moreover, MultiDK can exploit irregularities between molecular structure and property relations better than the linear regression method by employing multiple kernels - more than one kernel functions for a set of the input descriptors. The multiple kernels consist of the Tanimoto similarity function and a linear kernel for a set of binary descriptors and a set of non-binary descriptors, respectively. Using MultiDK, we achieve average performance of $r^2 = 0.92$ with a set of molecules for solubility prediction. We also extend MultiDK to predict pH-dependent solubility and apply it to solubility estimation of quinone molecules with ionizable functional groups as strong candidates of flow battery electrolytes.