Predicting Aqueous Solubility of Organic Molecules Using Deep Learning Models with Varied Molecular Representations

TL;DR

This study develops deep learning models using various molecular representations to predict aqueous solubility of organic molecules, achieving high accuracy and analyzing factors influencing model performance across a large dataset.

Contribution

It introduces a comprehensive comparison of molecular representations and neural network architectures for solubility prediction, highlighting the effectiveness of molecular descriptors and GNNs.

Findings

Molecular descriptors yield the best prediction performance.

Graph neural networks perform comparably well.

Model accuracy improves with larger datasets.

Abstract

Determining the aqueous solubility of molecules is a vital step in many pharmaceutical, environmental, and energy storage applications. Despite efforts made over decades, there are still challenges associated with developing a solubility prediction model with satisfactory accuracy for many of these applications. The goal of this study is to develop a general model capable of predicting the solubility of a broad range of organic molecules. Using the largest currently available solubility dataset, we implement deep learning-based models to predict solubility from molecular structure and explore several different molecular representations including molecular descriptors, simplified molecular-input line-entry system (SMILES) strings, molecular graphs, and three-dimensional (3D) atomic coordinates using four different neural network architectures - fully connected neural networks (FCNNs), recurrent neural networks (RNNs), graph neural networks (GNNs), and SchNet. We find that models using molecular descriptors achieve the best performance, with GNN models also achieving good performance. We perform extensive error analysis to understand the molecular properties that influence model performance, perform feature analysis to understand which information about molecular structure is most valuable for prediction, and perform a transfer learning and data size study to understand the impact of data availability on model performance.