The accurate prediction of molecular properties, such as lipophilicity and aqueous solubility, are of great importance and pose challenges in several stages of the drug discovery pipeline. Machine learning methods, such as graph-based neural networks (GNNs), have shown exceptionally good performance in predicting these properties. In this work, we introduce a novel GNN architecture, called directed edge graph isomorphism network (D-GIN). It is composed of two distinct sub-architectures (D-MPNN, GIN) and achieves an improvement in accuracy over its sub-architectures employing various learning, and featurization strategies. We argue that combining models with different key aspects help make graph neural networks deeper and simultaneously increase their predictive power. Furthermore, we address current limitations in assessment of deep-learning models, namely, comparison of single training run performance metrics, and offer a more robust solution.

Knowledge of the putative bound-state conformation of a molecule is an essential prerequisite for the successful application of many computer-aided drug design methods that aim to assess or predict its capability to bind to a particular target receptor. An established approach to predict bioactive conformers in the absence of receptor structure information is to sample the low-energy conformational space of the investigated molecules and derive representative conformer ensembles that can be expected to comprise members closely resembling possible bound-state ligand conformations. The high relevance of such conformer generation functionality led to the development of a wide panel of dedicated commercial and open-source software tools throughout the last decades. Several published benchmarking studies have shown that open-source tools usually lag behind their commercial competitors in many key aspects. In this work, we introduce the open-source conformer ensemble generator CONFORGE, which aims at delivering state-of-the-art performance for all types of organic molecules in drug-like chemical space. The ability of CONFORGE and several well-known commercial and open-source conformer ensemble generators to reproduce experimental 3D structures as well as their computational efficiency and robustness has been assessed thoroughly for both typical drug-like molecules and macrocyclic structures. For small molecules, CONFORGE clearly outperformed all other tested open-source conformer generators and performed at least equally well as the evaluated commercial generators in terms of both processing speed and accuracy. In the case of macrocyclic structures, CONFORGE achieved the best average accuracy among all benchmarked generators, with RDKit's generator coming close in second place.

In silico metabolism prediction models have become indispensable tools to optimize the metabolic properties of xenobiotics while preserving their intended biological activity. Among these, site-of-metabolism (SOM) prediction models are particularly valuable for pinpointing metabolically labile atomic positions. However, the practical utility of these models depends not only on their ability to deliver accurate predictions but also on their capacity to provide reliable estimates of predictive uncertainty. In this work, we introduce aweSOM, a graph neural network (GNN)-based SOM prediction model that leverages deep ensembling to model the total predictive accuracy and partition it into its aleatoric and epistemic components. We conduct a comprehensive evaluation of aweSOM's uncertainty estimates on a high-quality data set, identifying key challenges that currently constrain the performance of SOM prediction models. Based on these findings, we propose actionable insights to drive progress in the field of metabolism prediction.

The accurate prediction of molecular properties such as lipophilicity and aqueous solubility is of great importance in several stages of the drug discovery pipeline. Machine learning methods like graph-based neural networks have shown exceptionally good performance in predicting these properties. In this work we introduce a novel graph neural network architecture composed of two distinct sub-architectures that achieves an improvement in accuracy over its individual parts employing various learning-, and featurization strategies. We argue that combining models with different key aspects might help make graph neural networks deeper while simultaneously increasing their predictive power. Additionally, we want to highlight the need to move beyond comparing single performance metrics to show machine learning model superiority.

Accurate determination of the metabolic fate of xenobiotics is essential for ensuring their safety and efficacy. While in vivo and in vitro methods remain the gold standard for assessing metabolic properties, they are both costly and time-consuming. In silico metabolism prediction models offer complementary solutions to tackle the challenging task of improving a compound's metabolic stability without compromising its desired biological activity. This paper introduces aweSOM, a novel graph neural network (GNN)-based site-of-metabolism (SOM) predictor for phase 1 and 2 metabolic reactions. aweSOM leverages deep ensemble learning to decompose the total predictive uncertainty into its aleatoric and epistemic components. Through a detailed analysis of experimental results and case studies, we show how these uncertainty estimates enhance prediction reliability, provide valuable insights to guide future model improvements and data expansion strategies, and help to improve the efficiency of experimental metabolic characterization for novel compounds.

Enumerating protonation states and calculating micro-state pK _a values of small molecules is an important yet challenging task for lead optimization and molecular modeling. Commercial and non-commercial solutions have notable limitations such as restrictive and expensive licenses, high CPU/GPU hour requirements, or the need for expert knowledge to set up and use. We present a graph neural network model that is trained on 714,906 calculated mico-state pK _a predictions from molecules obtained from the ChEMBL database. The model is fine-tuned on a set of 5,994 experimental pK _a values significantly improving its performance on two challenging test sets. Combining the graph neural network model with Dimorphite-DL , an open-source program for enumerating ionization states, we have developed the open-source Python package pkasolver, which is able to generate and enumerate protonation states and calculate micro-state pK _a values with high accuracy.