MOTIVATION: The field of protein-ligand binding affinity prediction continues to face significant challenges. While deep learning (DL) models can leverage 3D structural information of protein-ligand complexes, they perform well only on heavily biased test sets containing information leaked from training sets. This lack of generalization arises from the limited availability of training data and the models' inability to effectively learn from protein-ligand interactions. Since these interactions are inherently time-dependent, molecular dynamics (MD) simulations offer a potential solution by incorporating conformational sampling and providing interaction rich information. RESULTS: We have developed MDbind, a dataset comprising 63 000 simulations of protein-ligand interactions, along with novel neural networks capable of learning from these simulations to predict binding affinity. By utilizing MD as data augmentation, our models achieved state-of-the-art performance on the PDBbind v.2016 core set and an external test set, the free energy perturbation (FEP) dataset. Additionally, when trained on the full MD simulations, the models demonstrated less biased predictions. AVAILABILITY AND IMPLEMENTATION: The code for neural networks is available at https://github.com/ICOA-SBC/MD_DL_BA. The models, the results and the training/validation/test sets are available for download at https://zenodo.org/records/10390550. The MDbind trajectories are being transferred to the MDDB: https://mmb-dev.mddbr.eu/#/browse? option=mdbind.

Pan Assays Interference Compounds (PAINS) are a significant problem in modern drug discovery: compounds showing non-target specific activity in high-throughput screening can mislead medicinal chemists during hit identification, wasting time and resources. Recent work has shown that existing structural alerts are not up to the task of identifying PAINS. To address this short-coming, we are in the process of developing a tool, PrePeP, that predicts PAINS, and allows experts to visually explore the reasons for the prediction. In the paper, we discuss the different aspects that are involved in developing a functional tool: systematically deriving structural descriptors, addressing the extreme imbalance of the data, offering visual information that pharmacological chemists are familiar with. We evaluate the quality of the approach using benchmark data sets from the literature and show that we correct several short-comings of existing PAINS alerts that have recently been pointed out.

Compound selectivity is an important issue when developing a new drug. In many instances, a lack of selectivity can translate to increased toxicity. Protein kinases are particularly concerned with this issue because they share high sequence and structural similarity. However, selectivity may be assessed early on using data generated from protein kinase profiling panels. To guide lead optimization in drug discovery projects, we propose herein two new selectivity metrics, namely window score (WS) and ranking score (RS). These metrics can be applied to standard in vitro data–including intrinsic enzyme activity/affinity (Ki, IC50 or percentage of inhibition), cell-based potency (percentage of effect, EC50) or even kinetics data (Kd, Kon and Koff). They are both easy to compute and offer different viewpoints from which to consider compound selectivity. We performed a comparative analysis of their respective performance on several data sets against already published selectivity metrics and analyzed how they might influence compound selection. Our results showed that the two new metrics bring additional information to prioritize compound selection. Two novel metrics were developed to better estimate selectivity of compounds screened on multiple proteins.