Journal of Cheminformatics最新文献

When data availability is limited, the prediction of properties through purely data-driven machine learning (ML) is challenging. Integrating physically-based modeling techniques into ML methods may lead to better performance. In a recent work by Chew et al. (“Advancing material property prediction: using physics-informed machine learning models for viscosity”) descriptors from classical molecular dynamics (MD) simulations were included into a quantitative structure–property relationship to accurately predict temperature-dependent viscosity of pure liquids. Through feature importance analysis, the authors found that heat of vaporization was the most relevant descriptor for the prediction of viscosity. In this comment, we would like to discuss the physical origin of this finding by referring to Eyring’s rate theory, and develop an alternative modeling approach using a thermodynamic-based architecture that requires less input data.

Cyclic peptides are promising drug candidates due to their ability to modulate intracellular protein–protein interactions, a property often inaccessible to small molecules. However, their typically poor membrane permeability limits therapeutic applicability. Accurate computational prediction of permeability can accelerate the identification of cell-permeable candidates, reducing reliance on time-consuming and costly experimental screening. Although deep learning has shown potential in predicting molecular properties, its application in permeability prediction remains underexplored. A systematic evaluation of these models is important to assess current capabilities and guide future development. In this study, we conduct a comprehensive benchmark of 13 machine learning models for predicting cyclic peptide membrane permeability. These models cover four types of molecular representations: fingerprints, SMILES strings, molecular graphs, and 2D images. We use experimentally measured PAMPA permeability data from the CycPeptMPDB database, comprising nearly 6000 cyclic peptides, and evaluate performance across three prediction tasks: regression, binary classification, and soft-label classification. Two data-splitting strategies, random split and scaffold split, are used to assess the generalizability of trained models. Our results show that model performance depends strongly on molecular representation and model architecture. Graph-based models, particularly the Directed Message Passing Neural Network (DMPNN), consistently achieve top performance across tasks. Regression generally outperforms classification. Scaffold-based splitting, although intended to more rigorously assess generalization, yields substantially lower model generalizability compared to random splitting. Comparing prediction errors with experimental variability highlights the practical value of current models while also indicating room for further improvement.

Bitter peptides (BPs), derived from the hydrolysis of proteins in food, play a crucial role in both food science and biomedicine by influencing taste perception and participating in various physiological processes. Accurate identification of BPs is essential for understanding food quality and potential health impacts. Traditional machine learning approaches for BP identification have relied on conventional feature descriptors, achieving moderate success but struggling with the complexities of biological sequence data. Recent advances utilizing protein language model embedding and meta-learning approaches have improved the accuracy, but frequently neglect the molecular representations of peptides and lack interpretability. In this study, we propose xBitterT5, a novel multimodal and interpretable framework for BP identification that integrates pretrained transformer-based embeddings from BioT5+ with the combination of peptide sequence and its SELFIES molecular representation. Specifically, incorporating both peptide sequences and their molecular strings, xBitterT5 demonstrates superior performance compared to previous methods on the same benchmark datasets. Importantly, the model provides residue-level interpretability, highlighting chemically meaningful substructures that significantly contribute to its bitterness, thus offering mechanistic insights beyond black-box predictions. A user-friendly web server (https://balalab-skku.org/xBitterT5/) and a standalone version (https://github.com/cbbl-skku-org/xBitterT5/) are freely available to support both computational biologists and experimental researchers in peptide-based food and biomedicine.

Accurate chemical reaction prediction is critical for reducing both cost and time in drug development. This study introduces ReactionT5, a transformer-based chemical reaction foundation model pre-trained on the Open Reaction Database—a large publicly available reaction dataset. In benchmarks for product prediction, retrosynthesis, and yield prediction, ReactionT5 outperformed existing models. Specifically, ReactionT5 achieved 97.5% accuracy in product prediction, 71.0% in retrosynthesis, and a coefficient of determination of 0.947 in yield prediction. Remarkably, ReactionT5, when fine-tuned with only a limited dataset of reactions, achieved performance on par with models fine-tuned on the complete dataset. Additionally, the visualization of ReactionT5 embeddings illustrates that the model successfully captures and represents the chemical reaction space, indicating effective learning of reaction properties.

Graphical Abstract

Purpose

Drug-induced liver injury (DILI) is a significant concern in drug development, often leading to the discontinuation of clinical trials and the withdrawal of drugs from the market. This study explores the application of graph neural networks (GNNs) for DILI prediction, using molecular graph representations as the primary input.

Methods

We evaluated several GNN architectures, including Graph Convolutional Networks (GCNs), Graph Attention Networks (GATs), Graph Sample and Aggregation (GraphSAGE), and Graph Isomorphism Networks (GINs), using the latest FDA DILI dataset and other molecular property prediction datasets. We introduce a novel approach that creates a custom graph dataset, driven by molecular optimisation, that incorporates detailed and realistic chemical features such as bond lengths and partial charges as input into the GNN models. We have named our model approach DILIGeNN.

Results

DILIGeNN achieved an AUC of 0.897 on the DILI dataset, surpassing the current state-of-the-art model in the DILI prediction task. Furthermore, DILIGeNN outperformed the state-of-the-art in other graph-based molecular prediction tasks, achieving an AUC of 0.918 on the Clintox dataset, 0.993 on the BBBP dataset, and 0.953 on the BACE dataset, indicating strong generalisation and performance across different datasets.

Conclusion

DILIGeNN, utilising a single graph representation as input, outperforms the state-of-the-art methods in DILI prediction that incorporate both molecular fingerprint and graph-structured data. These findings highlight the effectiveness of our molecular graph generation and the GNN training approach as a powerful tool for early-stage drug development and drug repurposing pipeline.

Scientific Contribution: DILIGeNN is a GNN framework that extracts graph features from 3D optimised molecular structures as is done in target-based drug discovery and molecular docking simulation. Our method is the first to encode spatial and electrostatic information into a single graph representation, as opposed to other work that require multiple graphs or additional chemical descriptors for feature representation. Our approach, using warm starts following repeated early stopping during training, outperforms the current state-of-the-art methods in liver toxicity (DILI), permeability (BBBP) and activity (BACE) prediction tasks.

Graphic Abstract

Inference of molecules with desired activities/properties is one of the key and challenging issues in cheminformatics and bioinformatics. For that purpose, our research group has recently developed a state-of-the-art framework mol-infer for molecular inference. This framework first constructs a prediction function for a fixed property using machine learning models, which is then simulated by mixed-integer linear programming to infer desired molecules. The accuracy of the framework heavily relies on the representation power of the descriptors. In this study, we highlight a typical class of non-isomorphic chemical graphs with reasonably different property values that cannot be distinguished by the standard “two-layered (2L) model" of mol-infer. To address this distinguishability problem of the 2L model, we propose a novel family of descriptors, named cycle-configuration (CC), which captures the notion of ortho/meta/para patterns that appear in aromatic rings, which was impossible in the framework so far. Extensive computational experiments show that with the new descriptors, we can construct prediction functions with similar or better performance for all 44 tested chemical properties, including 27 regression datasets and 17 classification datasets comparing with our previous studies, confirming the effectiveness of the CC descriptors. For inference, we also provide a system of linear constraints to formulate the CC descriptors as linear constraints. We demonstrate that a chemical graph with up to 50 non-hydrogen vertices can be inferred within a practical time frame.

Accurate prediction of toluene/water partition coefficients of neutral species is crucial in drug discovery and separation processes; however, data-driven modeling of these coefficients remains challenging due to limited available experimental data. To address the limitation of available data, we apply multi-fidelity learning approaches leveraging a quantum chemical dataset (low fidelity) of approximately 9000 entries generated by COSMO-RS and an experimental dataset (high fidelity) of about 250 entries collected from the literature. We explore the transfer learning, feature-augmented learning, and multi-target learning approaches in combination with graph neural networks, validating them on two external datasets: one with molecules similar to training data (EXT-Zamora) and one with more challenging molecules (EXT-SAMPL9). Our results show that multi-target learning significantly improves predictive accuracy, achieving a root-mean-square error of 0.44 (log {P}) units for the EXT-Zamora, compared to a root-mean-square error of 0.63 (log {P}) units for single-task models. For the EXT-SAMPL9 dataset, multi-target learning achieves a root-mean-square error of 1.02 (log {P}) units, indicating reasonable performance even for more complex molecular structures. These findings highlight the potential of multi-fidelity learning approaches that leverage quantum chemical data to improve toluene/water partition coefficient predictions and address challenges posed by limited experimental data. We expect the applicability of the methods used beyond just toluene/water partition coefficients.

This editorial presents an analysis of the articles published in the Journal of Cheminformatics Special Issue “AI in Drug Discovery”. We review how novel machine learning developments are enhancing structural-based drug discovery; providing better forecasts of molecular properties while also improving various elements of chemical reaction prediction. Methodological developments focused on increasing the accuracy of models via pre-training, estimating the accuracy of predictions, tuning model hyperparameters while avoiding overfitting, in addition to a diverse range of other novel and interesting methodological aspects, including the incorporation of human expert knowledge to analysing the susceptibility of models to adversary attacks, were explored in this Special Issue. In summary, the Special Issue brought together an excellent collection of articles that collectively demonstrate how machine learning methods have become an essential asset in modern drug discovery, with the potential to advance autonomous chemistry labs in the near future.

Graphical Abstract

Camelid heavy-chain only antibodies consist of two heavy chains and single variable domains (VHHs), which retain antigen-binding functionality even when isolated. The term “nanobody” is now more generally used for describing small, single-domain antibodies. Several antibody generative models have been developed for the sequence and structure co-design of the complementarity-determining regions (CDRs) based on the binding interface with a target antigen. However, these models are not tailored for nanobodies and are often constrained by their reliance on experimentally determined antigen–antibody structures, which are labor-intensive to obtain. Here, we introduce NanoDesigner, a tool for nanobody design and optimization based on generative AI methods. NanoDesigner integrates key stages—structure prediction, docking, CDR generation, and side-chain packing—into an iterative framework based on an expectation maximization (EM) algorithm. The algorithm effectively tackles an interdependency challenge where accurate docking presupposes a priori knowledge of the CDR conformation, while effective CDR generation relies on accurate docking outputs to guide its design. NanoDesigner approximately doubles the success rate of de novo nanobody designs through continuous refinement of docking and CDR generation.

Chemoinformatics is a rapidly advancing field that integrates chemistry, computer science, and data analysis to enhance the study and application of chemical systems. This interdisciplinary approach leverages computational tools and large datasets to drive innovation in various chemical disciplines, including drug discovery, materials science, and environmental chemistry. Recent advancements in artificial intelligence (AI) and machine learning (ML) have significantly improved the ability to analyze complex datasets, predict molecular properties, and design new compounds. Additionally, the expansion of open-access databases and collaborative platforms has facilitated broader access to chemical data and fostered global research collaboration. Sophisticated molecular modeling techniques, such as multi-scale modeling and free energy calculations, have enhanced the accuracy of predictions, while big data analytics has enabled the extraction of valuable insights from vast datasets. Emerging technologies, including quantum computing, hold promise for further revolutionizing the field by offering new capabilities for simulating and optimizing chemical processes. Despite these advancements, chemoinformatics faces challenges related to data integrity, computational demands, and interdisciplinary collaboration. Addressing these challenges is crucial for the continued growth and effectiveness of chemoinformatics. Overall, the field is poised to play a pivotal role in advancing chemical research and developing innovative solutions to address global challenges.

Scientific contribution This article highlights the growing impact of chemoinformatics in modern chemistry by integrating computational tools with molecular science to enhance data-driven discovery. It explores advancements in machine learning, artificial intelligence, and big data analytics, which improve molecular property predictions and accelerate chemical innovations. The study also discusses key applications in drug design and materials science, demonstrating how chemoinformatics drives efficiency and sustainability in research. Additionally, it outlines future challenges and opportunities, emphasizing the need for improved algorithms, data standardization, and interdisciplinary collaboration. This work contributes to the evolving role of chemoinformatics as a crucial pillar of modern chemical research.