Journal of Cheminformatics最新文献

Drug combination therapy is a well-established strategy for treating complex diseases. However, the vast combinatorial space renders exhaustive experimental screening impractical and costly. Recent studies have shown that deep learning techniques can effectively prioritize synergistic drug combinations by leveraging their powerful nonlinear modeling and automatic feature extraction capabilities. Meanwhile, Large Language Models (LLMs) offer great promise in drug discovery. In this paper, we propose CoSynLLM, an LLM-assisted predictive framework for predicting drug combination synergy. We fully leverage the latent knowledge embedded in LLMs to generate semantic-level chemical information, complemented by drug fingerprints to incorporate explicit structural details, while cell line gene expression profiles represent the cellular context. To effectively merge drug and cell line representations, a hierarchical feature fusion strategy is employed to progressively integrate features through multiple stages for predicting drug combination synergy. Extensive experiments on two benchmark datasets, NCI-ALMANAC and O'Neil, demonstrate that CoSynLLM achieves competitive performance, highlighting its effectiveness in predicting drug combination synergy. In summary, CoSynLLM effectively identifies synergistic drug combinations, offering a robust and practical computational framework for predicting drug combination synergy.

Limited experimental data remains a key challenge in applying machine learning to drug discovery, particularly for cancer-related targets. In this study, we present a data-efficient active meta-deep learning framework to predict mitogen-activated protein kinase 1 (MAPK1) inhibitors, which are promising candidates for cancer-related therapies. Our approach integrates active learning (AL) with a meta-model that combines four deep architectures: a convolutional neural network, an attention, a graph convolutional network, and a graph neural network-attention, trained on molecular descriptors and graph-based representations. These models generate four probability-based features that feed into an attention-based meta-learner, improving predictive performance by 5.12% in the area under the precision-recall curve (AUPRC) and 5.48% in the Matthews correlation coefficient (MCC) using only 10% of the training data. Among the AL sampling strategies evaluated, entropy sampling showed competitive performance in selecting informative molecules for model improvement. Overall, our framework achieves an AUPRC of 0.835 ± 0.017 and MCC of 0.817 ± 0.017, on par with a traditional training method despite using only 26.7% of the training data. Compared to a conventional random forest model trained on brute-force, a 100% full training set, our approach shows a 10.6% improvement in AUPRC and modest gains in MCC, confirming the effectiveness of the proposed framework. Under severe class imbalance, balanced accuracy steadily increased across AL iterations, reaching values greater than 0.85 at the final iteration for all uncertainty-driven strategies. Molecular docking confirmed successful prioritization of the top four predicted compounds. Evaluation on an external MAPK1 data set demonstrated generalizability, with our approach achieving an AUPRC of 0.818 and an MCC of 0.403, comparable to the independent test set. These results highlight the potential of combining intelligent data selection with deep learning architectures through the meta-model to accelerate predictive performance in data-scarce drug discovery. Scientific contribution: This study contributes a novel, data-efficient active meta-deep learning framework for predicting MAPK1 inhibitors, addressing the challenge of limited experimental data in a cancer-specific target. By integrating AL with a meta-model composed of four deep architectures, the approach significantly enhances the predictive performance using only a fraction of the training data. The framework achieves superior metrics compared to traditional training methods, highlighting its potential to accelerate drug discovery in data-scarce settings.

UDP-glucuronosyltransferases (UGTs) play a critical role in drug metabolism by catalyzing the glucuronidation of structurally diverse compounds. However, accurately predicting UGT-mediated sites of metabolism (SOMs) remains a challenge due to the limited availability of annotated data. In this study, we introduce UGTformer, a unified graph transformer-based framework that simultaneously performs UGT substrate classification and SOM prediction. UGTformer employs a hierarchical architecture integrating multi-hop message propagation with hop-aware and node-level transformer encoders. The model was pretrained on large-scale molecular graphs via chemically informed self-supervised tasks, and fine-tuned on a manually curated UGT metabolism data set covering four major metabolic reaction categories. In five-fold cross-validation, UGTformer achieved an AUC of 0.833 for substrate classification and 0.884 for SOM identification, outperforming multiple GNN baselines. On an independent external validation set, it maintained robust performance, demonstrating strong generalization to previously unseen molecules. By integrating chemically meaningful structural encodings and a joint learning paradigm, UGTformer delivers interpretable and biologically consistent predictions, offering a reliable and scalable approach for UGT-related metabolism prediction. The UGTformer model is freely accessible at https://lmmd.ecust.edu.cn/UGTformer/.

Discovering novel drug candidates remains a considerable challenge in pharmaceutical research. Generative AI models such as Generative Adversarial Networks (GANs) have shown considerable promise in de novo molecular generation. They demonstrated high potential in drug discovery applications, yet they often face challenges such as limited chemical coverage and mode collapse. In the present study, we developed MolGAN-QRL, a hybrid quantum-classical framework that introduced quantum-enhanced reinforcement learning within the MolGAN architecture to address these limitations. The proposed framework leveraged a hybrid reward mechanism to further optimize chemical validity, uniqueness, and drug-likeliness of the generated molecules. Experimental results demonstrated that MolGAN-QRL consistently achieved enhanced generative performances compared to classical MolGAN, with up to a 16-fold increase in the count of unique and valid generated compounds under certain conditions. These gains reflected the effectiveness of quantum-guided exploration and highlighted the known trade-off between uniqueness and validity in generative chemistry. Overall, our findings underlined the value of quantum-enhanced reward modeling in mitigating mode collapse and advancing molecular generation, and support the potential of hybrid quantum-classical methods to advance generative chemistry for drug discovery applications. SCIENTIFIC CONTRIBUTION: MolGAN-QRL introduces the first variant of the MolGAN framework, that is augmented with a variational quantum circuit (VQC) within the reinforcement-learning reward module, rather than the generator, the discriminator or the noise function. It leverages a hybrid reward mechanism that trains a quantum-classical function that lead to better mitigation of mode-collapse, through higher uniqueness scores and 16-fold more novel, valid and unique molecules generated.

We developed SPARFlow, an open-source KNIME workflow for structure-activity or structure-property relationship (SAR/SPR) analyses. The workflow integrates data preprocessing, chemical structure curation, similarity network construction, maximum common substructure detection, R-group decomposition, activity cliff identification, and database modelability assessment. It implements established indices, including SALI, SARI, MODI*, and RMODI, to characterize SAR landscapes and assess dataset suitability for predictive modeling. SPARFlow was validated using four datasets with distinct chemical and endpoint characteristics: cruzain inhibitors, biased μ-opioid receptor agonists, pesticides, and carbonyl compounds with hydration constants.Scientific ContributionThis work introduces SPARFlow, an KNIME-integrated workflow that combines data curation, activity-cliff detection, and modelability assessment for SAR and SPR studies. The workflow provides a unified implementation of key SAR analyses within a single KNIME pipeline. It updates implementations of established metrics, including MODI* and RMODI, together with complementary indices such as SARI and SALI. It ensures consistent data flow across all modules.

Accurate prediction of compound-protein interactions (CPIs) is crucial for chemical biology and drug discovery. Despite recent advancements, existing deep learning (DL)-based CPI models often struggle to simultaneously achieve high generalization performance, quantify prediction confidence, and ensure explainability. Here, we propose ChemGLaM, a chemical genomics language model designed to address these three crucial challenges, thereby enabling reliable and explainable CPI predictions. ChemGLaM integrates independently pre-trained chemical and protein language models through an interaction block with a cross-attention mechanism, achieving near state-of-the-art performance in predicting novel CPIs at a low computational cost. Incorporating uncertainty estimation and attention visualization enables ChemGLaM to enhance the success rate of virtual screening and to provide molecular insights into CPIs. To demonstrate the practical impact of ChemGLaM, we constructed a publicly available database containing large-scale CPI predictions for every possible pairing between all 20,434 human proteins and all 11,455 drugs and validated its practical applicability in a case study on amyotrophic lateral sclerosis. ChemGLaM marks an important step forward in addressing the challenges of AI-driven CPI exploration and drug discovery.Scientific ContributionThis study established a unified CPI prediction framework that simultaneously achieves high generalization performance, confidence quantification, and explainability. We leveraged this framework to create a community resource by constructing a comprehensive CPI database and demonstrated its practical utility by successfully prioritizing hit compounds and deconvoluting their targets in a phenotypic screening for amyotrophic lateral sclerosis.