Briefings in bioinformatics最新文献

G protein-coupled receptors (GPCRs) represent the largest membrane protein family and remain central targets in drug discovery. Ligand efficacy reflects the ability to modulate receptor conformational states and extends beyond binding affinity to underpin functional selectivity. However, most computational approaches still emphasize affinity prediction, with limited capacity to capture the conformational dynamics driving efficacy. Here, we introduce Dynamic-GLEP, a structure- and mechanism-aware framework that integrates molecular dynamics (MD)-derived conformational ensembles with transfer learning on equivariant graph neural networks. By constructing multi-conformation receptor-ligand complexes and fine-tuning the EquiScore model, Dynamic-GLEP identifies conformation-dependent interaction features to distinguish agonists from nonagonists. Applied to the 5-HT1A receptor, the framework achieved an area under the curve (AUC) of 0.74 in cross-validation and 0.71 on an external Food and Drug Administration (FDA)-related dataset. Comparative analyses showed that Holo-based models are advantageous for scaffold optimization, whereas Apo-derived ensembles provided greater adaptability to chemically diverse ligands. Furthermore, extension to the adenosine A2A receptor yielded high performance (AUC > 0.85), underscoring the method's robustness and transferability under data-scarce conditions. Collectively, these results highlight Dynamic-GLEP as a reliable and interpretable platform for ligand efficacy prediction in Class A GPCRs, with broad potential to support virtual screening, candidate prioritization, and mechanism-driven drug design.

Accurate prediction of ligand-induced activity for G-protein-coupled receptors (GPCRs) is a cornerstone of drug discovery, yet it is challenged by the need to model allosteric communication-the long-range signaling linking ligand binding to distal conformational changes. Prevailing sequence-based models often fail to capture these three-dimensional dynamics, a limitation frequently masked by averaged performance on simpler Class A targets. To address this, we introduce GPCRact, a novel framework that models the biophysical principles of allosteric modulation in GPCR activation. It first constructs a high-resolution, three-dimensional structure-aware graph from the heavy-atom coordinates of functionally critical residues at binding and allosteric sites. A dual attention architecture then captures the activation process: cross-attention encodes the initial ligand-protein interaction at the binding site, whereas self-attention learns the subsequent intra-protein signal propagation. This hierarchical architecture is built upon an E(n)-Equivariant Graph Neural Network (EGNN) to explicitly model conformational consequences of ligand binding, and is further refined with a tailored loss function and inference logic to mitigate error propagation. Underpinned by GPCRactDB, a comprehensive database we constructed for this study, GPCRact not only achieves state-of-the-art performance but also demonstrates robustly superior accuracy on a curated benchmark of allosterically complex receptors where existing models systematically underperform. Crucially, analysis of the learned attention weights confirms that the model identifies biologically validated allosteric pathways, offering a significant step toward resolving the black box nature of previous methods. Thus, GPCRact provides a more accurate, interpretable, and mechanistically-grounded solution to a long-standing challenge, paving the way for effective structure-guided drug discovery.

Molecular representation learning (MRL) is afoundation in leveraging computational methods for drug discovery, enabling the transformation of molecular structure and properties into numerical vectors. These vectors serve as input for machine learning models and facilitate the prediction and analysis of molecular attributes, functions, and reactions. The advent of foundation models has introduced both new opportunities and challenges to MRL. These models have improved generalizability and migration in scarce data. Through pretraining and fine-tuning, foundation models can be adapted to various domains. Their robust encoding and generative abilities also allow the transformation of molecular data into more expressive forms. This paper provides a detailed review of current mainstream molecular descriptors and datasets, focusing primarily on the representation of small molecules while excluding larger molecules such as proteins and peptides. It classifies foundation models into two primary categories based on the form of input: unimodal-based and multimodal-based models. For each category, representative models are identified and their advantages and disadvantages evaluated. Moreover, we systematically summarize four core pretraining strategies for MRL foundation models, analyzing their task designs, applicable scenarios, and impacts on downstream performance. In addition, the application of molecular representation foundation models in drug discovery and development is discussed, together with the current status of model interpretability. The paper concludes with insights into the future directions of MRL foundation models.

Understanding RNA conformational dynamics is essential to understand its roles in complex biological processes. While computational methods have revolutionized the prediction of static 3D RNA structures, predicting local flexibility directly from structure remains a significant challenge. We developed DeepRMSF, a deep learning-based method that leverages atomic-level descriptions of RNA to predict vibrational flexibility given a tertiary structure. Trained on MD-derived root-mean-square fluctuations(RMSF), DeepRMSF was benchmarked on 371 nonredundant RNAs, with 311 RNAs used for five-fold cross-validation (PCC = 0.7219-0.7464) and 60 RNAs as an independent test set (PCC = 0.734), ensuring minimal sequence/structural similarity between sets. DeepRMSF predicts the local flexibility of medium-sized RNAs (~75 nucleotides) in ~8.2 s, achieving >3000-fold speed-up over MD simulations while maintaining strong extrapolative accuracy. Rather than replacing MD, DeepRMSF offers a scalable and practical alternative for transcriptome-scale screening of RNA flexibility, facilitating studies on RNA structure-dynamics-function relationships and supporting computational modeling in RNA biology.

Protein-carbohydrate interactions play an important role in many biological processes and functions, like inflammation, signal transduction, and cell adhesion. In our work, we will study non-covalent carbohydrate binding sites. In this paper, we aim to build a deep-learning model to predict non-covalent protein-carbohydrate binding sites. We were motivated by the fact that experimental approaches for predicting these sites are expensive. So, computational tools are necessary for identifying these interactions. We explored several sequence-based features as well as structural features. We also leveraged protein language model embeddings. We analyzed different architectures and selected the most suitable deep learning architecture for our finalized prediction model, DeepCPBSite. DeepCPBSite is an ensemble model that combines three separate models with three approaches (random undersampling, weighted oversampling, and class-weighted loss) built on the ResNet+FNN architecture. We made separate datasets from three sources: RCSB, UniProt, and CASP. We also compared the structural features extracted from the structures predicted by AlphaFold and ESMFold in the context of our prediction tasks. We employed three different feature selection techniques and finally did a SHAP (SHapley Additive exPlanations) analysis on the structural features after categorizing the proteins based on their organism information. DeepCPBSite achieved 78.7% balanced accuracy and 59.6% sensitivity on the TS53 set, outperforming the second-best competitor, DeepGlycanSite, by 1.16% and 2.94%, respectively. Additionally, its F1, MCC, and AUPR scores outperformed other state-of-the-art methods, with improvements ranging from 3.77%-47.6%, 3.84%-32.7%, and 8.18%-60.21%, respectively.

Hypertrophic cardiomyopathy (HCM) is a condition where approximately 65% of patients exhibit myocardial fibrosis, indicated by late gadolinium enhancement, with the severity and extent of fibrosis being positively correlated with the risk of sudden cardiac death. While fibroblast activation in HCM has been noted in previous studies, the underlying regulatory mechanisms have not been thoroughly explored. In this study, we analyzed the latest single-nucleus sequencing (snRNA-seq) datasets related to HCM caused by the two most common mutations. We also examined the largest existing snRNA-seq and spatial transcriptomics datasets of HCM for external validation. Additionally, we conducted preliminary histopathological and molecular biology experiments to validate our findings and explore potential mechanisms. Our analysis revealed a phenotypic transformation of macrophages in both cases of HCM. These pro-inflammatory macrophages, driven by the high expression of ENPP2, mediated intercellular interactions that influenced fibroblast activation. The resulting increase in lysophosphatidic acid appeared to act as a plausible intermediary. Activated fibroblasts secreted substantial amounts of COL14A1, which is a critical component of myocardial fibrosis. These findings were consistent across different genetic backgrounds, suggesting their universal applicability in most HCM cases. Our study provides valuable insights into the mechanisms underlying myocardial fibrosis in HCM, highlighting the role of macrophage transformation and fibroblast activation. These findings offer potential for the identification of novel diagnostic or prognostic biomarkers and the development of targeted therapies with clinical translational potential.

Transcription factors (TFs) orchestrate cellular programs by activating or repressing gene expression in response to diverse stimuli. Although advances in experimental and computational biology have expanded our understanding of TFs, existing prediction methods still struggle to accurately capture TF-target regulatory relationships and determine their directionality (activation versus inhibition). Here, we propose ACNVE-K, an integrative framework combining k-mer decomposition with asymmetric covariance natural vector encoding to convert amino acid sequences into multidimensional feature vectors. Using Leveraging eXtreme Gradient Boosting (XGBoost), Gradient Boosting (GB), and Random Forest (RF) algorithms, we constructed five predictive models for TF identification, target gene inference, and regulatory direction classification. Benchmarking analyses demonstrated that XGBoost achieved the highest predictive performance across human and mouse genomes, particularly with updated genome annotations. The 5-mer configuration provided an optimal balance between feature richness and computational efficiency. Collectively, ACNVE-K offers a robust and interpretable framework for decoding transcriptional regulation, facilitating advances in precision medicine, regulatory genomics, and machine-learning-based gene network reconstruction.

Epistasis detection is hindered by multiple challenges, including the proliferation of analytic tools and the diverse methodological choices made in Genome-Wide Association Interaction Studies (GWAIS). These factors often produce inconsistent and only partially overlapping results, with individual methods emphasizing distinct aspects of epistasis. Although comparative evaluations of GWAIS approaches exist, they generally do not identify the factors responsible for methodological discrepancies or assess their implications for biomedical research. Consequently, it remains unclear which features of GWAIS strategies contribute most to these differences and which methods are most appropriate for revealing specific genetic architectures. Here, we present a workflow designed to characterize heterogeneity in GWAIS results and derive practical recommendations systematically. First, we assess non-replicability by comparing single nucleotide polymorphisms-pair rankings and Statistical Epistasis Networks (SENs)-graphs in which nodes represent genetic loci and edges denote epistatic interactions-to identify clusters of protocols with similar outcomes. SENs provide a structured framework for visualizing and comparing variation in epistasis detection, enabling prioritization of interactions recurrently identified across methods. Second, we propose strategies to reduce heterogeneity and enhance robustness, with particular emphasis on interpretability. Notably, we demonstrate that differences among SENs can be informative rather than disadvantageous, as they yield complementary perspectives on disease genetics. Finally, we highlight the benefits of informed SEN aggregation, showing how this approach can strengthen the utility of GWAIS for elucidating biological mechanisms relevant to disease prevention, diagnosis, and management.

Chemotherapy is an important treatment for cancer patients, but it comes with risks. Therefore, effective chemotherapy response prediction is crucial. While whole slide image provides high-resolution insights into tumour environments, existing weakly supervised learning frameworks struggle to effectively integrate molecular data, such as gene expression, limiting their predictive power in complex chemotherapy response and small-sample scenarios. We present a bimodal chemotherapy response multi-instance learning framework, BiChemoCLAM, a novel multimodal deep learning framework that combines attention-driven multiple instance learning with multimodal compact bilinear pooling for interpretable and data-efficient chemotherapy response prediction. It achieves an Area Under Curve (AUC) of 80.91%, 71.68%, and 75.80% on ovarian serous cystadenocarcinoma, colorectal adenocarcinoma, and bladder urothelial carcinoma cancer datasets, respectively. The experimental results show that BiChemoCLAM is an effective model for predicting response to chemotherapy.

Alzheimer's disease (AD) is the leading cause of dementia and represents a progressive, irreversible neurodegenerative disorder. Given the complexity and heterogeneity of AD, which involves numerous interrelated molecular pathways, large-scale proteomics datasets are essential for robust biomarker discovery. Comprehensive proteomic profiling enables the unbiased identification of novel biomarkers across diverse biological processes, thereby increasing the likelihood of finding sensitive and specific candidates for early diagnosis and therapeutic targeting. In this study, we analyzed 28 large-scale proteomics datasets obtained from the AD Knowledge Portal and published studies. The data comprise tandem mass tag, label-free quantification, and proximity extension assay measurements from brain tissue and cerebrospinal fluid. To enhance analytical power, we integrated these proteomic profiles with corresponding clinical information to construct comprehensive feature sets for subsequent machine learning analysis. Using Random Forest and Logistic Regression models, we identified a panel of proteins capable of distinguishing AD patients from healthy controls. Several of these biomarkers have been previously validated in the context of AD, while others represent novel candidates not yet reported as AD-associated. These newly identified biomarkers warrant further experimental validation and hold promise for improving early diagnosis as well as guiding the development of targeted therapies for AD.