Computational Biology and Chemistry最新文献

{"title":"A unified graph-based approach for protein function prediction using AlphaFold structures and sequence features","authors":"Thi-Tuyen Nguyen ,&nbsp;Wenqing Zheng ,&nbsp;Van-Nui Nguyen ,&nbsp;Nguyen Quoc Khanh Le ,&nbsp;Matthew Chin Heng Chua","doi":"10.1016/j.compbiolchem.2025.108609","DOIUrl":"10.1016/j.compbiolchem.2025.108609","url":null,"abstract":"<div><div>Predicting protein function is a key challenge in computational biology with broad implications for understanding biological systems and disease mechanisms. Traditional deep learning approaches rely heavily on protein sequence data and protein–protein interaction (PPI) networks, often neglecting structural information due to limited availability of experimentally resolved protein structures. The advent of AlphaFold, which predicts protein structures with near-atomic accuracy, provides an opportunity to integrate structural context into function prediction. In this study, we propose StructSeq2GO, a novel hybrid model that combines structural and sequence information. StructSeq2GO employs graph representation learning to extract structural features from AlphaFold-predicted protein structures and integrates them with sequence embeddings derived from the ProteinBERT language model to predict Gene Ontology (GO) labels. Experimental evaluations demonstrate that StructSeq2GO achieves state-of-the-art performance across three GO domains, with <span><math><msub><mrow><mi>F</mi></mrow><mrow><mi>m</mi><mi>a</mi><mi>x</mi></mrow></msub></math></span> scores of 0.485, 0.681, and 0.663, AUC scores of 0.764, 0.939, and 0.891, and AUPR scores of 0.688, 0.763, and 0.702 for the Biological Process (BPO), Cellular Component (CCO), and Molecular Function (MFO) ontologies, respectively. These results highlight the critical importance of structural information and the efficacy of ProteinBERT in enhancing protein function prediction, as structure provides spatial and biochemical context not captured by sequence alone. The model’s performance is influenced by the quality of AlphaFold structural predictions and may benefit from future improvements in structure confidence modeling. Additionally, extending StructSeq2GO to predict pathway-level or disease-related annotations could broaden its biological utility.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"120 ","pages":"Article 108609"},"PeriodicalIF":3.1,"publicationDate":"2025-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144841014","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimization enabled ResNet features with transfer learning for Alzheimer’s disease detection","authors":"Deepthi K. Moorthy ,&nbsp;P. Chinnasamy ,&nbsp;P. Nagaraj","doi":"10.1016/j.compbiolchem.2025.108613","DOIUrl":"10.1016/j.compbiolchem.2025.108613","url":null,"abstract":"<div><div>Millions of individuals worldwide suffer from Alzheimer's Disease (AD), a debilitating degenerative condition. Early detection of Alzheimer's disease is critical to ensure effective treatment and better patient outcomes. In the past few years, advanced medical imaging techniques, particularly MRI, have shown potential for diagnosing Alzheimer's disease. However, developing accurate and efficient techniques for Alzheimer's disease detection offcuts a demanding duty suitable to the complication of medical images and the limited availability of labelled data. The early detection of Alzheimer's disease is critical for effective treatment and management of this debilitating neurodegenerative condition. The research proposes a novel method for Alzheimer's disease detection using an optimization-enabled ResNet feature extraction technique with transfer learning that is proposed by combining LeNet and VGG networks. The pre-processing was done using image resizing and median filter and the featureextraction was conducted using the proposed Walrus Optimization Algorithm-Residual neural network (WOA-ResNet), where WOA is employed for training ResNet. The conducted experiments with the Alzheimer’s dataset achieved a higher accuracy using the proposed LeNet-VGG method. The findings suggest that optimization-enabled ResNet feature extraction with LeNet-VGG networks can significantly improve the accuracy of Alzheimer's disease detection. The presented method achieved maximum accuracy value of 95.37 %, sensitivity value of 97.24 % and specificity value of 93.73 %.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"119 ","pages":"Article 108613"},"PeriodicalIF":3.1,"publicationDate":"2025-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144862724","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A novel network for resolving subjective masking differences and accurate thyroid nodule diagnosis","authors":"Zhiyuan Ouyang ,&nbsp;Simei Huang ,&nbsp;Liuju Liang ,&nbsp;Jianing Xu ,&nbsp;Caifen Wei ,&nbsp;Yi Zhang ,&nbsp;Hancheng Jiang ,&nbsp;Haifeng Tang ,&nbsp;Lu Wang ,&nbsp;Lin Wang ,&nbsp;Xiangzhi Li ,&nbsp;Zhenbing Liu ,&nbsp;Ruojie Zhang ,&nbsp;Lian Qin ,&nbsp;Xiaobo Yang","doi":"10.1016/j.compbiolchem.2025.108572","DOIUrl":"10.1016/j.compbiolchem.2025.108572","url":null,"abstract":"<div><h3>Background:</h3><div>Over the past three decades, there has been a significant increase in the incidence of thyroid cancer. Ultrasound serves as a non-invasive tool in differentiating between benign and malignant thyroid nodules. However, its reliance on manual input can often lead to subjective bias.</div></div><div><h3>Purpose:</h3><div>This study proposes a novel network architecture committed to diminishing subjective bias led by manual masks and enhancing the accuracy of the current models. It amalgamates multi-scale features for the effective classification of thyroid nodules.</div></div><div><h3>Methods:</h3><div>The innovative model, deemed APSNet, finds inspiration from active and passive systems. It incorporates attention mechanisms to augment nodule recognition. The model underwent training on a localized ultrasound image dataset and was tested using an external datasets TDID and TN3K. The assessment of its performance involved metrics such as Dice, IoU, F1, Acc, Sen, Spe, Ppv, Npv, and AUC, followed by statistical tests including the Friedman and DeLong tests.</div></div><div><h3>Results:</h3><div>APSNet outperformed existing models across multiple metrics, achieving an Acc of 0.9259, F1 score of 0.9540, and AUC of 0.9243 on the TDID dataset, and an Acc of 0.9287, F1 score of 0.9001, sensitivity of 0.9273, and AUC of 0.9290 on the TN3K dataset. The DeLong test confirmed its superiority, indicating statistically significant improvements over other models. Ablation Study confirms the effectiveness of Dual-System design and the potention of Transformer-based backbone.</div></div><div><h3>Conclusions:</h3><div>APSNet offers a remarkable stride forward in thyroid nodule diagnosis by effectively addressing subjectivity and amplifying feature extraction capabilities. It proffers a more accurate and dependable diagnostic tool to clinicians.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"120 ","pages":"Article 108572"},"PeriodicalIF":3.1,"publicationDate":"2025-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144852412","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Systemic Lupus Erythematosus prediction using Epistatic-Quantile Fusion Transformer network with integrated multi-omics and clinical data","authors":"Manoj B. Chandak,&nbsp;Abhijeet R. Raipurkar,&nbsp;Sunita G. Rawat","doi":"10.1016/j.compbiolchem.2025.108617","DOIUrl":"10.1016/j.compbiolchem.2025.108617","url":null,"abstract":"<div><div>Systemic Lupus Erythematosus (SLE) is a complex autoimmune disorder with heterogeneous symptoms and overlapping clinical presentations, making early prediction extremely difficult. Traditional models often fail to integrate high-dimensional multi-omics data and EHR records effectively, primarily due to their inability to handle biological variability, data imbalance, and complex feature dependencies. To address these gaps, the study proposes Epistatic-Quantile Fusion Transformer (EQF-T), a unified framework that introduces multiple novel components. Initially, for pre-processing, the Beta-Variational Rank-ordered Quantile Autoencoder (Beta-VARQA) is used, which combines Beta-divergence, Rank-ordered Quantile Filtering, and Variational Autoencoding to denoise and normalize heterogeneous inputs, retaining biologically significant patterns. For feature extraction, the framework incorporates Epistatic Attention fused Multi-Omics Laplacian Transformer (EA-MLT), which captures intricate dependencies and Epistatic Synergistic effects, essential for understanding the dynamic progression of SLE. This EA-MLT employs Epistatic Attention to capture higher-order gene-gene interactions and integrates the Multi-Omics Laplacian Transformer (MOLT), which uses a Laplacian Attention Mechanism to model structural dependencies across omics layers. The final classification is performed by SLE-Net (SLE Prediction Network), an end-to-end deep learning model designed to analyze fused data and provide interpretable outputs. Together, these components enable EQF-T to effectively learn from complex, high-dimensional biological and clinical data. Further, the proposed model achieves superior performance with 99.82 % accuracy, 99.78 % precision, 99.76 % recall, 99.77 % F1-score, and 99.8 % ROC-AUC, demonstrating its reliability and potential for precise SLE prediction.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"120 ","pages":"Article 108617"},"PeriodicalIF":3.1,"publicationDate":"2025-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144852411","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Theoretical design of metal–porphin systems for selective interaction with trace geosmin","authors":"Ali Kadhim Wadday ,&nbsp;Sukaina Tuama Ghafel ,&nbsp;Suraa Reaad ,&nbsp;Mustafa M. Kadhim","doi":"10.1016/j.compbiolchem.2025.108604","DOIUrl":"10.1016/j.compbiolchem.2025.108604","url":null,"abstract":"<div><div>Geosmin, even at minuscule concentrations, can significantly compromise the taste and odor of drinking water, making its removal a critical challenge. In this study, an integrated computational strategy encompassing DFT, QTAIM, NCI analysis, RDG, DOS, and Molecular Dynamics simulations was used to probe how geosmin interacts with both pristine Porphin and its copper-coordinated variant. The findings clearly show that copper coordination markedly enhances geosmin adsorption. Specifically, the average isosteric heat increases from 18.66 to 20.67 kcal/mol, while the energy distribution narrows, suggesting more consistent interaction strength. Mulliken charge and electrostatic potential analyses indicate a partial charge transfer involving the hydroxyl group of geosmin, supporting the conclusion that van der Waals and hydrogen-bonding interactions play a significant role. Additionally, copper coordination dramatically reduces the HOMO–LUMO gap from 0.0756 eV to 0.00083 eV signaling increased electronic reactivity and a shift towards semi-metallic character. Molecular Dynamics simulations confirm that these frameworks are both structurally and thermally stable, with total energy fluctuations remaining within ±0.17 kcal/mol. Altogether, these results highlight copper-modified Porphin as a highly promising platform for geosmin adsorption, offering robust thermal stability, enhanced selectivity, and notable electronic property modulation. Such frameworks are potential candidates for next-generation VOC sensors and advanced water purification technologies.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"119 ","pages":"Article 108604"},"PeriodicalIF":3.1,"publicationDate":"2025-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144772988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning driven identification of therapeutic phytochemicals targeting Hepatocellular carcinoma","authors":"V Vanitha Jain,&nbsp;Madhu Anabala,&nbsp;Deepak Sharma,&nbsp;Rajiniraja Muniyan","doi":"10.1016/j.compbiolchem.2025.108608","DOIUrl":"10.1016/j.compbiolchem.2025.108608","url":null,"abstract":"<div><div>Hepatocellular carcinoma (HCC), being the most common liver cancer, remains a global health concern due to the high mortality rate. HCC is also attributed to severe alcohol abuse, further leading to liver cirrhosis and cytochrome expression. The known treatments for HCC are becoming less effective with high side effects, which highlights the need for promising phytochemicals, as antioxidants, anti-inflammatories, antitumor, and other pharmacological properties. This study comprises a majorly utilized <em>in vitro</em> model for HCC, i.e., Huh 7 cell line, which was considered for retrieving the IC50 values of experimentally known inhibitors using the ChemBL database. Followed by many subsequent steps, Extra Trees Classifier and Light Gradient Boosting Machine (LGBM) showed the best performance of Receiver Operative Characteristic (ROC) of 0.91 and 0.90, respectively, as robust ML-based QSAR models. Furthermore, screening of the unknown phytochemicals and ADMET analysis showed optimum results for the phytochemicals: Bilobol, Corlumine, and Oliveotilic acid. Additionally, HSP90AA1 and CTNNB1, being the major targets with corlumine, had the best docking score of −8.66 kcal/mol and −5.21 kcal/mol, respectively, than the reference compound −8.31 kcal/mol for HSP90AA1 and −4.83 for CTNNB1 kcal/mol respectively. Further studies of molecular dynamic simulation, such as RMSD, RMSF, RG, SASA, and H-bond formation for CTNNB1- corlumine complex showed comparatively better results than HSP90AA1- corlumine complex. In a nutshell, corlumine phytochemicals, as an outcome from this study, may be used for <em>in vitro</em> and <em>in vivo</em> model testing as a novel compound as a pharmaceutical drug molecule for HCC inhibition.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"119 ","pages":"Article 108608"},"PeriodicalIF":3.1,"publicationDate":"2025-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144773001","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"PreAIS: Prediction of A-to-I editing sites based on DNN-CNN deep learning models","authors":"Xingze Fang ,&nbsp;Yun Zuo ,&nbsp;Youxu Tan ,&nbsp;Xiangrong Liu ,&nbsp;Xiangxiang Zeng ,&nbsp;Zhaohong Deng","doi":"10.1016/j.compbiolchem.2025.108612","DOIUrl":"10.1016/j.compbiolchem.2025.108612","url":null,"abstract":"<div><div>Adenosine-to-inosine RNA editing is crucial in biological processes and diseases, making A-to-I site identification key for research and drug development. However, accurate identification remains challenging due to complexity, low accuracy, and poor generalization in current models. To overcome these, a deep learning model called PreAIS has been proposed for identifying A-to-I editing sites. PreAIS first employs K-mer algorithm for feature extraction, followed by DNN-CNN for model1 construction. Finally, the model1 was trained and evaluated using 10-fold cross-validation. Compared to state-of-the-art models, PreAIS(model1) demonstrated improvements of 3.01 %, 0.67 %, and 5.04 % in Accuracy (ACC), Specificity (Sp), and Sensitivity (Sn) on Dataset 1. Additionally, using a human A-to-I RNA editing site dataset validated by Sanger sequencing, PreAIS(model1) identified 55 out of 58 sites with 94.8 % accuracy, outperforming other classifiers. To further validate the model's generalization capability, Bi-profile Bayes features were extracted from Dataset 2 for model evaluation. While keeping other parameters unchanged, only the input dimensions were adjusted to construct model2. Results from the independent test set demonstrated that even on a different dataset, our model continued to exhibit superior performance, once again surpassing the current best predictive models. Additionally, CAM was employed to interpret the prediction of PreAIS. The predictive model PreAIS and the related dataset constructed in this study can be accessed on the following GitHub page: <span><span>https://github.com/xzfang00/PreAIS</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"119 ","pages":"Article 108612"},"PeriodicalIF":3.1,"publicationDate":"2025-07-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144739313","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Computationally guided design and synthesis of pyrimidine–oxazole hybrids as novel antidiabetic agents: kinetic and molecular interaction studies","authors":"Shoaib Khan ,&nbsp;Tayyiaba Iqbal ,&nbsp;Rafaqt Hussain ,&nbsp;Faez Falah Alshehri ,&nbsp;Zafer Saad Al Shehri ,&nbsp;Sobhi M. Gomha ,&nbsp;Magdi E.A. Zaki ,&nbsp;Hamdy Kashtoh","doi":"10.1016/j.compbiolchem.2025.108602","DOIUrl":"10.1016/j.compbiolchem.2025.108602","url":null,"abstract":"<div><div>Diabetes mellitus (DM) is one of the complex and chronic endocrine diseases often characterized by high blood glucose level. Diabetes is due to either pancreases not producing insulin which convert excess of blood glucose to glycogen, or the cell of body becoming unresponsive to insulin’s effect. Primary symptom of DM includes blurry vision, excess urination and slow healing sores but if not diagnosed earlier and treated, it is associated with some severe secondary impairment like cardiovascular diseases, diabetic neuropathy, diabetic nephropathy and Alzheimer’s diseases etc. The focus of current research work is to design and synthesized a novel pyrimidine based oxazole derivatives (1−10) having promising anti-diabetic activity. These derivatives were synthesized by using reagent grade starting material i.e. 4-chloro-6-methylpyrimidin-2-amine. Structural conformation of the synthesized derivative was acquired by ¹H-NMR and ¹³C-NMR and their molecular weight were confirmed by HREI-MS. These compound exhibit moderate to excellent biological potential against α-amylase and α-glucosidase in comparison to standard acarbose IC₅₀= 10.50 ± 0.20 μM and IC₅₀= 10.80 ± 0.10 μM. Among these derivatives, analog 8 having IC₅₀= 5.20 ± 0.10 μM against α-amylase and IC₅₀= 5.70 ± 0.10 μM against α-glucosidase emerged as a most potent compound of the series with excellent inhibitory potency of target enzyme. The biological interaction of the newly synthesized derivatives was studied through molecular docking to assess their enzyme inhibition potency. Furthermore, the molecular dynamic (MD) simulation, density functional theory (DFT) studies are also performed in order to assess their structural conformational changes, stability and reactivity under dynamic environment. Absorption distribution metabolism excretion and toxicity (ADMET) analysis showed that these potent compounds have no toxicological effect.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"119 ","pages":"Article 108602"},"PeriodicalIF":3.1,"publicationDate":"2025-07-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144724019","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advancing explainable AI in healthcare: Necessity, progress, and future directions","authors":"Rashmita Kumari Mohapatra ,&nbsp;Lochan Jolly ,&nbsp;Sarada Prasad Dakua","doi":"10.1016/j.compbiolchem.2025.108599","DOIUrl":"10.1016/j.compbiolchem.2025.108599","url":null,"abstract":"<div><div>Clinicians typically aim to understand the shape of the liver during treatment planning that could potentially minimize any harm to the surrounding healthy tissues and hepatic vessels, thus, constructing a precise geometric model of the liver becomes crucial. Over the years, various methods for liver image segmentation have emerged, with machine learning and computer vision techniques gaining rapid popularity due to their automation, suitability, and impressive results. Artificial Intelligence (AI) leverages systems and machines to emulate human intelligence, addressing real-world problems. Recent advancements in AI have resulted in widespread industrial adoption, showcasing machine learning systems with superhuman performance in numerous tasks. However, the inherent ambiguity in these systems has hindered their adoption in sensitive yet critical domains like healthcare, where their potential value is immense. This study focuses on the interpretability aspect of machine learning methods, presenting a literature review and taxonomy as a reference for both theorists and practitioners. The paper systematically reviews explainable AI (XAI) approaches from 2019 to 2023. The provided taxonomy aims to serve as a comprehensive overview of XAI method traits and aspects, catering to beginners, researchers, and practitioners. It is found that explainable modelling could potentially contribute to trustworthy AI subject to thorough validation, appropriate data quality, cross validation, and proper regulation.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"119 ","pages":"Article 108599"},"PeriodicalIF":3.1,"publicationDate":"2025-07-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144739286","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MJnet: A lightweight RNN-based model for microRNA target site prediction","authors":"Junhao Yu,&nbsp;Cong Hui,&nbsp;Jianhua Jia","doi":"10.1016/j.compbiolchem.2025.108603","DOIUrl":"10.1016/j.compbiolchem.2025.108603","url":null,"abstract":"<div><div>Accurate prediction of microRNA (miRNA) target sites is critical for understanding post-transcriptional gene regulation. While recent deep learning models have achieved high predictive accuracy, many suffer from excessive computational complexity and limited interpretability. In this study, we propose MJnet, a lightweight and efficient deep learning model based on a Bidirectional Gated Recurrent Unit (BiGRU) architecture, integrated with simple C2 encoding, a multi-scale one-dimensional convolutional network (TextCNN), and a self-attention mechanism. This framework captures both local sequence features and global contextual dependencies while maintaining low computational cost. Extensive experiments on experimentally validated datasets demonstrate that our model outperforms several traditional and deep learning-based baselines, including Mimosa, in terms of accuracy, F1-score, and robustness across balanced gene-level test sets. Ablation studies confirm the effectiveness of each module, and attention heatmaps reveal interpretable patterns aligned with known seed regions. Our approach offers a practical, reproducible, and interpretable solution for miRNA target site prediction in biologically relevant contexts.</div></div>","PeriodicalId":10616,"journal":{"name":"Computational Biology and Chemistry","volume":"119 ","pages":"Article 108603"},"PeriodicalIF":3.1,"publicationDate":"2025-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144721337","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}