Journal of Biomolecular Structure & Dynamics最新文献

{"title":"Understanding the change in calcium ion-binding, protein dynamics, and hydration behavior from the wild type for two pathogenic mutants (D3V and G34R) of <i>troponin-C</i>: molecular dynamics simulation and network analysis.","authors":"Pooja, Pradipta Bandyopadhyay","doi":"10.1080/07391102.2026.2736170","DOIUrl":"https://doi.org/10.1080/07391102.2026.2736170","url":null,"abstract":"<p><p>The protein Troponin-C is an important player in muscle contraction, and mutations in Cardiac Troponin-C (cTnC) have been associated with inherited cardiomyopathies. Till now, eight mutants of the N-domain of cTnC have been listed in the ClinVar database as pathogenic or likely pathogenic. Understanding the molecular mechanisms of these mutations is crucial for mitigating their effects. Previous work studied mutations A8V, A31S, and I4M and found that changes in the conformational landscape are a key feature of these mutations. In the current work, our focus is on two mutations, D3V and G34R, the only ones out of the eight mentioned that undergo a change of charge on mutation. We have performed molecular dynamics simulation and network analysis to calculate binding affinity between Ca²⁺ and mutants vis-à-vis the wild type, the change of Ca²⁺ coordination shell, the effect of water, protein dynamics, and the communication path from the mutation site to the binding site. Our findings show that both electrostatics and the location of the mutation are the key factors for the difference between WT and the two mutants. D3V, in a disordered region, disrupts long-range allostery through electrostatic and hydration changes, whereas G34R, near the binding site, induces local electrostatic and structural perturbations. The changed conformational landscape, as found for the previously studied mutants, is a general feature for these two mutations also; however, specific effects found here illustrate the diversity present in different mutants.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-18"},"PeriodicalIF":2.5,"publicationDate":"2026-10-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149224217","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Interaction of serum carrier protein with anticancer drugs, PCZ and PXN on binary and ternary systems through multi-spectroscopic, molecular modeling and quantum chemical approaches.","authors":"Samima Khatun, Riyazuddeen, Malik Raihan Ahmad, Rittik Bhati, Jayaraman Muthukumaran, Mohd Sajid Ali","doi":"10.1080/07391102.2026.2748694","DOIUrl":"https://doi.org/10.1080/07391102.2026.2748694","url":null,"abstract":"<p><p>We explored binding mechanism and conformational alteration of anticancer drugs, PCZ and PXN with HSA in binary and ternary combinations . The binding constant value for HSA-PCZ and HSA-PXN systems obtained from fluorescence quenching experiments are 2.7417 × 10⁴ L/mol and 1.603 × 10⁴ L/mol showed that the affinity of PCZ is more than PXN towards HSA. Far-UVCD spectroscopy study of HSA in presence of PCZ/PXN individually and in combination substantiated that they destabilized the secondary structure of the protein in the order of HSA-PXN > HSA-PXN-PCZ > HSA-PCZ . To see the conformational change in HSA due to drug binding, we have also performed UV-visible, synchronous and 3D fluorescence methods. All the methods proving the same fact that both PCZ and PXN alter the microenvironment of HSA near tryptophan and tyrosine amino acid residue due to complex formation. Experimental results further validated by computational techniques like docking, simulation and DFT. Molecular docking study shows the binding sites, nature of interactions which exactly supports our fluorescence result. Energy gap between Frontier Molecular Orbitals (FMO) of HSA-PCZ/PXN complex was found to be higher than free drug molecules which indicated that the greater molecular stability of complexes were formed. Additionally, we performed MD simulation to study the structural stability of HSA both unbound and bound to PCZ/PXN. This study is expected to be of paramount importance for getting insights of their transportation in human body and could be highly beneficial in further investigating their oncogenic properties.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-19"},"PeriodicalIF":2.5,"publicationDate":"2026-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149224227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Structure-dynamics analysis of insulin receptor modulation by plant-derived ligands suggests a potential regulatory role for Asn711.","authors":"Mahnaz Shojapour, Somayeh Farahmand, Robabeh Hajihosseinloo","doi":"10.1080/07391102.2026.2741934","DOIUrl":"https://doi.org/10.1080/07391102.2026.2741934","url":null,"abstract":"<p><p>The insulin receptor (IR) is a central regulator of glucose homeostasis and represents a promising yet underexplored target for antidiabetic drug development. In this study, an integrated computational workflow combining large-scale phytochemical screening, molecular docking, and 200 ns molecular dynamics (MD) simulations was employed to identify plant-derived ligands capable of stabilizing activation-relevant regions of IR. From an initial library of 140 phytochemicals, Magnoflorine and Nerolidol emerged as top candidates based on binding affinity, ADMET properties, and interaction persistence. Docking results revealed favorable engagement of both ligands within the IR ectodomain, followed by MD simulations that confirmed stable complex formation and reduced structural fluctuations relative to the apo receptor. Both ligands showed stable interactions with Asn711, indicating that this residue may play a previously underappreciated role in ligand recognition. Dynamic analyses further demonstrated ligand mobility between the two insulin-binding sites, supporting a dual-site engagement mechanism analogous to insulin. Overall, these findings provide new mechanistic insights into IR modulation and suggest Magnoflorine and Nerolidol as promising lead molecules for the development of plant-based therapeutics targeting insulin resistance.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-19"},"PeriodicalIF":2.5,"publicationDate":"2026-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149182123","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Coarse grained molecular dynamics analysis of menthol interactions with ergosterol in a fungal mimetic membrane.","authors":"Alaa Aldabet","doi":"10.1080/07391102.2026.2743847","DOIUrl":"https://doi.org/10.1080/07391102.2026.2743847","url":null,"abstract":"<p><p>Menthol, a monoterpene from peppermint oil, has antifungal activity against Candida species. However, how menthol interacts directly with fungal membrane sterols is still not well understood. In this study, coarse-grained molecular dynamics simulations were used with the Martini force field to investigate menthol in a fungal-mimetic POPC/ergosterol (8:2) membrane. Radial distribution function analysis revealed preferential spatial association between menthol and ergosterol, including a local maxima at approximately 5.0 Å and a broader peak in the 9.5 Å range. Contact analysis further demonstrated persistent transient interactions, with an average of 1.3 ± 0.4 menthol-ergosterol contacts per frame and at least one contact detected in approximately 94% of analyzed frames. Mean square displacement (MSD) analysis indicated laterally mobile menthol behavior within the membrane, with a diffusion coefficient of approximately 13.45 Ų/ps, (R² = 0.94). Density profile analysis revealed a broad membrane distribution of menthol with partial overlap with the ergosterol region. The membrane remained structurally stable throughout the simulation, showing an average sterol-region thickness of approximately 22.95 Å and an area per lipid of approximately 75.03 Ų. Membrane ordering analysis revealed a high orientational order parameter (SCD = 0.7983), indicating a well-organized liquid-ordered phase, with no evidence of pore formation or membrane disruption. Statistical robustness was confirmed <i>via</i> bootstrapping (95% CI: 0.6977-0.9347). The results indicate that menthol preferentially associates with ergosterol-containing regions while retaining considerable lateral mobility, engaging in frequent but transient interactions rather than forming stable complexes. This dynamic behavior provides a plausible molecular basis for the membrane-associated antifungal effects of menthol reported experimentally.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-12"},"PeriodicalIF":2.5,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149166564","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Computational analysis of pathogenic nsSNPs in the human TNF gene and their role in neuroinflammation in Adrenoleukodystrophy.","authors":"Sarita Maurya, Chakresh Kumar Jain","doi":"10.1080/07391102.2026.2743170","DOIUrl":"https://doi.org/10.1080/07391102.2026.2743170","url":null,"abstract":"<p><p>Adrenoleukodystrophy (ALD) is an X-linked neurodegenerative disorder caused by pathogenic variants in the ABCD1 gene and characterized by progressive demyelination and neuroinflammatory responses. Tumor necrosis factor-alpha (TNF-α) is an important mediator of inflammatory signaling and has been implicated in the neuroinflammatory environment associated with ALD. However, the structural and functional consequences of genetic variations within TNF remain insufficiently understood. The study focuses to find high-risk non-synonymous single nucleotide polymorphisms (nsSNPs) in the TNF gene and their impact on protein structure dynamics. A comprehensive computational methodology integrated with several pathogenicity prediction tools, such as SIFT, PolyPhen-2, MutPred2, MetaLR, CADD, REVEL, MutationAssessor, and PredictSNP along with I-Mutant 2.0 and DynaMut to re-examine protein stability, and ConSurf to assess evolutionary conservation have been employed. Selective variations on protein structure were subjected to 200 ns Molecular Dynamics (MD) simulations and analysed. The results identified I118F and I118N as variants with strong computational evidence of deleteriousness and predicted destabilizing effects at a conserved residue.MD simulations demonstrated that these mutations result in heightened structural fluctuations, as indicated by elevated RMSD and Radius of Gyration (Rg) values, signifying a reduction in structural compactness. The PCA and FEL analyses showed that the wild-type protein keeps a stable energy minimum, whereas DCCM analysis indicated that the I118F and I118N mutations exhibit weaker dynamic coupling with more than one metastable state, leading to lesser stable structure. These results suggest that the stability and coordinated dynamics of the TNF trimer are compromised by mutations at position 118. These findings suggest that the identified TNF variants may influence TNF protein stability and conformational dynamics; however, their relevance to ALD-associated neuroinflammation remains hypothetical and requires experimental and clinical validation.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-22"},"PeriodicalIF":2.5,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149164474","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Unravelling the synergy of curcumin and quercetin in the adipogenesis pathway: highlighting molecular dynamics and epigenetic reversibility.","authors":"Sunandini Ghosh, Indraneel Rakshit, Pritha Bhattacharjee","doi":"10.1080/07391102.2026.2741932","DOIUrl":"https://doi.org/10.1080/07391102.2026.2741932","url":null,"abstract":"<p><p>Obesity, the global pandemic, has been observed to be associated with metabolic diseases, including cardiovascular diseases, type 2 diabetes mellitus, polycystic ovarian syndrome, hypothyroidism, arthritis, and even cancer. Conventional drugs, popular in weight management, have been reported to have several adverse effects on cardiovascular and gastrointestinal health. Modern drug design, considering natural bioactive compounds with minimal side effects and ligand-receptor dynamic conformations, has become a necessity of the hour. In this context, the current study objective was to investigate the anti-obesity efficacy of two significant bioactive compounds, curcumin and quercetin, as potential therapeutics in obesity management. Molecular docking, followed by MD simulation, was the technique adopted in our study to identify the capability of curcumin and quercetin as potential anti-obesity drugs. GLP1R, PPARG, and LEPR were selected as the receptors. The structural stability of these proteins is required for their successful role in metabolic pathways. <i>In vitro</i> validation of the suitable doses of curcumin and quercetin was performed on 3T3L1 adipocytes. <i>In-silico</i> analyses, including root mean square deviation, radius of gyration, and solvent-accessible surface area, identified the specific ligand-dependent receptor activities. Quercetin was observed to enhance the compactness and stability of GLP1R and LEPR. In contrast, curcumin was found to maintain the conformation of PPARG. Treatment with a combination dose of curcumin and quercetin (15 µM + 20µM) on mature adipocytes for 72 hr reflected the most pronounced effects on the DNA methylation (4.83 ± 0.15%, <i>p</i> < 0.001), candidate gene expression (<i>ADIPOQ</i>: 0.00006 ± 0.00001-fold decrease; <i>PPARG</i>: 0.019 ± 0.0004-fold decrease; <i>GLP1R</i>: 0.096 ± 0.023-fold decrease; <i>p</i> < 0.001), and histone modifications. The prospects of this study include the consideration of a combination therapy of curcumin and quercetin for the most effective weight management outcome.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-25"},"PeriodicalIF":2.5,"publicationDate":"2026-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149127556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Structure-function studies of four disulfide-containing insulins from <i>Conus textile</i> and <i>Conus memiae</i> and a snail-guided human insulin analogue.","authors":"Shweta Dhannura, Kashibai Patil, Krishnappa Radhakrishna, Ganesh Karemunge, Prasanna D Revanasiddappa, Anil Kumar Pasupulati, Konkallu Hanumae Gowd","doi":"10.1080/07391102.2026.2741936","DOIUrl":"https://doi.org/10.1080/07391102.2026.2741936","url":null,"abstract":"<p><p>Cone snail insulins represent a unique class of insulin-like polypeptides that can regulate the human insulin receptor, offering valuable lead compounds for the treatment of diabetes and its complications. Among these, signalling insulins constitute the predominant class of snail insulins and are characterized by the presence of an additional fourth disulfide bond in the sequence. However, studies examining their affinity for the human insulin receptor remain limited. Database analysis revealed signalling insulins from <i>Conus textile</i> (_Scon-ins Tx1) and <i>Conus memiae</i> (_Scon-ins Me1) with chain lengths comparable to that of human insulin. In this study, computational approaches were employed to predict the 3D structures of these signalling insulins and evaluate their binding affinity toward the human insulin receptor (hIR). The 3D structures of _Scon-ins Tx1 and _Scon-ins Me1 were computed using AlphaFold 3 and modelling-simulation methods, yielding in distinct conformations for the same polypeptide. AlphaFold 3 models retained the canonical three-helix fold, whereas the modelled-simulated structures exhibited reduced helical content. Further, AlphaFold 3 models possess higher disulfide strain energy compared to the modelled-simulated structures. Binding affinity calculations showed that _Scon-ins Tx1 exhibits comparable affinity to human insulin at the site-1 pocket of hIR but lower affinity at the site-2 pocket. Furthermore, a newly designed snail-guided human insulin (_SHum-Ins-A4-B2-4SS) exhibited structural features and binding affinities comparable to those of _Scon-ins Tx1, further supporting the findings of this study. The results of current computational findings identify _Scon-ins Tx1 as a suitable candidate for future experimental validation and potential development as a diabetes therapeutic.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-19"},"PeriodicalIF":2.5,"publicationDate":"2026-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149127550","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Significance of BCR-ABL P-loop point mutations (L248R/V) on binding to FDA approved drugs: molecular dynamics simulations and DFT studies.","authors":"Y Indra Neela, Lalitha Guruprasad","doi":"10.1080/07391102.2026.2735399","DOIUrl":"https://doi.org/10.1080/07391102.2026.2735399","url":null,"abstract":"<p><p>The BCR-ABL fusion protein with dysregulated tyrosine kinase activity is the cause of chronic myelogenous leukemia (CML). Mutations in the P-loop, gatekeeper, catalytic and activation loop, within its kinase domain leads to drug resistance while treating CML. The mutation driven drug resistance mechanisms at the atomic level in BCR-ABL needs to be unravelled. This work aims to study the structural, dynamical, and functional effects of the P-loop point mutations on BCR-ABL kinase. In this work, the five main FDA approved drugs Imatinib, Dasatinib, Nilotinib, Ponatinib, Bosutinib and one investigational drug Rebastinib were analysed using molecular dynamics (MD), post-MD analyses and quantum chemical (QM) approaches. The P-loop residue L248 and its point mutations (L248R/V) forming direct interactions with the drug were validated. The root-mean-square deviation analysis suggests that Imatinib, Dasatinib and Nilotinib are promising drugs for L248 and L248R/V point mutations, followed by Rebastinib, Ponatinib, and Bosutinib, despite minor variations. The R and C-spines of the wild and mutant systems differ significantly, indicating that the protein backbone is directly affected by mutation. For all the considered wild and mutated amino acid-drug complexes, the interaction energy for L248R complexes is high compared to L248 and L248V complexes. The frontier molecular orbitals, HOMO-LUMO gap, and electrostatic potential maps were analysed. The MD simulations and QM studies suggest that L248 residue and its point mutations (L248R/V) are responsible for high affinity binding of BCR-ABL kinase with the drugs considered for the study.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-33"},"PeriodicalIF":2.5,"publicationDate":"2026-10-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149110377","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrating deep learning model for precision design of MERS antibodies with augmented affinity against SARS-CoV-2.","authors":"Areesha Irshad, Atta Ur-Rehman, Taskeen Koser, Mudasir Alvi, Hunza Usman, Rana Rehan Khalid","doi":"10.1080/07391102.2026.2741688","DOIUrl":"https://doi.org/10.1080/07391102.2026.2741688","url":null,"abstract":"<p><p>The rapid emergence of SARS-CoV-2 has renewed interest in cross-reactive antibody therapeutics targeting conserved epitopes across divergent coronaviruses. Building on our deep-learning driven protein engineering pipeline, we sought to redirect experimentally validated MERS-CoV antibodies toward the SARS-CoV-2 receptor-binding domain. A structural comparison identified a conserved Site 2 epitope capable of supporting cross-neutralizing interactions and three MERS-CoV antibodies showing measurable compatibility with this region were selected as templates. Using the customized ProteinMPNN based design framework, with multi-parameter biophysical filtering, we introduced context-specific substitutions to strengthen interfacial complementarity with the SARS-CoV-2 RBD. The study extends our earlier workflow by adding two evaluation parameters, antigen processing and T cell immunogenicity, to assess mutational impact on developability and immune compatibility. Top-ranked variants underwent molecular dynamics simulations and post-trajectory analyses, including stability profiling, hydrogen-bond and hydrophobic network evaluation, and per-residue binding energetics. Developability assessments, including aggregation propensity and immunogenicity profiling, were integrated to ensure therapeutic readiness. Across all scaffolds, several engineered variants, most notably D12(T721E), MERS27(D722N), and JC57-14(S618G), exhibited enhanced interface organization, favorable energy landscapes, and can improved structural stability relative to their parental antibodies. Many also showed reduced aggregation potential and lower predicted antigen presentation across MHC class I and II alleles, further refined by the added processing and immunogenicity metrics. Together, these findings suggest that MERS-CoV antibody scaffolds can be repurposed to target SARS-CoV-2 through a conserved epitope, highlighting a promising strategy for broad-spectrum coronavirus antibodies and providing strengthened leads for experimental development.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-28"},"PeriodicalIF":2.5,"publicationDate":"2026-10-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149110412","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DeepRAGIL-2: a retrieval-augmented protein language model framework for sensitive and accurate prediction of IL-2-inducing peptides.","authors":"Juan Peter Timothy Yuune, Van The Le, Yu-Yen Ou","doi":"10.1080/07391102.2026.2741122","DOIUrl":"https://doi.org/10.1080/07391102.2026.2741122","url":null,"abstract":"<p><p>Interleukin-2 (IL-2) is a pleiotropic cytokine central to T-cell activation, proliferation, and immunological tolerance, yet existing computational tools for predicting IL-2-inducing peptides suffer from near-zero sensitivity under realistic class-imbalanced conditions, rendering them impractical for positive-class discovery. Here we present DeepRAGIL-2, a sequence-based framework integrating ESM-2 protein language model embeddings, dual-window convolutional feature extraction, and retrieval-augmented embedding fusion (RAG) into a unified classification pipeline. Trained on 3,825 redundancy-filtered experimentally validated sequences from IEDB and evaluated on an independent test set under a 1:4 class imbalance, DeepRAGIL-2 achieves a sensitivity of 0.7501, specificity of 0.9708, MCC of 0.7605, and AUC of 0.8654, outperforming conventional machine learning classifiers and published IL-2 prediction tools on sensitivity and MCC. Across ten independent runs, performance was stable at MCC of 0.7555 ± 0.0173, confirming that the reported result is representative rather than an artefact of a single weight initialisation. Ablation experiments identify RAG integration as the single most consequential architectural component, with MCC rising from 0.3453 to 0.7605 upon its inclusion, demonstrating that retrieval of experimentally validated reference sequences at inference time substantially compensates for the limited positive-class training data available for this task. DeepRAGIL-2 represents the first IL-2 prediction framework with demonstrated sensitivity for positive-class recovery under realistic conditions, establishing a framework applicable in principle to cytokine-specific immunological sequence classification.</p>","PeriodicalId":15272,"journal":{"name":"Journal of Biomolecular Structure & Dynamics","volume":" ","pages":"1-16"},"PeriodicalIF":2.5,"publicationDate":"2026-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149110381","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}