Molecular Diversity最新文献

{"title":"Synthesis and biological evaluation of Phorbazole D analogues identify IO-5a as a promoter of hUC-MSC chondrogenic differentiation.","authors":"Ai-Zhu Wang, Xiao-Mei Liu, Li-Ming Zhang, Lu-Lu Fu, Ping Li, Jie-Ping Zhang, Hua Han","doi":"10.1007/s11030-026-11683-3","DOIUrl":"10.1007/s11030-026-11683-3","url":null,"abstract":"<p><p>Mesenchymal stem cells (MSCs) are a promising cell source for osteoarthritis and intervertebral disc degeneration, but their application is limited by inefficient and unstable chondrogenic differentiation. Here, nine Phorbazole D analogues were synthesized and evaluated for promoting chondrogenesis in human umbilical cord mesenchymal stem cells (hUC-MSCs) under TGF-β/dexamethasone based induction conditions. Among the tested analogues, IO-5a was selected for further evaluation based on its comparatively higher pro-chondrogenic activity and no detectable cytotoxicity. Image-based Alcian blue analysis showed that the positive staining area increased from approximately 17% in the control group to approximately 45% after IO-5a treatment. RT-qPCR further showed that 10 µM IO-5a upregulated COL2A1, SOX9, ACAN, and COMP by approximately 2.6-fold, 1.1-fold, 1.35-fold, and 1.4-fold, respectively. Immunofluorescence, western blotting, and 3D pellet culture analyses supported increased cartilage associated marker expression and cartilage like matrix deposition. Preliminary structure-activity relationship analysis suggested that retention of the A-ring phenolic hydroxyl group and favorable A-/C-ring substituent matching may contribute to activity. In silico target prediction and molecular docking nominated PTPN1 (PTP1B) as a potential candidate protein, although this predicted interaction requires further experimental validation. Collectively, IO-5a represents a bioactive Phorbazole D analogue with chondrogenesis promoting activity in hUC-MSCs.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":"7551-7568"},"PeriodicalIF":4.3,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148617877","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Unraveling the molecular landscape and therapeutic strategies for acute kidney injury: insights from transcriptomics, network pharmacology, virtual screening, and in vitro experiments.","authors":"Guoqiang Li, Dianjie Zeng, Yinhuai Wang, Jiachen Liu, Dong Yang","doi":"10.1007/s11030-026-11528-z","DOIUrl":"10.1007/s11030-026-11528-z","url":null,"abstract":"<p><p>Acute kidney injury (AKI) is a clinical syndrome characterized by a rapid decline in renal function, high morbidity and mortality, and a lack of effective early diagnostic markers or targeted therapies. To address this critical unmet need, we employed an integrated multi-omics and network pharmacology approach to systematically investigate the molecular mechanisms and potential therapeutic targets of AKI. Core targets were identified through differential gene expression (DEG) analysis combined with weighted gene co-expression network analysis (WGCNA), followed by exploration using protein-protein interaction (PPI) networks and pathway enrichment analyses. Inflammation, oxidative stress, and energy metabolism emerged as key pathways involved in AKI pathogenesis. Using ten CytoHubba algorithms and the MCODE module for comprehensive screening, we identified three hub genes-ACO2, FBP1, and PFKL. Their expression patterns and cellular specificity were further characterized using single-cell RNA sequencing data from AKI renal tissues. Additionally, we constructed a miRNA-hub gene regulatory network, providing insights into miRNA-based therapeutic strategies. Molecular docking analysis identified three approved drugs-Ajmaline, Cimetidine, and Tretinoin-with strong binding affinities to the hub proteins, suggesting their potential for repurposing in AKI treatment. Finally, by reviewing knockout mouse models from the Mouse Genome Informatics (MGI) database and conducting in vitro cell experiments, we explored the in vivo and in vitro roles of these core targets, providing experimental evidence of their physiological relevance. Overall, this study integrates cross-cohort transcriptomic profiling, network-based hub prioritization, single-nucleus cell-type localization, translational drug repurposing analyses, and in vitro experimental validation thereby providing a multi-layered framework to prioritize candidate biomarkers for AKI.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":"8313-8334"},"PeriodicalIF":4.3,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13635034/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147939837","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluation of toll-like receptor inhibition by novel thymoquinone analogues as potential remedy for diabetic nephropathy.","authors":"Nashwa H Zaher, Reham M M El-Hazek, Mohamed I Mostafa","doi":"10.1007/s11030-026-11730-z","DOIUrl":"https://doi.org/10.1007/s11030-026-11730-z","url":null,"abstract":"<p><p>Diabetic nephropathy (DN) is a major worldwide cause of chronic kidney disease and end-stage renal failure, impacting approximately 20-50% of individuals with diabetes. DN frequently progresses to end-stage renal disease. The prevalence of DN continues to increase significantly, accompanied by rising rates of death and cardiovascular complications. To date, there is no specific drug approved exclusively for the treatment of diabetic nephropathy. In this work 13 novel disubstituted TQ analogues were synthesized, recrystallized and structurally verified via microanalytical and spectral data. Screening against TLR-4 activity was performed, the most potent disubstituted propyl ether TQ13 was subjected to more in-vivo biological investigations in STZ induced hyperglycemia rats. TQ13 showed anti-hyperglycemic, antioxidant, anti-inflammatory and antiapoptotic effects in addition to amelioration of kidney function markers, TLR-2 & TLR-4 activities. The findings of this investigational study suggest that disubstituted propyl ether TQ13 may represent a promising radio-stable renoprotective candidate in DN therapeutics which needs more tentative studies on toxicity and exact mechanism of action involving advanced staging pathways.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":""},"PeriodicalIF":4.3,"publicationDate":"2026-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149056846","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"In silico directed evolution of humanized peptide transporters via computational epistatic rescue.","authors":"Alper Karagöl, Taner Karagöl","doi":"10.1007/s11030-026-11710-3","DOIUrl":"https://doi.org/10.1007/s11030-026-11710-3","url":null,"abstract":"<p><p>The human intestinal peptide transporter 1 (SLC15A1, PepT1) is a critical determinant of oral drug bioavailability, yet its thermodynamic characterization and structural tractability remain challenging. While bacterial orthologues like the Escherichia coli DtpA transporter offer highly stable structural surrogates, significant cross-species variations in the binding microenvironment limit their pharmacological fidelity. Traditional structure-based engineering aimed at engrafting the human pharmacophore onto bacterial scaffolds is frequently hindered by combinatorial explosion and severe thermodynamic frustration, as non-native side-chains disrupt co-evolved local packing. To bypass the empirical bottlenecks of in vitro directed evolution, we present a fully in silico co-evolutionary pipeline leveraging deep contextual protein language models (PLMs). By mapping 19 pharmacologically critical human PepT1 residues onto the DtpA sequence, we utilized the 650-million parameter ESM-2 model to conduct zero-shot mutational profiling. We identified 14 primary humanizing mutations to which the model assigned very low likelihoods (below 10 - 5). To resolve these energetic conflicts, we deployed an automated epistatic rescue algorithm to sweep the flanking microenvironmental topologies. This computationally efficient heuristic successfully identified localized, non-native compensatory mutations for all high-risk targets without relying on traditional molecular dynamics. Most notably, the highly deleterious Q41R and R305F substitutions were buffered by Y38D and I304K secondary mutations, yielding 5,215-fold and 11,614-fold increases in the model likelihood assigned to the primary substitution, respectively; these are changes in sequence likelihood and not calculated free energies. Binding dynamics were assessed via docking calculations and all-atom membranous molecular dynamics simulations. This deep learning-driven framework computationally rationalizes the engraftment of the human binding pocket, reducing the potential experimental screening space and yielding a humanized surrogate that requires experimental validation of folding, binding and transport before use in pharmacological screening.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":""},"PeriodicalIF":4.3,"publicationDate":"2026-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149056840","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Discovery of a new SARS-CoV-2 Mpro inhibitor from a natural product and natural product-like compound database using in silico and in vitro studies.","authors":"Sudesna Das, Uttam Pal, Yasmin Begum, Akshay Joshi, Nittu Singh, Krishan Gopal Thakur, Umesh Prasad Singh","doi":"10.1007/s11030-026-11724-x","DOIUrl":"https://doi.org/10.1007/s11030-026-11724-x","url":null,"abstract":"<p><p>This current study employs structure-guided virtual screening to identify potential inhibitors from 69,075 natural product or natural product-like compounds in the InterBioscreen (IBS) database against the SARS-CoV-2 Main protease (Mpro), a well-validated target. From high-throughput structure-guided virtual screening, subsequently followed by more rigorous re-docking, top seven compounds were chosen as hits based on their high binding affinity and specific H-bond interactions with the active-site residues of Mpro. All these compounds were new for this target as well as the virus. The stability of the seven protein-ligand complexes was further assessed using molecular dynamics (MD) simulations and the MM-GBSA method. The ADME/Tox profile of these seven compounds was also predicted theoretically. After MD simulation, only three compounds, viz. STOCK1N-86169, STOCK1N-88750, and STOCK1N-65657 were identified as high-potential candidates for Mpro inhibition. To confirm the in silico analysis results, an in vitro FRET-based Mpro inhibition assay and an ELISA-based SARS-CoV-2 inhibition assay were performed. From these in vitro investigations, STOCK1N-86169 was identified as the only potent Mpro inhibitor, exhibiting an IC₅₀ of 3.8 µM. Independent confirmation of STOCK1N-86169 binding to Mpro was obtained by differential scanning calorimetry (DSC), which revealed an increase of approximately 15 °C in the melting temperature (Tm) of the Mpro-STOCK1N-86169 complex relative to the apo-protein. In the cytotoxicity assay, it showed CC₅₀ > 100 µM for HeLa, ~ 100 µM for the Huh-7 cell line, and ~ 33.5 µM for the Vero E6 cell line. In the Vero cell-based SARS-CoV-2 infection assay, it showed an EC₅₀ of ~ 7 µM, with a reasonable Selective Index (S.I.) of ~ 5-fold. Our study suggests that the natural product like compound STOCK1N-86169, which presents a new scaffold for Mpro inhibition and also SARS-CoV-2, may be a promising candidate for further development as an effective inhibitor of SARS-CoV-2 Mpro.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":""},"PeriodicalIF":4.3,"publicationDate":"2026-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149029408","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Discovery of potent and selective MAO-B inhibitors from a morpholine-substituted chalcone library for neurodegenerative disorders.","authors":"Asmar Taghiyeva, Jaeyeop Lim, Abanish Biswas, Subodh Mondal, Ritika Uppal Mukherjee, Vinay N Basavanakatti, Nilüfer Bayrak, Venkatesan Jayaprakash, Man-Jeong Paik, Hoon Kim, Amaç Fatih TuYuN","doi":"10.1007/s11030-026-11733-w","DOIUrl":"https://doi.org/10.1007/s11030-026-11733-w","url":null,"abstract":"<p><p>A series of targeted chalcones (TCs) was designed and synthesised using a microwave-assisted synthetic method to introduce diverse substituents at different positions on the phenyl ring. The structures of all nineteen TC analogues were confirmed by NMR, FTIR, and HRMS. In vitro tests against acetyl- and butyrylcholinesterases (AChE and BuChE), β-Site Amyloid Precursor Protein Cleaving Enzyme 1 (BACE1), and monoamine oxidases (MAO-A and MAO-B) showed that MAO-B was the most potent and selective. Among them, TC12 was the strongest inhibitor (IC₅₀ = 11.00 ± 1.0 nM; SI > 3636) and worked competitively (K_i = 7.16 ± 1.20 nM). In terms of IC₅₀ = 50.0 ± 1.0 nM and SI > 800, TC9 was ranked second. TC12 showed no time-dependent inhibition of MAO-B, and its MAO-B inhibition was reversible. Molecular docking studies revealed that TC12 establishes favourable interactions within the MAO-B active site, particularly through π-π stacking with aromatic cage residues (TYR398 and TYR435), while showing limited engagement with key regions of MAO-A. ADME predictions indicated that TC12 exhibits favourable drug-like properties, including compliance with Lipinski's rules, moderate CNS permeability, and a pharmacokinetic profile supportive of oral bioavailability. In vitro Parallel Artificial Membrane Permeability Assay (PAMPA) and microsomal stability studies revealed that TC9 exhibited high blood-brain barrier (BBB) permeability comparable to carbamazepine, while TC12 showed moderate permeability but was consistent with its predicted CNS penetration. Both TC analogues demonstrated favourable metabolic stability profiles in rat and human liver microsomes, supporting their potential for further pharmacological evaluation. Overall, TC12 represents a promising candidate for selective MAO-B inhibition.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":""},"PeriodicalIF":4.3,"publicationDate":"2026-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149029453","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploring the molecular mechanism of benzo[a]pyrene affecting prostate cancer based on network toxicology and external validation.","authors":"Siqi Zhu, Zhuang Li, Kehua Jiang, Fa Sun, Jianguo Zhu","doi":"10.1007/s11030-026-11729-6","DOIUrl":"https://doi.org/10.1007/s11030-026-11729-6","url":null,"abstract":"<p><p>Benzo[a]pyrene (BaP) is a polycyclic fragrant hydrocarbon contaminant commonly establish throughout the surroundings. The International Agency for Research on Cancer has classified it as a Group 1 carcinogen; however, its exact contribution to the onset of prostate cancer (PCa) is still not well defined. This study systematically explores the mechanism underlying the association between BaP exposure and prostate cancer using approaches including network toxicology, machine learning, transcriptomic validation, immune infiltration assessment, single-cell analysis, molecular docking and external validation. The results showed that BaP and prostate cancer shared 975 overlapping targets, which were predominantly enriched in signaling pathways such as PI3K-Akt. Four core genes were screened out through differential analysis and machine learning, namely CAV1, TWIST1, PRKCA, and GDF15. Transcriptome verification showed that TWIST1 and GDF15 were up-regulated, CAV1 and PRKCA were down-regulated, and the AUC of the four-gene combined diagnosis reached 0.990. Core genes are associated with cellular growth and immune cell recruitment, and single-cell sequencing verified their specific cellular distribution and immunological shifts. Molecular docking showed that BaP binds well to the core target. External validation confirmed that BaP may promotes the progression of PCa, with upregulated expression of GDF15 and downregulated expression of PRKCA in PCa. In conclusion, BaP may promote the development of PCa by regulating pathways such as CAV1, TWIST1, GDF15 and PRKCA.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":""},"PeriodicalIF":4.3,"publicationDate":"2026-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149029534","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Structure-based drug repurposing and in vitro evaluation of VAV2-associated candidates for suppression of macrophage foam-cell formation in atherosclerosis.","authors":"Prerna Bansal, Kamna Srivastava","doi":"10.1007/s11030-026-11713-0","DOIUrl":"https://doi.org/10.1007/s11030-026-11713-0","url":null,"abstract":"<p><p>Cardiovascular disease remains a leading cause of mortality worldwide, and macrophage-derived foam-cell formation represents an important event in atherosclerosis. VAV2, a Rho-family guanine nucleotide-exchange factor, participates in signalling pathways associated with oxidized low-density lipoprotein uptake and macrophage lipid accumulation. In this study, a structure-based drug-repurposing workflow was used to identify clinically relevant compounds with predicted affinity for a full-length VAV2 structural model. Virtual screening, molecular docking, ADMET prediction, three independent 200-ns molecular-dynamics simulations, principal-component analysis, and MM-PBSA calculations prioritized talazoparib, tivozanib, and isradipine for further evaluation. In THP-1-derived macrophages, talazoparib reduced oxLDL-induced intracellular neutral-lipid accumulation at a sub-cytotoxic concentration. In a complementary DiI-oxLDL assay, all three compounds reduced cell-associated DiI-oxLDL fluorescence relative to vehicle control. These findings provide preliminary computational and cellular support for further investigation of the selected compounds. However, direct binding to or inhibition of VAV2 has not been demonstrated and requires biochemical, genetic, pharmacological, and in vivo validation.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":""},"PeriodicalIF":4.3,"publicationDate":"2026-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149007862","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial intelligence-guided discovery of fungal ergosterol derivatives as selective LXRβ agonists targeting the cholesterol dependency of glioblastoma.","authors":"Qi Li, Chunxue Zhang, Tiantian Hu, Zhenzhen Zhang, Haigang Wu","doi":"10.1007/s11030-026-11728-7","DOIUrl":"https://doi.org/10.1007/s11030-026-11728-7","url":null,"abstract":"<p><p>Glioblastoma (GBM) is the most lethal primary brain malignancy, for which no new systemic therapy has achieved regulatory approval in over past decade. A molecular vulnerability of GBM is its comprehensive co-dependency on exogenous cholesterol to sustain oncogenic signaling and membrane biogenesis, rendering the liver X receptor beta (LXRβ)-the principal transcriptional regulator of cholesterol efflux in the central nervous system-an attractive therapeutic target. However, the development of LXRβ-selective, blood-brain barrier (BBB)-penetrant agonists that avoid the hepatotoxicity associated with LXRα co-activation has remained an unresolved challenge. Here, we developed an integrated artificial intelligence-based virtual screening pipeline. The pipeline integrates ML-QSAR modelling, D-MPNN, and deep learning-based DTI prediction. It was applied to systematically interrogate a curated library of ~ 1.2 million natural products. Sequential filtering by predicted LXRβ affinity, LXRβ/LXRα selectivity, and integrated pharmacokinetic scoring, refined by molecular docking and MM-GBSA binding free energy calculations, identified four high-confidence candidates. Experimental validation by CCK-8 cytotoxicity assays across five GBM cell lines and six normal cell models revealed that 5,6-Epoxyergosterol is a potent anti-glioma agent. Furthermore, a 200 ns molecular dynamics simulation of the LXRβ-5,6-Epoxyergosterol complex demonstrated stable binding, with dominant hydrogen-bonding contacts at His435 and Trp443 and a free energy landscape consistent with a single dominant agonist-bound conformation. These findings establish fungal ergosterol derivatives as a novel class of LXRβ-targeted therapeutics and provide a validated AI-driven framework for accelerating natural product drug discovery in neuro-oncology.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":""},"PeriodicalIF":4.3,"publicationDate":"2026-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148954613","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"HD-SKRG: a hierarchical dual-view and structure-knowledge relation graph enhancement network for few-shot molecular property prediction.","authors":"Luyi Jia, Mingyang Wang, Zeming Wang, Xianjie Wang","doi":"10.1007/s11030-026-11719-8","DOIUrl":"https://doi.org/10.1007/s11030-026-11719-8","url":null,"abstract":"<p><p>Few-shot molecular property prediction aims to adapt to new property tasks using only a few labeled molecules. However, existing methods are still limited by insufficient molecular representations and biased intra-task relation construction under scarce supervision. To address these limitations, this paper proposes HD-SKRG, a hierarchical dual-view and structure-knowledge relation graph enhancement network. HD-SKRG improves few-shot prediction from two complementary aspects: molecular representation learning and intra-task relation propagation. For molecular representation, HD-SKRG builds atom-level and functional-group-level graphs, injects elemental knowledge into atom representations, and transfers local atomic information to functional-group representations. A frequency-aware aggregation module further produces molecular-level knowledge representations. For intra-task relation propagation, the structure relation graph provides the main feature-propagation pathway, whereas the knowledge relation graph supplies semantically related neighbors and refines relation weights. This design reduces graph-construction bias caused by relying on structural similarity alone. The dual-view encoders are pretrained by cross-view contrastive learning, and the full model is meta-trained under the MAML framework. Experiments on Tox21, SIDER, MUV, and ToxCast under 1-shot and 10-shot settings show that, HD-SKRG achieves the best results in five of eight settings and the second-best results in the remaining three. Ablation studies further confirm the contribution of each module.</p>","PeriodicalId":708,"journal":{"name":"Molecular Diversity","volume":" ","pages":""},"PeriodicalIF":4.3,"publicationDate":"2026-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148954606","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}