Browse Articles

3Br-MGD: few-shot toxicity prediction with a three-branch deep encoder and meta-learning framework

Abstract Predicting the toxicity of pharmaceutical compounds remains a major challenge in drug discovery. Early and accurate toxicity assessment is essential for eliminating harmful candidates before costly preclinical and clinical testing, thereby improving patient safety, reducing development costs, and accelerating the drug development process. Despite advances in computational toxicology, existing methods often struggle to capture complex molecular characteristics and maintain robust performance under limited-data conditions. To address these challenges, we propose 3Br-MGD, a novel three-branch framework that integrates deep learning and meta-learning for molecular toxicity prediction. The architecture combines complementary molecular representations: FingerprintMLP encodes Morgan fingerprint descriptors, Graph Convolutional Networks (GCNs) capture structural information from molecular graphs, and one-dimensional Deep Convolutional Neural Networks (1D-CNNs) extract sequential features from SMILES strings. These embeddings are integrated within a Prototypical Network-based few-shot learning framework, enabling rapid adaptation to new prediction tasks with limited labeled samples and improving generalization in low-resource settings. Experimental results on benchmark toxicity datasets demonstrate that 3Br-MGD consistently outperforms conventional baselines in predictive accuracy, robustness, and generalization. Furthermore, the integration of heterogeneous molecular encoders reduces dependence on large training datasets while enhancing interpretability through the exploitation of complementary chemical information from multiple molecular views.

Cascade Oxidation of Ethylene and Propylene over a Redox Heterometallic Cluster

Investigation in the selective targeting of homologous recombination-deficient cells by benzophenanthridine alkaloid nitidine

Altermagnetic Metal–Organic Frameworks

Sampling strategies for enhancing the analytical performance of a hybrid spectroscopy platform in biomedical applications

Abstract Laser-Induced Breakdown Spectroscopy (LIBS) of liquids poses significant analytical challenges, especially for biological samples such as blood and saliva, where only limited volumes are available. Maximising spectral information from these scarce samples is therefore essential. Among various liquid sampling approaches, the drop-coating method has shown particular promise for efficient sample utilisation. This study focuses on advancing the drop-coating approach by integrating enhanced sampling and laser excitation strategies to improve spectral performance while preserving LIBS’s intrinsic advantages. Specifically, pulsed laser-based Surface-Enhanced LIBS (SELIBS) and Nanoparticle-Enhanced LIBS (NELIBS) techniques were employed to amplify signal intensity and detection sensitivity. Systematic optimisation of key experimental parameters revealed effective conditions for achieving reproducible and high-intensity spectra. The proposed methodology provides a practical approach to enhance LIBS performance for the trace-level analysis of limited-volume biological samples, laying the groundwork for sensitive, non-destructive diagnostics and forensic applications. Furthermore, this work implements a hybrid spectroscopic platform designed for comprehensive elemental and molecular diagnosis of biological samples. This integrated architecture utilises a dual-laser, single-spectrograph configuration that enables coordinated excitation and collection of LIBS and Raman signals. This configuration simplifies the experimental workflow, reduces alignment and calibration, eases the analysis process and enhances the depth of information that can be extracted from biological samples, yielding a more holistic, comprehensive chemical composition of limited-volume bio samples.

Quantification of Binding of Small Molecules to Native Kinases by Flow Cytometry Reveals Divergence from Biochemical Affinities

Pregnancy outcomes following maternal GLP-1 receptor agonist exposure: a systematic review and meta-analysis

Dimensional Evolution from a Giant Molybdenum-Red Cage-like {Mo ₂₀₀ } to 1D Chains Enabling Ultrahigh Proton Conduction

Oral health knowledge, attitude, and practices of parents of autistic children

Adaptive Protein Corona Nanoassemblies Couple Cytokine Signaling with Endogenous Antigen Transport for Systemic Cancer Immunity

Mindfulness and artificial intelligence self-efficacy among university students: the mediating role of artificial intelligence literacy

Exploring the Structure and Chemistry of One-Dimensional and Two-Dimensional Lepidocrocite TiO ₂ at Atomic Resolution

Discovery of indolylchalcone benzenesulfonamides as selective inhibitors of tumor-associated carbonic anhydrase IX and XII

Abstract Indole, characterized by its favorable bioavailability, unique structural attributes, and a broad pharmacological profile, represents a privileged scaffold in anticancer drug discovery. Chalcones constitute another important scaffold in medicinal chemistry and numerous chalcone derivatives have been reported to exhibit diverse biological activities, including anticancer and enzyme inhibitory effects. Several chalcone-based compounds have also been investigated as carbonic anhydrase inhibitors. Sulfonamides are a classical class of carbonic anhydrase (CA, EC 4.2.1.1) inhibitors with diverse therapeutic relevance, being employed as diuretics, anticonvulsants, topical antiglaucoma agents and in the management of obesity and cancer. Herein, we designed and synthesized a novel series of indolylchalcone–benzenesulfonamide hybrids ( 15a − m and 16a − d ) and evaluated their inhibitory activities against a panel of four human carbonic anhydrases (hCA isoforms I, II, IX and XII). Interestingly, most of the tested compounds inhibited the tumor-associated hCA IX and XII isoforms with single- to double-digit nanomolar inhibition constants ( K i s). In particular, compound 15h (( E )-4-(3-(5-cyano-1 H -indol-3-yl)acryloyl)benzenesulfonamide) showed potent inhibition of hCA IX ( K i = 8.9 nM) with marked selectivity over hCA I, II, and XII (SI = 61.9, 7.2, and 5.7, respectively), whereas compound 15j (( E )-4-(3-(6-bromo-1 H -indol-3-yl)acryloyl)benzenesulfonamide) exhibited strong inhibition of hCA XII ( K i = 4.9 nM) with pronounced selectivity over hCA I, II, and IX (SI = 628.4, 12.8, and 11.9, respectively), compared with the reference inhibitor acetazolamide (AAZ). Structure–activity relationship analysis revealed that electron withdrawing groups, particularly cyano and bromo substituents on the indole scaffold, conferred enhanced selectivity toward tumor-associated carbonic anhydrase isoforms. Molecular docking demonstrated that both compounds adopted the canonical sulfonamide-binding mode, coordinating with Zn²⁺ and engaging in stable hydrogen-bond interactions with key active-site residues. Complementary molecular dynamics simulations confirmed the persistence of these interactions, as reflected by lower RMSD values and the maintenance of critical contacts throughout the simulation period. In addition, in silico ADME prediction using the SwissADME server suggested that compound 15j possesses a more favorable balance of lipophilicity and polarity, indicating improved membrane permeability and oral absorption potential compared to 15h and acetazolamide (AAZ), whereas 15h exhibited relatively higher polarity that may limit permeability despite comparable lipophilicity. Accordingly, these findings suggest that the novel indolylchalcone-benzenesulfonamide hybrids 15h and 15j represent promising leads for the development of potent and selective inhibitors of tumor-associated carbonic anhydrases.

Synergy of Dual-Atomic Sites and Interfacial Dynamics for 83.7% Selective Hydroxylamine Electrosynthesis

Comparative evaluation of domain-specific and general-purpose transformer models for Arabic poet classification

Abstract Arabic poet classification presents distinctive challenges stemming from the morphological richness and stylistic diversity inherent in both classical and modern Arabic verse. This study conducts an extensive comparative evaluation of several neural language models to assess their ability to represent poetic expression and capture authorial characteristics. Two carefully curated datasets are utilised: FrequentPoets , representing prolific authors with extensive verse collections, and CrossEraPoets , encompassing poets from distinct historical periods to examine temporal stylistic variation. A comparative evaluation framework is introduced to contrast domain-specific and general-purpose language models across prolific authorship and cross-era stylistic variation. The domain-adapted AraPoemBERT consistently achieves superior performance, attaining 73.11% accuracy (73.00% F1) on FrequentPoets and 77.06% accuracy (77.04% F1) on CrossEraPoets , whereas the general-purpose GPT-4o demonstrates considerably lower performance under zero-shot and few-shot evaluation settings. The results highlight the significance of domain-adapted pretraining for morphologically complex languages like Arabic and suggest the potential advantage of transformer-based architectures in modelling stylistic and linguistic nuances unique to Arabic verse. These findings also suggest that temporal diversity may play an important role in model generalisation across different poetic styles. The study contributes to Arabic Natural Language Processing (NLP) and digital humanities by enabling computational authorship attribution, stylistic analysis, and cross-historical exploration of Arabic literary heritage. Overall, the proposed framework provides a robust foundation for future research in Arabic poetry analytics and domain-specific language modelling.

Free-Anionic Duality of Interstitial Electrons Drives Superconductivity in Electride La ₃ In

Retraction Note: An enzymatic BioBattery based on multicopper oxidases with reduced substrate diffusion constraints for sustainable energy harvesting

Photocatalytic Hydrogen Cycling via Benzotrithiophene-Based Mesoporous Covalent Organic Frameworks for Efficient Fine Chemical Coproduction

An optimized ensemble learning model for interpretable and efficient obesity classification

Discovery and Biosynthesis of Nitrilobacillins by Post-Translational Introduction of C-Terminal Nitrile Groups