Search Results (3,770)

Drug–target affinity (DTA) prediction is a critical step in drug discovery and precision medicine. Although graph neural networks (GNNs) have achieved remarkable progress, existing graph pooling methods rely on fixed ratios, failing to adapt to the structural diversity of molecules and proteins, which leads to information loss or redundant feature retention. To address this issue, we propose the Adaptive Self-Attention Graph Pooling (ASAGPooling) mechanism, which introduces a learnable pooling ratio that dynamically adjusts node retention during training. Furthermore, we develop ASAG-DTA, a multi-modal framework that integrates GNNs with Transformers to jointly model molecular graphs, protein contact maps, SMILES sequences, and FASTA sequences. While ASAGPooling achieves competitive prediction accuracy (MSE = 0.186 on Davis), we acknowledge that it does not surpass the state-of-the-art DynHeter-DTA (MSE = 0.130), which incorporates a more complex dynamic heterogeneous graph architecture. Instead, the key contribution of ASAGPooling lies in its adaptability, interpretability, and computational efficiency. It can eliminate the need for manually tuned pooling ratios, enable direct visualization of retained key residues/atoms, and reduce model complexity. This makes ASAG-DTA a practical lightweight alternative for large-scale virtual screening scenarios where computational resources are constrained. Full article

Accurate short-term power forecasting is essential for enhancing the efficiency and reliability of energy systems. Nonetheless, conventional techniques for forecasting struggle to detect nonlinear patterns in power time series, as maintaining both stability and accuracy in predictions is tough. This research presents a unique prediction framework that integrates a Multilayer Perceptron (MLP) with survivor-based evolutionary selection strategies. The proposed neural network architecture comprises three hidden layers containing 32, 16, and 8 neurons, respectively. This enables the network to extract features while preserving essential information progressively. A Survivor selection process is employed to enhance the model’s efficacy. This approach retains the optimal training models for subsequent training phases. This technique enhances both predictive accuracy and training efficiency. The amalgamation of Survivor-based selection methodologies with MLP architectures for short-term power generation forecasting is overlooked in the existing literature, although it holds promise. Thus, the proposed model is evaluated against established baselines, including Linear Regression (LR), Random Forest (RF), and Support Vector Regression (SVR). The results from 30 distinct trials indicate that the proposed MLP (32-16-8) combined with the Survivor approach exhibits the minimal prediction errors, with a mean absolute error (MAE) of 5.3588 and a root mean square error (RMSE) of 10.0216. This strategy is superior in minimizing errors compared to alternative methods. Furthermore, statistical analyses utilizing the Wilcoxon signed-rank test and paired t-test indicate that the proposed method significantly outperforms SVR and RF, while displaying performance comparable to LR. The findings indicate that including a Survivor-based selection mechanism in the MLP training process is an effective and reliable method for forecasting short-term generation power. Full article

Background: Metal-based anticancer drugs, particularly platinum and gold complexes, play a central role in chemotherapy but are often limited by systemic toxicity, resistance, and suboptimal selectivity. Peptide conjugation has emerged as a versatile strategy to modulate the pharmacokinetic and biological properties of metal complexes, enabling targeted delivery, improved uptake, and controlled activation. This review aims to critically analyze platinum- and gold-peptide bioconjugates in cancer therapy, focusing on directly reactive metal complexes and redox-activated prodrug systems. Methods: Relevant literature from the past two decades was surveyed across major scientific databases, focusing on the design, conjugation strategies, biological activity, and mechanisms of action of Pt- and Au-peptide bioconjugates. Results: Reviewed studies reveal distinct behavior for platinum- and gold-based systems. Pt(II)-peptide conjugates primarily retain DNA-reactive interaction, with peptides mainly enhancing cellular uptake, selective targeting and solubility, although improved cytotoxicity is not consistently achieved. In contrast, Pt(IV)-peptide conjugates function as prodrugs, where axial peptide functionalization allows greater structural versatility and sometimes improved selectivity, with therapeutic efficacy strongly depending on intracellular reduction kinetics. Au(I)-peptide conjugates act as directly reactive species targeting thiol- and selenol-containing proteins, whereas Au(III) bioconjugates often behave as redox-activated prodrugs, with peptide conjugation influencing stability and cellular fate. Conclusions: Overall, peptide conjugation represents a powerful but non-trivial approach for optimizing metal-based anticancer agents. The success of metal-peptide bioconjugates critically depends on balancing peptide-mediated delivery with the intrinsic reactivity and activation pathways of the metal center. A function-guided design of bioconjugates is essential to achieve genuine selectivity and therapeutic benefit. Full article

Four alloy coatings were deposited via laser cladding on EH40 steel: a Co-based coating (HG), a Ni-based coating (P0), and two Ni-based composite coatings containing 15 wt.% WC (P15) and 30 wt.% WC (P30). Their low-temperature mechanical properties—hardness, tensile strength, shear strength, and impact toughness—were systematically investigated. Hardness increased with WC content, with P30 being the hardest. HG exhibited the highest tensile strength (577 MPa), exceeding the EH40 substrate baseline. Shear tests revealed strong anisotropy: P0 was stronger longitudinally, but WC addition reversed this trend. P30 showed critically low longitudinal shear strength (187.1 MPa), while HG demonstrated high, nearly isotropic shear performance. Impact toughness decreased for all coatings at lower temperatures (−40 °C to −80 °C). P30 maintained good impact energy at −40 °C and −60 °C but suffered severe embrittlement at −80 °C, correlating with its poor longitudinal shear strength. HG offered the best balance of high strength and isotropic properties. P15 provided a reasonable compromise between enhanced hardness and retained toughness. This study highlights the critical trade-off between surface strength and bulk impact toughness for laser claddings on high-strength steel in low-temperature service. Full article

CO₂ is a promising fracturing fluid for tight reservoirs because it avoids water-phase damage and offers low viscosity, high diffusivity, and strong penetration into fine pore throats, but its pore-scale flow in pore-fracture systems remains difficult to evaluate because thermodynamic state, fractures, and mass transfer act together. In this study, a radial microfluidic model containing randomly distributed microfractures was used with a temperature- and pressure-controlled visualization platform to compare CO₂–oil and water–oil flow. Image segmentation and areal-fraction statistics quantified swept area and final fluid distribution. Gaseous CO₂ at ambient pressure and compressed-liquid CO₂ below the critical temperature differ substantially in density and viscosity, but both retain a discernible CO₂–oil interface and exhibit pressure-driven preferential-path flow. The gaseous case shows strong fracture guidance and fingering, whereas the compressed-liquid velocity series demonstrates increasingly rapid advancement and stronger channeling at excessive velocity. Under near-critical supercritical conditions (35 °C, 8 MPa), progressive oil-color fading ahead of the displacement front shows that dissolution participates while flow expands through matrix pores. Under higher-temperature supercritical conditions, disappearance of the sharp interface and continuous color attenuation identify dissolution-assisted diffusion as a significant transport mechanism and produce diffuse redistribution across the pore space. Water undergoes immiscible channelized displacement and remains capillary-trapped in small throats and low-permeability regions. The results identify three flow regimes: distinct-interface pressure-driven displacement, near-critical convection–dissolution coupling, and higher-temperature supercritical dissolution-assisted diffuse redistribution. Full article

Background: While vaccines are designed to elicit targeted immune responses, in some cases, the immunogenic molecules employed can inherently interact with broader host cellular pathways as a secondary consequence. This phenomenon can be exemplified by COVID-19 vaccines. COVID-19 vaccines, including mRNA platforms, use the SARS-CoV-2 spike protein as an immunogen to induce the production of neutralizing antibodies. The spike protein binds the ACE2 (angiotensin-converting enzyme 2) receptor on human cells, mediating viral entry and infection. ACE2 is widely expressed across multiple tissues and is a key component of the renin–angiotensin–aldosterone system (RAAS) that acts as a homeostatic regulator of systemic and local blood flow, blood pressure, cardiac function, fluid balance and immunity. Some studies have proposed the interaction between the spike protein and ACE2 as a possible contributing factor to rare adverse effects observed following COVID-19 vaccination, including myocarditis, pericarditis, thrombosis, and reported alterations in blood pressure, though these mechanisms remain to be fully elucidated. Objectives: As a proof-of-concept approach in vaccine antigen development, we engineered SARS-CoV-2 spike mutants with impaired binding to the host receptor ACE2. Methods: By rational design, we produced and validated in vitro and in vivo spike point mutants that do not effectively bind ACE2. Results: The engineered spike mutants do not effectively bind the human entry receptor ACE2 while retaining the immunogenic properties equal to or better than the wild type spike and thus generate a protective response in animals when used as a vaccination agent. Conclusions: By establishing a straightforward molecular strategy for rational vaccine design, this work demonstrates the feasibility of limiting specific antigen–host receptor interactions while maintaining immunogenicity. This approach may be applicable to future vaccination strategies where antigen interaction with host cells could potentially interfere with physiological pathways. Full article

Tool-calling LLM agents are vulnerable to indirect prompt injection: externally retrieved data can redirect tool calls without system-prompt access, and prompt-level defences leave three harm classes undefended (path traversal, user-guided exfiltration, high-frequency tool abuse). We present a Policy Enforcement Point (PEP) that intercepts at the tool-call boundary with declarative rules over a cross-step information-flow label system (source integrity, data sensitivity) and a synchronous SHA-256 hash-chained audit log. On a controlled dataset across four attack classes, the full system cuts the attack success rate (ASR) from 40.0% to 5.0% (deepseek-v4-pro, five repeats) versus 35.0% for the strongest prompt-only baseline; disabling cross-step label propagation raises the call-level false-negative rate by 26.4 points. The 30.0% task-level false-positive rate is dominated by by-design least-privilege capability-token denials, not rule false positives—an expanded 30-task benign set yields 0/30 rule false positives under scripted isolation. A conservative-DS mitigation (intent-taint) closes the constructed denied-read reconstruction blind-spot variant (ASR 100% to 0%) at no cost on standard workflows. The audit log detects all three tested tamper classes; the in-process enforcement overhead is sub-millisecond per call. Across four further backends, ASR drops under the full system, though LLaMA-3.3-70B retains 16.7% (a rule-coverage gap). A preliminary run over a real MCP stdio transport (an official filesystem server) shows the mechanism operates at a real boundary with a sub-millisecond execution-path increment. We frame these as mechanism-coverage evidence on a controlled benchmark, not a deployability claim for production MCP workloads. Code, data, and metrics are openly available in the replication repository. Full article

Non-intrusive load monitoring (NILM) estimates component-level energy use from aggregate measurements, but low-frequency data limit appliance signatures and make overlapping or weakly observed loads difficult to separate. This paper proposes a relation-aware multi-driver pipeline for interpretable low-frequency load attribution under partial observability. The method does not require supervised component labels or predefined appliance models. It combines semantic feature typing, heterogeneous relation discovery, feature-family construction, mechanism-aware evidence modeling, conservative allocation, event-background separation, and role-based attribution. Only evidence-supported load is assigned to feature families, while unsupported variation is retained as unexplained demand or residual load. The method is evaluated in a simulated EV-focused building case and through measured-building validation on nine ADRENALIN buildings. In the EV case, the selected EV-aligned family achieved a correlation of 0.990 and an NMAE of 0.100 against the withheld EV reference, while heat-pump and base-load recovery was weaker, with NMAE values of 0.565 and 0.895. In the ADRENALIN validation, temperature-associated families achieved median NMAE values of 0.594 using the restricted feature set and 0.576 using the full feature set. Additional comparison, ablation, sensitivity, diagnostic, and runtime analyses show that the pipeline is most effective for dominant event-driven loads, remains limited for smoother or masked lower-magnitude components, and treats unexplained variation explicitly. The results demonstrate a practical framework for interpretable driver-based load attribution when component labels are unavailable or incomplete. Full article

Background/Objectives: Germ cell tumors (GCTs) are thought to arise from primordial germ cells that fail to appropriately differentiate and instead retain pluripotency programs. PRDM14 is a key regulator of pluripotency and primordial germ cell specification and is aberrantly expressed in multiple GCT subtypes. However, the role of PRDM14 in GCT malignancy remains unclear. In this study, we investigated whether PRDM14 functions in GCTs through CBFA2T2, a transcriptional corepressor previously identified as a PRDM14-interacting partner in pluripotent stem cells and developmental models. Methods: To determine the presence and level of PRDM14 and CBFA2T2 in GCT, a panel of GCT lines were assessed for RNA and protein expression and interaction. Then, to better understand the biological effects of PRDM14 and CBFA2T2 within GCTs, PRDM14 and CBFA2T2 knockdowns were employed. Results: We show that PRDM14 and CBFA2T2 are expressed across GCT cell lines, colocalize predominantly in the nucleus, and cooperate as a complex in GCT cell lines. Knockdown of either PRDM14 or CBFA2T2 resulted in reduced expression of key pluripotency genes and a significant impairment of cell proliferation, indicating a shared role in maintaining an undifferentiated, proliferative state. Transcriptomic analysis following PRDM14 or CBFA2T2 depletion revealed extensive overlap in differentially expressed genes and convergent alterations in developmental and metabolic signaling pathways. Conclusions: Together, these findings suggest that PRDM14 and CBFA2T2 form a functional complex that sustains pluripotency and proliferation in GCT cells. This supports a model in which persistence of germline regulatory mechanisms contributes to GCT malignancy, highlighting this interaction as a novel component of GCT biology. Full article

The synergistic use of natural bamboo fiber and synthetic aramid fiber in asphalt mixtures has received limited research attention, particularly regarding the optimal blending ratio under a constant total fiber content and the underlying reinforcement mechanisms. This study systematically investigated the co-blending of bamboo and aramid fibers in SMA-13 asphalt mixtures with a fixed total fiber content of 0.3%. Five mixture groups were prepared, LF (0.3% lignin fiber, control), BF (0.3% bamboo fiber), as well as three hybrid groups: ABF-1 (0.27% bamboo + 0.03% aramid, 9:1), ABF-2 (0.24% bamboo + 0.06% aramid, 4:1), and ABF-3 (0.21% bamboo + 0.09% aramid, 7:3). The mixtures were evaluated using rutting tests, low-temperature flexural beam tests, moisture stability tests, and AMPT dynamic modulus testing. The results demonstrate that hybrid-fiber mixtures outperform single-fiber mixtures, with ABF-2 exhibiting the best overall performance. Compared with LF and BF, ABF-2 achieved a dynamic stability of 6921 passes/mm (increases of 97.7% and 52.7%, respectively); flexural tensile strength increased by 43.1% and 32.1%; maximum flexural tensile strain increased by 42.6% and 35.0%; and retained stability improved by 10.8% and 12.5%. AMPT results indicated a higher dynamic modulus and lower phase angle for the hybrid system, suggesting an enhanced elastic response. The superior performance of ABF-2 is attributed to the complementary adsorption–stabilization effect of bamboo fiber and bridging–reinforcement effect of aramid fiber. This study provides quantitative evidence for the beneficial combination of natural and synthetic fibers in asphalt mixtures and identifies key limitations that warrant future investigation. Full article

Wearable electrocardiogram (ECG) monitoring enables continuous cardiovascular assessment, yet signals acquired in ambulatory environments are inevitably corrupted by baseline wander, electrode motion artifacts, and muscle interference, which obscure diagnostically critical waveform features. Existing deep learning denoisers rely on heuristic attention mechanisms and time-domain-only losses, lacking principled control over what information the network retains or discards. To address this limitation, we propose EFIB-Net, an information bottleneck-guided multi-resolution network for robust ECG denoising. The framework introduces two complementary components: an efficient frequency-guided attention module that derives temporal attention weights directly from the energy distribution of parallel multi-resolution convolutional branches, requiring only four learnable parameters while providing physically interpretable feature selection that naturally highlights QRS complexes, and a variational information bottleneck constraint at the encoder–decoder bottleneck that forces the latent representation to retain only reconstruction-relevant information and discard noise, guided by a spectral–temporal composite loss. To the best of our knowledge, we are among the first to explicitly introduce the information bottleneck principle into deep-learning-based ECG signal denoising. Experiments on the MIT-BIH Arrhythmia Database show that EFIB-Net outperforms ten traditional and deep learning baselines across four standard metrics—signal-to-noise ratio (SNR), root mean square error, percentage root-mean-square difference, and correlation coefficient; at an input SNR of −5 dB it reaches $8.12$ dB output SNR, surpassing the strongest attention-based competitor by 1.77 dB ($p<0.01$) while using only 0.45 M parameters and 10.8 ms inference latency per segment; downstream evaluation further demonstrates that the denoised signals achieve 99.18% R-peak detection sensitivity and 91.26% heartbeat classification F1-score, both within approximately one percentage point of the clean-signal upper bound, making it practical for real-time cardiac monitoring on resource-constrained wearable devices. Zero-shot cross-database evaluation on the QT Database further confirms generalizability, with only 0.54 dB degradation without retraining. Full article

The Fifth Industrial Revolution (Industry 5.0) requires human-centric mechanisms that preserve the integrity, reproducibility, and interpretability of AI-driven decisions in smart manufacturing. Injection molding generates heterogeneous, imbalanced, and weakly labeled process data, posing reliability and integrity risks to data-driven quality control. This study proposes an integrity-verified and reproducibility-instrumented secure machine learning framework for operating-regime analysis in injection molding that integrates (i) SHA-256-based data-integrity verification at ingestion, (ii) Pearson correlation-based feature selection, and (iii) a Gaussian Mixture Model (GMM) under a passive-adversary threat model with Transport Layer Security (TLS)-secured transmission. Evaluated on real industrial data (n = 6719 cycles, seven process variables), correlation-based feature selection retained four non-redundant variables and improved the GMM Silhouette Score from 0.274 ± 0.075 (all features) to 0.323 ± 0.014 (95% CI [0.318, 0.329]), a +18.2% relative improvement (paired t(29) = 3.39, p = 0.002; Cohen’s d = 0.62; Wilcoxon p = 0.022), while lowering the Davies–Bouldin Index from 1.63 to 1.17. The Silhouette standard deviation of 0.014 over 30 seeds meets the σ ≤ 0.02 reproducibility target. The GMM resolves four interpretable operating regimes—one low-load regime consistent with nominal operation and three elevated-load regimes (left-side, right-side, and bilateral)—with operator-readable per-variable signatures. Relative to hard-partition and projection baselines, the GMM is not Silhouette-optimal but provides an interpretable, generative regime model that meets the σ ≤ 0.02 reproducibility target. The framework operationalizes human-centric manufacturing security as measurable integrity, reproducibility, and interpretability. Full article

The influence of ausforming temperature on the isothermal bainitic transformation behavior of 42CrMo steel was systematically investigated using thermo-mechanical simulation, dilatometric analysis, and electron backscatter diffraction (EBSD). The results show that ausforming significantly accelerates the bainitic transformation kinetics, whereas lower ausforming temperatures lead to a progressive reduction in the final bainite fraction. This apparently contradictory behavior originates from the competitive interaction between deformation-induced mechanical stabilization of austenite and dislocation-assisted heterogeneous nucleation of bainitic ferrite. Lower ausforming temperatures result in higher retained dislocation densities, which promote early-stage nucleation while simultaneously increasing resistance to transformation interface migration and hindering carbon redistribution. As a consequence, the bainitic ferrite microstructure is markedly refined, exhibiting reduced lath thickness and length. Crystallographic analysis reveals that the bainitic ferrite predominantly follows the Kurdjumov–Sachs orientation relationship with prior austenite, and that strong variant selection is induced by ausforming, particularly at lower deformation temperatures. The reduced variant multiplicity within individual prior austenite grains further contributes to the refinement and preferential orientation of the bainitic microstructure. These findings highlight the critical role of ausforming temperature in governing the coupled evolution of transformation kinetics, phase fraction, and crystallographic characteristics during bainitic transformation and provide guidance for microstructural control of bainitic steels through temperature-dependent thermo-mechanical processing. Full article

Alpine permafrost and seasonally frozen ground threaten the long-term safe operation of highway infrastructures. Aiming at the structural performance optimization of prefabricated corrugated steel plate retaining walls in alpine permafrost regions, this study adopted finite element numerical simulation combined with field test validation to systematically explore the influences of wall height, plate thickness, corrugation geometry, and tie reinforcement layout on structural deformation and internal force, and carried out targeted parameter optimization. The core innovations include the following: (1) Structural lateral displacement and internal force rise nonlinearly with the increase in wall height, and high retaining walls exhibit an accelerated growth trend of deformation and stress. (2) Increasing plate thickness can effectively reduce structural displacement and stress, while the improvement effect gradually weakens after exceeding a critical thickness. Specifically, when the thickness increases from 4 mm to 5 mm, the displacement decreases by 33.13%. (3) Appropriately increasing corrugation pitch and height improves structural equivalent stiffness and optimizes stress distribution. Increasing the corrugation pitch from 75 mm to 400 mm and corrugation height from 25 mm to 150 mm reduces the maximum horizontal displacement by 52.6%. This demonstrates that larger corrugation profiles significantly improve structural stiffness. For walls higher than 6 m, the spacing should be reduced to 0.8 m × 1.0 m to provide additional lateral restraint. (4) Furthermore, seasonal freeze–thaw cycles and a non-uniform temperature field significantly amplify structural displacement and stress. After 12 months of freeze–thaw cycles, the maximum horizontal displacement increases by 49.7% and the maximum equivalent stress increases by 56.9% compared to the initial state. This study clarifies the parameter control mechanism and temperature coupling effect and provides a reliable theoretical basis and design reference for the engineering application of prefabricated corrugated steel plate retaining walls in alpine permafrost areas. Full article

Crop leaf diseases cause 10–40% annual yield losses, yet timely field diagnosis remains difficult. Vision-language models (VLMs) lift recognition accuracy with rich textual descriptions, but multimodal pipelines are too slow for real-time field use because they require text processing at inference. We present MTL-AWL, a framework built on a training–inference asymmetry: VLM text serves as privileged training-time supervision, and two coupled mechanisms—one retaining VLM semantics in the image encoder and one exploiting them—enable image-only deployment at multimodal accuracy. A modal-dropout strategy ($p=0.6$) intermittently masks the VLM text sequence during training, forcing the image encoder to retain cross-modal representations independently. An adaptive multi-task loss jointly optimizes InfoNCE contrastive alignment, attention diversity, and modality consistency under learnable softmax weights, consistently converging to a dominant contrastive weight (55% on soybean, 68% on PlantDoc)—identifying cross-modal alignment as the primary mechanism of VLM knowledge transfer. At inference, the model reaches 818 FPS (3.7× faster than multimodal methods) at only 0.41% accuracy cost, attaining 99.30%/98.89% (multimodal/image-only) on soybean and 72.65%/68.80% on PlantDoc—compact enough for real-time, offline field screening. Full article