Amyloid fibrils from soybean 7S and 11S proteins can improve emulsion stability by creating a bridge-like structure at the oil-water interface. To clarify the structural features and potential for stabilising highinternal-phase emulsions of 7S/11S protein amyloid fibrils in varying proportions, more investigation is necessary. This study examined the properties of various emulsion types with oil contents of 30 % and 85 %, as well as the characteristics of amyloid fibrils of different proportions of 7S/11S soybean protein (1:0, 0:1, 1:1, 2:1, and 1:2). The 7S/11S ratio of 2:1 had the highest fibrils beta-sheet content (51.40 %), the highest ThT fluorescence intensity and the strongest fibrils-forming ability. After amyloid fibrillation treatment, emulsions made with different amounts of 7S/11S soy protein showed improved stability compared to the control group. Additionally, amyloid fibrils consisting of 85 % of oil emulsion had superior characteristics than those of 30 %. The emulsion's EAI and ESI initially increased and subsequently decreased as 7S percentage increased. When the 7S/11S soybean protein ratio was 2:1, the emulsion had the highest viscoelasticity, EAI (89.72 m2/g) and ESI (78.96 min) were the highest, and the smallest particle size (90.9 nm) and lowest TSI values, indicating that the 7S/11S soybean protein amyloid fibrils have excellent potential for preparing high-internal-phase emulsion. This study offers a theoretical foundation for employing soy protein in high internal phase emulsions and for comprehending the structure of the various 7S and 11S protein components from soy amyloid fibrils.
Protein, as a core parameter for assessing the rice nutritional quality, shows a close correlation between its content and cooking flavor characteristics. Aiming at complicated operation process, high time-cost and high reagent consumption of traditional chemical methods for detecting protein content, this study investigated the rapid detection method for protein content in milled rice with different particle sizes of pulverization (0.5 mm, 1.0 mm, 1.5 mm, and whole particle) using near-infrared spectroscopy. By combining wavelength selections with multivariate quantitative corrections, the optimal construction method of the spectral regression model for rice protein content with different particle sizes was explored. Following comparative analyses, it was determined that the 1.0 mm particle size model constructed by the uninformative variable elimination combined with the deep extreme learning mechanism exhibited optimal performance. Coefficients of determination for the validation and external test sets were found to be 0.9697 and 0.9641, with relative root mean square errors of 2.50 % and 3.01 %, and residual prediction deviations of 5.8383 and 5.4370, respectively. This method not only provides a reliable nonlinear modeling method for the analysis of grain quality, but also provides an innovative solution for the adaptation of multi-scale data and efficient algorithms in the spectral quantitative analysis of agricultural products.
Efficient and accurate identification of millet weeds enables weeding robots to perform precise pesticide application or mechanical weeding. To facilitate the effective deployment of the model on mobile devices with limited resources, a novel lightweight weed detection method, LRS-YOLOv8, is proposed in this research. The algorithm is composed of the Lightweight Shared Convolutional Group Normalization Head (LWSCGNH) detection head, the RepConv Lightweight Cross Stage Partial (RLWCSP) module, and the Inner Wise Shape-Intersection over Union (IWS-IoU) loss function. With the LWSCGNH, the model achieves better detection of small target weeds and retains its lightweight properties. The RLWCSP module is designed to enhance multi-scale feature extraction and gradient flow capabilities, while computation and storage costs are reduced. The IWS-IoU loss function improves bounding box regression performance and enhances the detection of small target weeds, while also addressing the weed occlusion problem. The experimental results indicate that, for the millet weed dataset collected in a field environment, compared to YOLOv8n, the improved model enhances the precision, recall, and mAP50 metrics by 4.5 %, 1.9 %, and 0.6 %, respectively. Moreover, the model achieves a 48.3 % reduction in parameters, a 38.7 % decrease in size, and a 44.4 % drop in floating-point operations per second. The detection speed reaches 203 FPS, satisfying the requirements for lightweight and real-time performance, and surpassing other comparable object detection methods. Model optimization and deployment were successfully implemented on the RK3588 mobile platform for efficient multi-target detection. It provides a reference for deploying millet weed detection models on resource-constrained mobile devices.
Staphylococcus aureus is a key pathogen in bovine mastitis, and antibiotic therapy is challenged by resistance and residue concerns. Milk-derived extracellular vesicles emerge as promising natural antimicrobials. This study aimed to evaluate the antimicrobial activity and explore potential associated mechanisms of milk-derived extracellular vesicles against S. aureus. Milk-derived EV-enriched fractions (mEVs) from healthy (HmEVs) and mastitic (MmEVs) bovine milk suppressed S. aureus growth in vitro and were associated with oxidative imbalance, with MmEVs showing stronger inhibition. In addition, MmEVs significantly reduced biofilm biomass, extracellular matrix production, and the expression of key biofilm-associated genes (sarA, icaB, fnbA, clfB, cidA). Small RNA sequencing revealed distinct miRNA profiles between HmEVs and MmEVs; in particular, MmEVs were enriched in miRNAs predicted to target the S. aureus biofilm-associated gene clfB. Although we did not directly demonstrate uptake of mEV-derived miRNAs by bacteria or their regulation of bacterial gene expression in this study, our small RNA sequencing data together with subsequent bioinformatic predictions suggest that vesicular miRNAs should be regarded as candidate contributors, rather than demonstrated mediators, of the observed antibacterial and antibiofilm effects. Taken together, these findings indicate the potential of mEVs as residue-free adjuncts for controlling bovine mastitis, while recognizing that the present conclusions are mainly derived from in vitro experiments with S. aureus and bioinformatic analyses. Therefore, functional validation of candidate miRNAs, in vivo studies, and evaluation of activity against other mastitis-associated pathogens are still required to clarify the underlying mechanisms, therapeutic potential, and spectrum of activity of mEVs.
As electronic systems continue to scale in integration and thermal density, managing high heat fluxes poses a persistent challenge. Here we propose a hybrid topology optimization framework for microchannel heat sinks (MCHS), integrating node-based density parameterization, a three-field Heaviside projection (expansion, erosion, baseline), and density-to-level-set mapping, augmented by a temperature-gradient-driven dynamic hole generation mechanism. This approach enables sharply resolved structural boundaries and adaptively improves coolant distribution. Compared to conventional straight microchannel designs under identical thermal loads and flow rates, the optimized configurations reduce the average surface temperature by 5.7-9.5 % and enhance the convective heat transfer coefficient by 71-104 %, while incurring a 176-219 % increase in pressure drop. To assess the thermal-hydraulic trade-off, a figure of merit (FOM)-defined as the normalized ratio of heat transfer enhancement to pumping power penalty-is employed. Across all tested conditions, FOM values exceed unity, peaking at approximately 1.65, indicating an overall gain in system-level performance. Experimental measurements agree closely with numerical predictions, with average temperature deviations confined within 1-5 % under multiple operating conditions.
