A new interpretable streamflow prediction approach based on SWAT-BiLSTM and SHAP.

Streamflow is a crucial variable for assessing the available water resources for both human and environmental use. Accurate streamflow prediction plays a significant role in water resource management and assessing the impacts of climate change. This study explores the potential of coupling conceptual hydrological models based on physical processes with machine learning algorithms to enhance the performance of streamflow simulations. Four coupled models, namely SWAT-Transformer, SWAT-LSTM, SWAT-GRU, and SWAT-BiLSTM, were constructed in this research. SWAT served as a transfer function to convert four meteorological features, including precipitation, temperature, relative humidity, and wind speed, into six hydrological features: soil water content, lateral flow, percolation, groundwater discharge, surface runoff, and evapotranspiration. Machine learning algorithms were employed to capture the underlying relationships between these ten feature variables and the target variable (streamflow) to predict daily streamflow in the Sandu-River Basin (SRB). Among the four coupled models and the calibrated SWAT model, SWAT-BiLSTM exhibited the best streamflow simulation performance. During the calibration period (training period), it achieved R2 and NSE values of 0.92 and 0.91, respectively, and maintained them at 0.90 during the validation period (testing period). Additionally, the performance of all four coupled models surpassed that of the calibrated SWAT model. Compared to the tendency of the SWAT model to underestimate streamflow, the absolute values of PBIAS for all coupled models are below 10%, which indicates that there is no significant systematic bias evident. SHapley Additive exPlanations (SHAP) were used to analyze the impact of different feature variables on streamflow prediction. The results indicated that precipitation contributed the most to streamflow prediction, with a global importance of 29.7%. Hydrological feature variable output by the SWAT model played a dominant role in the Bi-LSTM's prediction process. Coupling conceptual hydrological models with machine learning algorithms can significantly enhance the predictive performance of streamflow. The application of SHAP improves the interpretability of the coupled models and enhances researchers' confidence in the prediction results.

Unraveling the antimicrobial potential of Lactiplantibacillus plantarum strains TE0907 and TE1809 sourced from Bufo gargarizans: advancing the frontier of probiotic-based therapeutics.

Introduction: In an era increasingly defined by the challenge of antibiotic resistance, this study offers groundbreaking insights into the antibacterial properties of two distinct Lactiplantibacillus plantarum strains, TE0907 and TE1809, hailing from the unique ecosystem of Bufo gargarizans. It uniquely focuses on elucidating the intricate components and mechanisms that empower these strains with their notable antibacterial capabilities. Methods: The research employs a multi-omics approach, including agar diffusion tests to assess antibacterial efficacy and adhesion assays with HT-29 cells to understand the preliminary mechanisms. Additionally, gas chromatography-mass spectrometry (GC-MS) is employed to analyze the production of organic acids, notably acetic acid, and whole-genome sequencing is utilized to identify genes linked to the biosynthesis of antibiotics and bacteriocin-coding domains. Results: The comparative analysis highlighted the exceptional antibacterial efficacy of strains TE0907 and TE1809, with mean inhibitory zones measured at 14.97 and 15.98 mm, respectively. A pivotal discovery was the significant synthesis of acetic acid in both strains, demonstrated by a robust correlation coefficient (cor ≥ 0.943), linking its abundance to their antimicrobial efficiency. Genomic exploration uncovered a diverse range of elements involved in the biosynthesis of antibiotics similar to tetracycline and vancomycin and potential regions encoding bacteriocins, including Enterolysin and Plantaricin. Conclusion: This research illuminates the remarkable antibacterial efficacy and mechanisms intrinsic to L. plantarum strains TE0907 and TE1809, sourced from B. gargarizans. The findings underscore the strains' extensive biochemical and enzymatic armamentarium, offering valuable insights into their role in antagonizing enteric pathogens. These results lay down a comprehensive analytical foundation for the potential clinical deployment of these strains in safeguarding animal gut health, thereby enriching our understanding of the role of probiotic bacteria in the realm of antimicrobial interventions.

Corrigendum: The Effect of High-Diet and Exercise Intervention on the TNF-α Level in Rat Spleen.

[This corrects the article DOI: 10.3389/fimmu.2021.671167.].

The Effect of High-Fat Diet and Exercise Intervention on the TNF-α Level in Rat Spleen.

High-fat diet (HFD) consumption can trigger chronic inflammation in some tissues. However, it remains unclear if HFD induces chronic inflammation in the spleen. This investigation aims to address the effect of HFD consumption and exercise intervention on the level of tumor necrosis factor alpha (TNF-α) in the spleen. Rats were subjected to HFD feeding and/or moderate-intensity treadmill running. The TNF-α levels in plasma and spleen were detected by ELISA. The mass and total cell numbers of the spleen were measured. In addition, the expression of TNF-α and its relevant gene mRNAs in macrophages from the spleen were analyzed by qRT-PCR. We found that HFD consumption did not significantly affect the mass and total cell numbers of the spleen. However, HFD consumption significantly increased splenic TNF-α level, the expression of TNF-α, toll-like receptor 4, and nuclear factor κB p65 mRNAs. In contrast, the expression of nicotinic acetylcholine receptor alpha 7 subunit (α7nAChR) mRNA in macrophages was downregulated. Additionally, exercise abolished the increase in splenic TNF-α level as well as the abnormal expression of TNF-α and related gene mRNAs in macrophages in HFD-fed rats. In conclusion, our results reveal that HFD consumption increases TNF-α level in the spleen, which is along with upregulation of the expression of TLR4 and NF-κB mRNAs as well as downregulation of the expression of α7nAChR mRNA in splenic macrophages in rats. Exercise abolished detrimental effects of HFD on TNF-α level in the spleen and prevented abnormal expression of these genes in the macrophages from rat spleen.

Design and synthesis of affinity ligands and relation of their structure with adsorption of proteins.

Affinity chromatography has played an increasingly important role both in the pharmaceutical industry and academic research. In the present study, we report our preliminary investigation on the relationship between the affinity ligand structure and its adsorption to multi-protein samples. The structure of the ligands, including the size of the ring (cyclic group) and the length of the chain (linear group), has a great impact on the adsorption of ligands to proteins. Meanwhile, the functional groups that the ligands carry are also closely related to the adsorption of ligands to proteins. This research provides good guidance for the design and synthesis of affinity materials in affinity chromatography. It is also useful to other protein-ligand interaction-related research.

Pre-fractionation of rat liver cytosol proteins prior to mass spectrometry-based proteomic analysis using tandem biomimetic affinity chromatography.

Efficient and high resolution separation of the protein mixture prior to trypsin digestion and mass spectrometry (MS) analysis is generally used to reduce the complexity of samples, an approach that highly increases the probability of detecting low-copy-number proteins. Our laboratory has constructed an affinity ligand library composed of thousands of ligands with different protein absorbance effects. Structural differences between these ligands result in different non-bonded protein-ligand interactions, thus each ligand exhibits a specific affinity to some protein groups. In this work, we first selected out several synthetic affinity ligands showing large band distribution differences in proteins absorbance profiles, and a tandem composition of these affinity ligands was used to distribute complex rat liver cytosol into simple subgroups. Ultimately, all the fractions collected from tandem affinity pre-fractionation were digested and then analyzed by LC-MS/MS, which resulted in high confidence identification of 665 unique rat protein groups, 1.8 times as many proteins as were detected in the un-fractionated sample (371 protein groups). Of these, 375 new proteins were identified in tandem fractions, and most of the proteins identified in un-fractionated sample (290, 80%) also emerged in tandem fractions. Most importantly, 430 unique proteins (64.7%) only characterized in specific fractions, indicating that the crude tissue extract was well distributed by tandem affinity fractionation. All detected proteins were bioinformatically annotated according to their physicochemical characteristics (such as MW, pI, GRAVY value, TM Helices). This approach highlighted the sensitivity of this method to a wide variety of protein classes. Combined usage of tandem affinity pre-fractionation with MS-based proteomic analysis is simple, low-cost, and effective, providing the prospect of broad application in proteomics.