Objective: To analyze the association between each regional fat mass and total body bone mineral density (BMD) in children and adolescents aged 7-17 years in Beijing, so as to provide theoretical basis and practical guidance for implementing interventions. Methods: From September to December 2020, a stratified cluster random sampling method was used to select 1 423 children and adolescents aged 7-17 years in Tongzhou District, Beijing. Dual energy X-ray absorptiometry (DXA) was employed to measure regional body composition and total body BMD. Multiple linear regression was used to analyze the association between regional fat mass and total body BMD. Results: The median (interquartile range) fat mass values for total body, upper limbs, abdomen, hips, and thighs were 13.51(8.84, 19.21), 1.59(1.08, 2.23), 0.73(0.39, 1.29), 2.32(1.46, 3.26), 5.29(3.59, 7.21)kg, respectively. After adjusting for covariates, the results of multiple linear regression analysis showed that total body fat mass (β=0.010), abdominal fat mass (β=-0.100), and hip fat mass (β=0.104) were significant associations with total body BMD (all P<0.01). Sexstratified analysis revealed that in boys, total body fat mass (β=0.008) and hip fat mass (β=0.058) were positively associated with BMD, while thigh fat mass (β=-0.038) showed a negative association with total body BMD (all P<0.05). In girls, total body fat mass (β=0.013), hip fat mass (β=0.163), and thigh fat mass (β=0.023) were positively associated with total body BMD, whereas abdominal fat mass (β=-0.196) showed a negative association with total body BMD (all P<0.05). Among children and adolescents with body fat percentage below the standard range, within the standard range and above the standard range, total body fat masses were positively associated with total body BMD (β=0.021, 0.016, 0.015); among children and adolescents with body fat percentage within the standard range while upper limb (β=-0.042), abdominal (β=-0.067), and thigh fat mass (β=-0.018) showed negative associations with total body BMD, and hip fat mass demonstrated a positive association with total body BMD (β=0.082) (all P<0.05). Conclusion Regional fat distribution is associated with total body BMD in children and adolescents, with the nature of these associations varying by sex and body fat percentage.

Computational analysis can accurately detect drug-gene interactions (DGIs) cost-effectively. However, transductive learning models are the hotspot to reveal the promising performance for unknown DGIs (both drugs and genes are present in the training model), without special attention to the unseen DGIs (both drugs and genes are absent in the training model). In view of this, this study, for the first time, proposed an inductive learning-based model for the precise identification of unseen DGIs. In our study, by integrating disease nodes to avoid data sparsity, a multi-relational drug-disease-gene (DDG) graph was constructed to achieve effective fusion of data on DDG intro-relationships and inter-actions. Following the extraction of graph features by utilizing graph embedding algorithms, our next step was the retrieval of the attributes of individual gene and drug nodes. In this way, a hybrid feature characterization was represented by integrating graph features and node attributes. Machine learning (ML) models were built, enabling the fulfillment of transductive predictions of unknown DGIs. To realize inductive learning, this study generated an innovative idea of transforming known node vectors derived from the DDG graph into representations of unseen nodes using node similarities as weights, enabling inductive predictions for the unseen DGIs. Consequently, the final model was superior to existing models, with significant improvement in predicting both external unknown and unseen DGIs. The practical feasibility of our model was further confirmed through case study and molecular docking. In summary, this study establishes an efficient data-driven approach through the proposed modeling, suggesting its value as a promising tool for accelerating drug discovery and repurposing.

Accurate prediction of drug-induced adverse drug reactions (ADRs) is crucial for drug safety evaluation, as it directly impacts public health and safety. While various models have shown promising results in predicting ADRs, their accuracy still needs improvement. Additionally, many existing models often lack interpretability when linking molecular structures to specific ADRs and frequently rely on manually selected molecular fingerprints, which can introduce bias. To address these challenges, we propose ToxBERT, an efficient transformer encoder model that leverages attention and masking mechanisms for simplified molecular input line entry system (SMILES) representations. Our results demonstrate that ToxBERT achieved area under the receiver operating characteristic curve (AUROC) scores of 0.839, 0.759, and 0.664 for predicting drug-induced QT prolongation (DIQT), rhabdomyolysis, and liver injury, respectively, outperforming previous studies. Furthermore, ToxBERT can identify drug substructures that are closely associated with specific ADRs. These findings indicate that ToxBERT is not only a valuable tool for understanding the mechanisms underlying specific drug-induced ADRs but also for mitigating potential ADRs in the drug discovery pipeline.

Efficient drug response prediction is crucial for reducing drug development costs and time, but current computational models struggle with limited experimental data and out-of-distribution issues between in vitro and in vivo settings. To address this, we introduced drug response prediction meta-learner (metaDRP), a novel few-shot learning model designed to enhance predictive accuracy with limited sample sizes across diverse drug-tissue tasks. metaDRP achieves performance comparable to state-of-the-art models in both genomics of drug sensitivity in cancer (GDSC) drug screening and in vivo datasets, while effectively mitigating out-of-distribution problems, making it reliable for translating findings from controlled environments to clinical applications. Additionally, metaDRP's inherent interpretability offers reliable insights into drug mechanisms of action, such as elucidating the pathways and molecular targets of drugs like epothilone B and pemetrexed. This work provides a promising approach to overcoming data scarcity and out-of-distribution challenges in drug response prediction, while promoting the integration of few-shot learning in this field.

Objective: To explore the mechanism of action of hepatic leukemia factor (HLF) in diabetes mellitus and to construct a conditional animal model of mice with islet β ⁃cell⁃specific HLF gene knockout. Methods: At the cellular level , the effects of HLF inhibition or overexpression on the proliferation of MIN6 cells was verified by the CCK⁃8 assay. The effects of HLF inhibition or overexpression were detected at the mRNA level and protein level by Cre + / - mice (C57BL/6J) to obtain offspring mice. The genotypes of the mice were identified by the PCR method.The differences in the expression levels of the HLF gene at the mRNA and protein levels in islet β ⁃cell knockout mice (HLFflox/flox Cre + / - ) and control mice (HLFflox/flox ) were detected by RT⁃qPCR technology and Western blot technology to verify the knockout effect. At the same time , the islet tissues of the mice in two groups were taken to make paraffin sections and analyzed by hematoxylin⁃eosin (HE) staining. Results: HLF gene inhibition or overex⁃pression had no significant effect on the proliferation of MIN6 cells. When the HLF gene was inhibited in MIN6 cells , the mRNA expression level decreased by 74% compared with the control group , and the protein expression level decreased by 60% compared with the control group. After overexpressing the HLF gene , the mRNA expres⁃sion level was 2. 13 times compared with that of the control group , and the protein expression level was 1. 8 times compared with that of the control group. The mRNA expression level of the HLF gene in the knockout mice de⁃creased by 89% compared with the control group , and the protein expression level decreased by 65% compared with the control group. The results of HE staining showed that there was no significant difference in the cell mor⁃phology in the islet tissues between the knockout mice and the control mice. Inhibiting HLF increased the glycogen content in MIN6 cells by approximately 20% . Conclusion The HLF gene knockout mice are successfully con⁃structed , providing an animal model for studying the role of HLF in the pathogenesis of diabetes mellitus.

Accurate prediction of drug-induced adverse drug reactions(ADRs)is crucial for drug safety evaluation,as it directly impacts public health and safety.While various models have shown promising results in predicting ADRs,their accuracy still needs improvement.Additionally,many existing models often lack interpretability when linking molecular structures to specific ADRs and frequently rely on manually selected molecular fingerprints,which can introduce bias.To address these challenges,we propose ToxBERT,an efficient transformer encoder model that leverages attention and masking mechanisms for simplified molecular input line entry system(SMILES)representations.Our results demonstrate that ToxBERT achieved area under the receiver operating characteristic curve(AUROC)scores of 0.839,0.759,and 0.664 for predicting drug-induced QT prolongation(DIQT),rhabdomyolysis,and liver injury,respectively,outperforming previous studies.Furthermore,ToxBERT can identify drug substructures that are closely associated with specific ADRs.These findings indicate that ToxBERT is not only a valuable tool for understanding the mechanisms underlying specific drug-induced ADRs but also for mitigating potential ADRs in the drug discovery pipeline.

Computational analysis can accurately detect drug-gene interactions(DGIs)cost-effectively.However,transductive learning models are the hotspot to reveal the promising performance for unknown DGIs(both drugs and genes are present in the training model),without special attention to the unseen DGIs(both drugs and genes are absent in the training model).In view of this,this study,for the first time,proposed an inductive learning-based model for the precise identification of unseen DGIs.In our study,by integrating disease nodes to avoid data sparsity,a multi-relational drug-disease-gene(DDG)graph was constructed to achieve effective fusion of data on DDG intro-relationships and inter-actions.Following the extraction of graph features by utilizing graph embedding algorithms,our next step was the retrieval of the attributes of individual gene and drug nodes.In this way,a hybrid feature charac-terization was represented by integrating graph features and node attributes.Machine learning(ML)models were built,enabling the fulfillment of transductive predictions of unknown DGIs.To realize inductive learning,this study generated an innovative idea of transforming known node vectors derived from the DDG graph into representations of unseen nodes using node similarities as weights,enabling inductive predictions for the unseen DGIs.Consequently,the final model was superior to existing models,with significant improvement in predicting both external unknown and unseen DGIs.The practical feasibility of our model was further confirmed through case study and molecular docking.In summary,this study establishes an efficient data-driven approach through the proposed modeling,suggesting its value as a promising tool for accelerating drug discovery and repurposing.

Objective: To explore sex difference in the cardiovascular health (CVH) status of 6-8 year old children in Beijing, so as to inform the early intervention of CVH related lifestyles. Methods: Based on the Beijing Children s Growth and Health Cohort (PROC), baseline physical examination, sequential questionnaire survey, and laboratory tests were conducted among 1 914 grade 1 students. Children s CVH and its subscales (health behaviors and health factors) scores were calculated according to the Life s Essential 8 (LE 8) index and categorized into high, moderate, and low CVH. CVH scores were reported as medians and interquartile ranges; sex differences were compared using the Chi square test and Wilcoxon test. Results: Among the 1 914 participants, the percentages of high, moderate, and low CVH were 35.7%, 63.5%, and 0.8%, respectively, and the percentages of high, moderate, and low health behavior scores were 25.9%, 67.5%, and 6.6%, respectively, with no statistically significant differences between sex ( χ 2=2.30, 0.07, P >0.05). The rates of high, moderate, and low health factor scores for boys and girls were 61.1%, 36.0%, 2.9% and 71.1%, 28.4%, 0.5%, respectively, with a statistically significant sex difference ( χ 2=31.88, P < 0.01). The overall CVH score was 76.0(70.0, 83.0), 76.0(69.0, 82.0) for boys, and 77.0(71.0, 83.0) for girls. Among the health behavior metrics, sleep scores were the best and physical activity scores were the worst[100.0(90.0,100.0), 40.0(20.0, 80.0 )]; among the health factor metrics, blood glucose scores were the best and lipid scores were the worst[100.0(100.0,100.0), 60.0(40.0,100.0)]. In respect to health factors, there were significant gender differences in body mass index, blood lipids, blood sugar, and blood pressure scores ( Z =-6.92, 3.01, -6.60, -2.30, <0.05), but there were no significant gender differences in diet, physical activity, nicotine exposure, or sleep scores with regards to health behaviors ( Z =0.99, 0.88, -0.13, 0.36, P > 0.05 ). Compared to boys, girls in the low and moderate CVH groups had high health factor scores despite low health behavior scores. Conclusion Most 6 to 8-year-old children in Beijing were found to have relatively good CVH, and optimization of children s CVH status can be achieved by promoting healthier lifestyles and monitoring health factors, especially among boys.

Objective: To explore the longitudinal association between organophosphate esters (OPEs) exposure and blood pressure in children, so as to provide a reference for identifying the effects of OPEs exposure on child health. Methods: A total of 404 children from the Beijing Child Growth and Health Cohort (PROC) were enrolled using a case cohort study design, baseline physical examination, urine collection, questionnaires survey were administered in 2018 and follow up surveys in 2019-2020 and 2023. Participants were divided into case group ( n =140) and control group ( n =264) according to the observation of new onset of high blood pressure during the follow up period. High performance liquid chromatography tandem mass spectrometry was used to detect diethyl phosphate (DEP)，bis (2-chloroethyl) phosphate (BCEP)，bis (1-chloro-2-propyl) phosphate, (BCIPP), diphenyl phosphate(DPHP), dibutyl phosphate (DnBP), bis (1,3-dichloro-2-propyl) phosphate(BDCIPP), bis(2-butoxyethyl) phosphate(BBOEP), bis (2-butoxyethyl) 2-hydroxyethyl phosphate (BBOEHEP), 4-hydroxyphenyl diphenyl phosphate (4-OH-TPHP). Generalized linear mixed models and Quantile g computation models were developed to analyze the longitudinal associations between OPEs individual/mixed exposure and blood pressure in children. Results: The detection range of 9 OPEs metabolites,including DEP, BCEP, BCIPP, DPHP, DnBP, BDCIPP, BBOEP, BBOEHEP and 4-OH-TPHP at three time points (baseline, first follow up and second follow up) were 27.7%-92.1%, 24.0%-99.3% and 39.2%-90.9% respectively. Without adjustment for covariates such as gender, age, body mass index, Tanner stage, parental education, and monthly household income, and family history of hypertension, the increase of BDCIPP concentration and mixed exposure of OPEs may reduce children s systolic blood pressure( β= -0.85，-2.40，95%CI=-1.69--0.01，-3.30--1.50，P <0.05). After adjusting for the covariates, the longitudinal association of individual OPEs or mixed exposure with pediatric BP was not statistically significant ( P >0.05). Conclusion Children are commonly exposed to OPEs, and although no significant longitudinal associations are observed between exposure to OPEs and blood pressure among school aged children in Beijing, it is recommended that child exposure should be minimized whenever possible.

Purpose To investigate the expression of C1GALT1 in gastric cancer and its effect on the biological be-havior of BGC-823 in gastric cancer cells.Methods The ex-pression of C1GALT1 mRNA and protein in gastric cancer tis-sues and normal gastric mucosa,gastric cancer cells and normal gastric mucosal cells was analyzed by bioinformatics,qRT-PCR and Western blot;the transient transfection of siRNA into BGC-823 cells was designed with C1GALT1 cDNA sequence as the target.Transwell assay was used to detect the effect of C1GALT1-siRNA on the migration and invasion ability of BGC-823 cells in gastric cancer.Western blot method detected the expression of epithelial-mesenchymal transition(EMT)-related proteins in BGC-823 after transfection of C1GALT1-siRNA.Re-sults C1GALT1 was highly expressed in gastric cancer tissues and cell lines BGC-823,SGC-7901 and MGC-803,and the ex-pression levels were positively correlated with gastric cancer pathological stages Ⅰ and Ⅱ(P＜0.05).After interfering with C1GALT1 in BGC-823 cells,the ability of migration and inva-sion decreased(P＜0.05),epithelial cell markers E-cadherin and Claudin-1 protein expression increased,while mesenchymal cell markers vimentin and Slug protein expression decreased(P＜0.05).Conclusion C1GALT1 is highly expressed in gastric cancer tissues and cells,silencing of C1GALT1 can inhibit mi-gration and invasion ability of gastric cancer,the mechanism may be related to EMT.