Objective To determine the knee joint cartilage thickness using different methods and explore the feasibility of mathematical statistical models of dataset for the prediction of cartilage thickness.Methods A total of 304 patients diagnosed as knee osteoarthritis(OA)combined with varus deformity and undergoing unilateral total knee arthroplasty at the Second Affiliated Hospital of Army Medical University from March 2023 to March 2024 were selected for the study.All patients had complete preoperative and postoperative clinical data.The healthy cartilage at four anatomical sites of patients,including the distal femur lateral condyle,lateral tibial plateau,posterior medial femoral condyle,and posterior lateral femoral condyle were selected,and the knee joint cartilage thickness was determined based on preoperative MRI analysis,robotic navigation system tracing,tissue section of surgical specimen and digital vernier caliper.The baseline indicators of demographics,disease and imaging ffor patients were collected to construct a dataset,and four models of linear regression analysis,principal component analysis,Least Absolute Shrinkage and Selection Operator(LASSO)regression analysis,and K-nearest neighbors(KNN)analysis were established for predicting the accuracy,determination coefficient(R2)and root mean square error(RMSE),and the regression equation for predicting cartilage thickness was established.Results The knee joint cartilage thicknesses determined by preoperative MRI analysis,robotic navigation system tracing,tissue section of surgical specimen had no statistically significant difference with that by digital vernier caliper(P>0.05).The predictive efficiencies of models of linear regression analysis,principal component analysis,and LASSO regression analysis for the knee joint cartilage thickness all failed to meet the expectations(R2<0.3,RMSE>0.03).The predictive effect of KNN model on the cartilage thickness of the distal femur lateral condyle and lateral tibial plateau was not ideal(R2=0.23,RMSE=0.29),while it had potential predictive value(accuracy=0.21,accuracy=0.15).Conclusion The prediction model of knee joint cartilage thickness based on individual parameters has certain scientificity,and the feasibility of KNN model is relatively high.However,due to insufficient sample size and unclear individual parameter weight,the efficiencies of the four established prediction models are not ideal,which fails to provide definite prediction equations.Therefore,the construction scheme of the prediction model still needs to be further optimized.

Based on whole-genome identification of the TPS gene family in Perilla frutescens and screening, cloning, bioinformatics, and expression analysis of the synthetic enzyme for the insect-resistant component germacrene D, this study lays the foundation for understanding the biological function of the TPS gene family and the insect resistance mechanism in P. frutescens. This study used bioinformatics tools to identify the TPS gene family of P. frutescens based on its whole genome and predicted the physicochemical properties, systematic classification, and promoter cis-elements of the proteins. The relative content of germacrene D was detected in both normal and insect-infested leaves of P. frutescens, and the germacrene D synthase was screened and isolated. Gene cloning, bioinformatics analysis, and expression profiling were then performed. The results showed that a total of 99 TPS genes were identified in the genome, which were classified into the TPS-a, TPS-b, TPS-c, TPS-e/f, and TPS-g subfamilies. Conserved motif analysis showed that the TPS in P. frutescens has conserved structural characteristics within the same subfamily. Promoter cis-element analysis predicted the presence of light-responsive elements, multiple hormone-responsive elements, and stress-responsive elements in the TPS family of P. frutescens. Transcriptome data revealed that most of the TPS genes in P. frutescens were highly expressed in the leaves. GC-MS analysis showed that the relative content of germacrene D significantly increased in insect-damaged leaves, suggesting that it may act as an insect-resistant component. The germacrene D synthase gene was screened through homologous protein binding gene expression and was found to belong to the TPS-a subfamily, encoding a 64.89 kDa protein. This protein was hydrophilic, lacked a transmembrane structure and signal peptide, and was predominantly expressed in leaves, with significantly higher expression in insect-damaged leaves compared to normal leaves. In vitro expression results showed that germacrene D synthase tended to form inclusion bodies. Molecular docking showed that farnesyl pyrophosphate(FPP) fell into the active pocket of the protein and interacted strongly with six active sites. This study provides a foundation for further research on the biological functions of the TPS gene family in P. frutescens and the molecular mechanisms underlying its insect resistance.
Perilla frutescens/chemistry* ; Plant Proteins/chemistry* ; Multigene Family ; Sesquiterpenes, Germacrane/metabolism* ; Alkyl and Aryl Transferases/chemistry* ; Phylogeny ; Gene Expression Regulation, Plant

Objective:To develop a preoperative differentiation model for colorectal mucinous adenocarcinoma and non-mucinous adenocarcinoma using a combination of contrast-enhanced CT radiomics and deep learning methods.Methods:This is a retrospective cohort study. Clinical data of colorectal cancer patients confirmed by postoperative pathological examination were retrospectively collected from January 2016 to December 2023 at Shanghai General Hospital Affiliated to Shanghai Jiao Tong University School of Medicine (Center 1, n=220) and the First Affiliated Hospital of Bengbu Medical University (Center 2, n=51). Among them, there were 108 patients diagnosed with mucinous adenocarcinoma, including 55 males and 53 females, with an age of (68.4±12.2) years (range: 38 to 96 years); and 163 patients diagnosed with non-mucinous adenocarcinoma, including 96 males and 67 females, with an age of (67.9±11.0) years (range: 43 to 94 years). The cases from Center 1 were divided into a training set ( n=156) and an internal validation set ( n=64) using stratified random sampling in a 7∶3 ratio, and the cases from Center 2 were used as an independent external validation set ( n=51). Three-dimensional tumor volume of interest was manually segmented on venous-phase contrast-enhanced CT images. Radiomics features were extracted using PyRadiomics, and deep learning features were extracted using the ResNet-18 network. The two sets of features were then combined to form a joint feature set. The consistency of manual segmentation was assessed using the intraclass correlation coefficient. Feature dimensionality reduction was performed using the Mann-Whitney U test and the least absolute shrinkage and selection operator regression. Six machine learning algorithms were used to construct models based on radiomics features, deep learning features, and combined features, including support vector machine, logistic regression, random forest, extreme gradient boosting, k-nearest neighbors, and decision tree. The discriminative performance of each model was evaluated using receiver operating characteristic curves, the area under the curve (AUC), DeLong test, and decision curve analysis. Results:After feature selection, 22 features with the most discriminative value were finally retained, among which 12 were traditional radiomics features and 10 were deep learning features. In the internal validation set, the Random Forest algorithm based on the combined features model achieved the best performance (AUC=0.938, 95% CI: 0.875 to 0.984), which was superior to the single-modality radiomics feature model (AUC=0.817, 95% CI: 0.702 to 0.913, P=0.048) and the deep learning feature model (AUC=0.832, 95% CI: 0.727 to 0.926, P=0.087); in the independent external validation set, the Random Forest algorithm with the combined features model maintained the highest discriminative performance (AUC=0.891, 95% CI: 0.791 to 0.969), which was superior to the single-modality radiomics feature model (AUC=0.770, 95% CI: 0.636 to 0.890, P=0.045) and the deep learning feature model (AUC=0.799, 95% CI: 0.652 to 0.911, P=0.169). Conclusion:The combined model based on radiomics and deep learning features from venous-phase enhanced CT demonstrates good performance in the preoperative differentiation of colorectal mucinous from non-mucinous adenocarcinoma.

Ferroptosis is a new form of programmed cell death with iron-dependent lipid peroxidation as its core.A variety of metabolites are involved in the regulation of ferroptosis,among which lipid metabo-lism plays an important role.The most classic lipid metabolism-mediated ferroptosis mechanism is that phospholipids containing polyunsaturated fatty acyl chain(PUFA-PLs)located on biological membranes undergo super-threshold peroxidation,which leads to the destruction of membrane structure and function.In addition,special lipids containing polyunsaturated fatty acid(PUFA),such as phospholipid with dia-cyl-PUFA tails(PL-PUFA2),polyunsaturated ether phospholipid(PUFA-ePL),cholesterol ester con-taining polyunsaturated fatty acyl chain(PUFA-CE)have also been found to be involved in the ferropto-sis process by providing PUFA for peroxidation.Lipid droplets was also found to regulate the sensitivity of ferroptosis through storing and releasing PUFA.Intermediates and derivatives of cholesterol metabolism are mainly involved in the negative regulation of ferroptosis,whereas different classes of sphingolipids were reported to have inconsistent regulatory directions for ferroptosis.A large number of previous studies have confirmed that ferroptosis is closely related to the metastasis and drug resistance of gastrointestinal tumors,therefore,we further summarized the lipid metabolism mechanisms related to ferroptosis resist-ance in gastrointestinal tumor cells,such as weakening the anabolism and peroxidation processes of PU-FA-PLs,enhancing the ferroptosis defense system and so on.At the same time,we elaborated on the re-lationship between cholesterol metabolism,lipid drop metabolism,sphingolipid metabolism,and ferropto-sis resistance in gastrointestinal tumors.Targeting these specific lipids,metabolic enzymes,and pathways to regulate ferroptosis has important clinical potential value.It is expected to provide new ideas for finding new diagnostic and prognostic markers,therapeutic drugs,and reversing chemotherapy resistance for gas-trointestinal tumors.

Objective To determine the knee joint cartilage thickness using different methods and explore the feasibility of mathematical statistical models of dataset for the prediction of cartilage thickness.Methods A total of 304 patients diagnosed as knee osteoarthritis(OA)combined with varus deformity and undergoing unilateral total knee arthroplasty at the Second Affiliated Hospital of Army Medical University from March 2023 to March 2024 were selected for the study.All patients had complete preoperative and postoperative clinical data.The healthy cartilage at four anatomical sites of patients,including the distal femur lateral condyle,lateral tibial plateau,posterior medial femoral condyle,and posterior lateral femoral condyle were selected,and the knee joint cartilage thickness was determined based on preoperative MRI analysis,robotic navigation system tracing,tissue section of surgical specimen and digital vernier caliper.The baseline indicators of demographics,disease and imaging ffor patients were collected to construct a dataset,and four models of linear regression analysis,principal component analysis,Least Absolute Shrinkage and Selection Operator(LASSO)regression analysis,and K-nearest neighbors(KNN)analysis were established for predicting the accuracy,determination coefficient(R2)and root mean square error(RMSE),and the regression equation for predicting cartilage thickness was established.Results The knee joint cartilage thicknesses determined by preoperative MRI analysis,robotic navigation system tracing,tissue section of surgical specimen had no statistically significant difference with that by digital vernier caliper(P>0.05).The predictive efficiencies of models of linear regression analysis,principal component analysis,and LASSO regression analysis for the knee joint cartilage thickness all failed to meet the expectations(R2<0.3,RMSE>0.03).The predictive effect of KNN model on the cartilage thickness of the distal femur lateral condyle and lateral tibial plateau was not ideal(R2=0.23,RMSE=0.29),while it had potential predictive value(accuracy=0.21,accuracy=0.15).Conclusion The prediction model of knee joint cartilage thickness based on individual parameters has certain scientificity,and the feasibility of KNN model is relatively high.However,due to insufficient sample size and unclear individual parameter weight,the efficiencies of the four established prediction models are not ideal,which fails to provide definite prediction equations.Therefore,the construction scheme of the prediction model still needs to be further optimized.

Ferroptosis is a new form of programmed cell death with iron-dependent lipid peroxidation as its core.A variety of metabolites are involved in the regulation of ferroptosis,among which lipid metabo-lism plays an important role.The most classic lipid metabolism-mediated ferroptosis mechanism is that phospholipids containing polyunsaturated fatty acyl chain(PUFA-PLs)located on biological membranes undergo super-threshold peroxidation,which leads to the destruction of membrane structure and function.In addition,special lipids containing polyunsaturated fatty acid(PUFA),such as phospholipid with dia-cyl-PUFA tails(PL-PUFA2),polyunsaturated ether phospholipid(PUFA-ePL),cholesterol ester con-taining polyunsaturated fatty acyl chain(PUFA-CE)have also been found to be involved in the ferropto-sis process by providing PUFA for peroxidation.Lipid droplets was also found to regulate the sensitivity of ferroptosis through storing and releasing PUFA.Intermediates and derivatives of cholesterol metabolism are mainly involved in the negative regulation of ferroptosis,whereas different classes of sphingolipids were reported to have inconsistent regulatory directions for ferroptosis.A large number of previous studies have confirmed that ferroptosis is closely related to the metastasis and drug resistance of gastrointestinal tumors,therefore,we further summarized the lipid metabolism mechanisms related to ferroptosis resist-ance in gastrointestinal tumor cells,such as weakening the anabolism and peroxidation processes of PU-FA-PLs,enhancing the ferroptosis defense system and so on.At the same time,we elaborated on the re-lationship between cholesterol metabolism,lipid drop metabolism,sphingolipid metabolism,and ferropto-sis resistance in gastrointestinal tumors.Targeting these specific lipids,metabolic enzymes,and pathways to regulate ferroptosis has important clinical potential value.It is expected to provide new ideas for finding new diagnostic and prognostic markers,therapeutic drugs,and reversing chemotherapy resistance for gas-trointestinal tumors.

Objective:To develop a preoperative differentiation model for colorectal mucinous adenocarcinoma and non-mucinous adenocarcinoma using a combination of contrast-enhanced CT radiomics and deep learning methods.Methods:This is a retrospective cohort study. Clinical data of colorectal cancer patients confirmed by postoperative pathological examination were retrospectively collected from January 2016 to December 2023 at Shanghai General Hospital Affiliated to Shanghai Jiao Tong University School of Medicine (Center 1, n=220) and the First Affiliated Hospital of Bengbu Medical University (Center 2, n=51). Among them, there were 108 patients diagnosed with mucinous adenocarcinoma, including 55 males and 53 females, with an age of (68.4±12.2) years (range: 38 to 96 years); and 163 patients diagnosed with non-mucinous adenocarcinoma, including 96 males and 67 females, with an age of (67.9±11.0) years (range: 43 to 94 years). The cases from Center 1 were divided into a training set ( n=156) and an internal validation set ( n=64) using stratified random sampling in a 7∶3 ratio, and the cases from Center 2 were used as an independent external validation set ( n=51). Three-dimensional tumor volume of interest was manually segmented on venous-phase contrast-enhanced CT images. Radiomics features were extracted using PyRadiomics, and deep learning features were extracted using the ResNet-18 network. The two sets of features were then combined to form a joint feature set. The consistency of manual segmentation was assessed using the intraclass correlation coefficient. Feature dimensionality reduction was performed using the Mann-Whitney U test and the least absolute shrinkage and selection operator regression. Six machine learning algorithms were used to construct models based on radiomics features, deep learning features, and combined features, including support vector machine, logistic regression, random forest, extreme gradient boosting, k-nearest neighbors, and decision tree. The discriminative performance of each model was evaluated using receiver operating characteristic curves, the area under the curve (AUC), DeLong test, and decision curve analysis. Results:After feature selection, 22 features with the most discriminative value were finally retained, among which 12 were traditional radiomics features and 10 were deep learning features. In the internal validation set, the Random Forest algorithm based on the combined features model achieved the best performance (AUC=0.938, 95% CI: 0.875 to 0.984), which was superior to the single-modality radiomics feature model (AUC=0.817, 95% CI: 0.702 to 0.913, P=0.048) and the deep learning feature model (AUC=0.832, 95% CI: 0.727 to 0.926, P=0.087); in the independent external validation set, the Random Forest algorithm with the combined features model maintained the highest discriminative performance (AUC=0.891, 95% CI: 0.791 to 0.969), which was superior to the single-modality radiomics feature model (AUC=0.770, 95% CI: 0.636 to 0.890, P=0.045) and the deep learning feature model (AUC=0.799, 95% CI: 0.652 to 0.911, P=0.169). Conclusion:The combined model based on radiomics and deep learning features from venous-phase enhanced CT demonstrates good performance in the preoperative differentiation of colorectal mucinous from non-mucinous adenocarcinoma.

Objective To calculate the leakage rate of the automated watering system in Tsinghua University Laboratory Animal Resources Center, to evaluate the safety of the system, and provide references for selection, maintenance, and management of automated watering systems in animal facilities.MethodsThis study investigated the automated watering system installed in South and North Barriers of Tsinghua University Laboratory Animal Resources Center (Phase II). Water leakage monitoring was conducted over two periods, one over a period of 3 years and the other over 1.5 years. The occurrence of water leakage events at the two barriers during the monitoring period was statistically analyzed, classifying the causes into four categories: mishandling by personnel, animal behavior, obstruction by foreign objects, and deformation of fittings. The total daily leakage rate due to these causes and the daily leakage rate caused by quality issues, i.e. obstruction by foreign objects and deformation of fittings were calculated. Further analysis and discussion focused on the causes of water leakage and its impact on the facilities. At the same time, the number of caretakers at the end of the monitoring period in the Phase I facility without automated watering system and the Phase II facility with automated watering system were counted. Finally the difference in the number of cages per capita under the two watering systems was calculated.ResultsA total of 52 water leakage incidents occurred in both areas during the monitoring period, with a total daily leakage rate of 0.000 13%. Among them, 31 were caused by personnel mishandling, accounting for approximately 60% of total leakages. Enhanced training, supervision, inspection, and effective reminder measures could reduce leakage caused by personnel mishandling. There were 2 cases of water leakages caused by animal behavior, 0 leakage due to obstruction by foreign objects, and 19 leakages due to system quality issues, with a daily leakage rate of 0.000 07%. According to the operation data of Tsinghua University Laboratory Animal Resources Center, the average number of cages managed per person in facilities equipped with the automated watering system was 908, compared to 570 cages in facilities without the automated watering system. This represents an approximate 59% increase in the number of cages managed per person with the adoption of the automated watering system.Conclusion The daily leakage rate of the automated watering system in the Tsinghua University Laboratory Animal Resources Center is significantly lower than the theoretical design rate of 0.003%, which demonstrates the system's safety and effectiveness. Additionally, the adoption of an automated watering system can significantly enhance caretaking efficiency. While initial investments in the system are required, the subsequent increase in efficiency leads to a continuous decrease in labor costs, thereby reducing the total operational expenses of the facility. In the context of modernizing animal facility construction, automated watering systems are becoming an essential consideration in facility design and operation.

Circular RNAs(CircRNAs)are a class of non-coding RNAs with a covalently closed-loop structure,high stability,and tissue specificity,with the production mechanisms different from linear RNAs.Recent studies have discovered that some CircRNAs can encode proteins via cap-independent translation mechanisms such as internal ribosome entry site,N6-methyladenosine,and rolling loop translation.The encoded proteins regulate homologous linear proteins or downstream signaling pathways via protein bait or other mecha-nisms,thereby exerting biological functions.Studies have shown that CircRNAs play a role in various diseases,especially in tumor progression,proliferation,invasion,and metastasis and immune regulation.Therefore,by elucidating the expression and roles of proteins encoded by CircRNAs in tumorigenesis and development,this pa-per is expected to provide new tumor markers and potential targets for tumor diagnosis and treatment.

Due to the limitations of current anti-HBV therapies, the HBV core (HBc or HBcAg) protein assembly modulators (CpAMs) are believed to be potential anti-HBV agents. Therefore, discovering safe and efficient CpAMs is of great value. In this study, we established a HiBiT-based high-throughput screening system targeting HBc and screened novel CpAMs from an in-house marine chemicals library. A novel lead compound 8a, a derivative of the marine natural product naamidine J, has been successfully screened for potential anti-HBV activity. Bioactivity-driven synthesis was then conducted, and the structure‒activity relationship was analyzed, resulting in the discovery of the most effective compound 11a (IC₅₀ = 0.24 μmol/L). Furthermore, 11a was found to significantly inhibit HBV replication in multiple cell models and exhibit a synergistic effect with tenofovir disoproxil fumarate (TDF) and IFNa2 in vitro for anti-HBV activity. Treatment with 11a in a hydrodynamic-injection mouse model demonstrated significant anti-HBV activity without apparent hepatotoxicity. These findings suggest that the naamidine J derivative 11a could be used as the HBV core protein assembly modulator to develop safe and effective anti-HBV therapies.