An extensive benchmark study on biomedical text generation and mining with ChatGPT.

MOTIVATION: In recent years, the development of natural language process (NLP) technologies and deep learning hardware has led to significant improvement in large language models (LLMs). The ChatGPT, the state-of-the-art LLM built on GPT-3.5 and GPT-4, shows excellent capabilities in general language understanding and reasoning. Researchers also tested the GPTs on a variety of NLP-related tasks and benchmarks and got excellent results. With exciting performance on daily chat, researchers began to explore the capacity of ChatGPT on expertise that requires professional education for human and we are interested in the biomedical domain. RESULTS: To evaluate the performance of ChatGPT on biomedical-related tasks, this article presents a comprehensive benchmark study on the use of ChatGPT for biomedical corpus, including article abstracts, clinical trials description, biomedical questions, and so on. Typical NLP tasks like named entity recognization, relation extraction, sentence similarity, question and answering, and document classification are included. Overall, ChatGPT got a BLURB score of 58.50 while the state-of-the-art model had a score of 84.30. Through a series of experiments, we demonstrated the effectiveness and versatility of ChatGPT in biomedical text understanding, reasoning and generation, and the limitation of ChatGPT build on GPT-3.5. AVAILABILITY AND IMPLEMENTATION: All the datasets are available from BLURB benchmark https://microsoft.github.io/BLURB/index.html. The prompts are described in the article.

A general hypergraph learning algorithm for drug multi-task predictions in micro-to-macro biomedical networks.

The powerful combination of large-scale drug-related interaction networks and deep learning provides new opportunities for accelerating the process of drug discovery. However, chemical structures that play an important role in drug properties and high-order relations that involve a greater number of nodes are not tackled in current biomedical networks. In this study, we present a general hypergraph learning framework, which introduces Drug-Substructures relationship into Molecular interaction Networks to construct the micro-to-macro drug centric heterogeneous network (DSMN), and develop a multi-branches HyperGraph learning model, called HGDrug, for Drug multi-task predictions. HGDrug achieves highly accurate and robust predictions on 4 benchmark tasks (drug-drug, drug-target, drug-disease, and drug-side-effect interactions), outperforming 8 state-of-the-art task specific models and 6 general-purpose conventional models. Experiments analysis verifies the effectiveness and rationality of the HGDrug model architecture as well as the multi-branches setup, and demonstrates that HGDrug is able to capture the relations between drugs associated with the same functional groups. In addition, our proposed drug-substructure interaction networks can help improve the performance of existing network models for drug-related prediction tasks.

Unraveling the molecular mechanisms driving enhanced invasion capability of extravillous trophoblast cells: a comprehensive review.

Precise extravillous trophoblast (EVT) invasion is crucial for successful placentation and pregnancy. This review focuses on elucidating the mechanisms that promote heightened EVT invasion. We comprehensively summarize the pivotal roles of hormones, angiogenesis, hypoxia, stress, the extracellular matrix microenvironment, epithelial-to-mesenchymal transition (EMT), immunity, inflammation, programmed cell death, epigenetic modifications, and microbiota in facilitating EVT invasion. The molecular mechanisms underlying enhanced EVT invasion may provide valuable insights into potential pathogenic mechanisms associated with diseases characterized by excessive invasion, such as the placenta accreta spectrum (PAS), thereby offering novel perspectives for managing pregnancy complications related to deficient EVT invasion.

A meta-analysis of treatment for early-stage cervical cancer: open versus minimally invasive radical trachelectomy.

BACKGROUND: In previous systematic reviews, meta-analysis was lacking, resulting in the statistical difference between the data of different surgeries being impossible to judge. This meta-analysis aims to contrast the fertility results and cancer outcomes between open and minimally invasive surgery. METHOD: We systematically searched databases including PubMed, Embase, Cochrane, and Scopus to collect studies that included open and minimally invasive radical trachelectomy. A random-effect model calculated the weighted average difference of each primary outcome via Review Manager V.5.4. RESULT: Eight studies (1369 patients) were incorporated into our study. For fertility results, the Open group excels MIS group in pregnancies-Third trimester delivery [OR = 2.68; 95% CI (1.29, 5.59); P = 0.008]. Nevertheless, there is no statistical difference in clinical pregnancy, miscarriage, and second-trimester rate. Concerning cancer outcomes, no difference was detected in the overall survival [OR = 1.56; 95% CI (0.70, 3.45); P = 0.27] and recurrence [OR = 0.63; 95% CI (0.35, 1.12); P = 0.12]. Concerning surgery-related outcomes, the comprehensive effects revealed that the estimated blood loss of the Open group was higher than that of the MIS group[MD = 139.40; 95% CI (79.05, 199.75); P < 0.0001]. However, there was no difference between the postoperative complication rate in the two groups [OR = 1.52; 95% CI (0.89, 2.60); P = 0.12]. CONCLUSION: This meta-analysis suggested that the fertility result of the Open group may be better than the MIS group, while the MIS group has better surgery-related outcomes. Owing to the poor cases of our study, a more robust conclusion requires more relevant articles in the future. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42022352999.

Effects of high manganese-cultivated seedlings on cadmium uptake by various rice (Oryza sativa L.) genotypes.

Cadmium (Cd) contamination in paddy soil threatens rice growth and food safety, enriching manganese (Mn) in rice seedlings is expected to reduce Cd uptake by rice. The effects of 250 µM Mn-treated seedlings on reducing Cd uptake of four rice genotypes (WYJ21, ZJY1578, HHZ, and HLYSM) planted in 0.61 mg kg-1 Cd-contaminated soil, were studied through the hydroponic and pot experiments. The results showed that the ZJY1578 seedling had the highest Mn level (459 µg plant-1), followed by WYJ21 (309 µg plant-1), and less Mn accumulated in the other genotypes. The relative expression of OsNramp5 (natural resistance-associated macrophage protein) was reduced by 42.7 % in ZJY1578 but increased by 23.3 % in HLYSM. The expressions of OsIRT1 (iron-regulated transporter-like protein) were reduced by 24.0-56.0 % in the four genotypes, with the highest reduction in ZJY1578. Consequently, a greater reduction of Cd occurred in ZJY1578 than that in the other genotypes, i.e., the root and shoot Cd at the tillering were reduced by 27.8 % and 48.5 %, respectively. At the mature stage, total Cd amount and distribution in the shoot and brown rice were also greatly reduced in ZJY1578, but the inhibitory effects were weakened compared to the tillering stage. This study found various responses of Cd uptake and transporters to Mn-treated seedlings among rice genotypes, thus resulting in various Cd reductions. In the future, the microscopic transport processes of Cd within rice should be explored to deeply explain the genotypic variation.

The temporal response of dissolved heavy metals to landscape indices in the Le'an river, China.

Heavy metals in watersheds are a serious concern due to their toxicity, abundance, and persistence in the environment, especially in mining areas. Source analyses and exploration of other related factors are one of the most important methods to help with effective prevention and control of heavy metal pollution in watersheds. In this study, the concentrations of Cr, Co, Ni, Cu, Zn, As, Cd, Sb, Ba and Pb were measured in the Le'an River, and PCA (principal component analysis) and APCS-MLR (absolute principal component scores - multivariate linear regression) methods were used to identify the sources of the dissolved heavy metals. Additionally, a CA (correlation analysis) method was used to explore the correlations between landscape indices and concentrations of heavy metals. Results show that the main sources for these dissolved heavy metals are mining activities, fertilizers, pesticides, and natural sources. Specific results of PCA and APCS-MLR suggest that Cu, Zn, Cd, Ba are mainly related to mining activities, Cr and Pb are due to fertilizers and pesticides, and Co and Ni are mainly due to natural sources. Correlations between landscapes and heavy metals revealed significant temporal variations, with the strongest responses of dissolved heavy metals to landscape indices appearing in December and March. The propensity of positive or negative responses of the heavy metals to landscape indices are determined by the sources, and their temporal variations may be related to the seasonal changes of rainfall and plant metabolism.

Robust weakly supervised learning for COVID-19 recognition using multi-center CT images.

The world is currently experiencing an ongoing pandemic of an infectious disease named coronavirus disease 2019 (i.e., COVID-19), which is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Computed Tomography (CT) plays an important role in assessing the severity of the infection and can also be used to identify those symptomatic and asymptomatic COVID-19 carriers. With a surge of the cumulative number of COVID-19 patients, radiologists are increasingly stressed to examine the CT scans manually. Therefore, an automated 3D CT scan recognition tool is highly in demand since the manual analysis is time-consuming for radiologists and their fatigue can cause possible misjudgment. However, due to various technical specifications of CT scanners located in different hospitals, the appearance of CT images can be significantly different leading to the failure of many automated image recognition approaches. The multi-domain shift problem for the multi-center and multi-scanner studies is therefore nontrivial that is also crucial for a dependable recognition and critical for reproducible and objective diagnosis and prognosis. In this paper, we proposed a COVID-19 CT scan recognition model namely coronavirus information fusion and diagnosis network (CIFD-Net) that can efficiently handle the multi-domain shift problem via a new robust weakly supervised learning paradigm. Our model can resolve the problem of different appearance in CT scan images reliably and efficiently while attaining higher accuracy compared to other state-of-the-art methods.

Marine aquaculture regulates dissimilatory nitrate reduction processes in a typical semi-enclosed bay of southeastern China.

Marine aquaculture in semi-enclosed bays can significantly influence nutrient cycling in coastal ecosystems. However, the impact of marine aquaculture on the dynamics of dissimilatory nitrate reduction processes (DNRPs) and the fate of reactive nitrogen remain poorly understood. In this study, the rates of DNRPs and the abundances of related functional genes were investigated in aquaculture and non-aquaculture areas. The results showed that marine aquaculture significantly increased the denitrification (DNF) and dissimilatory nitrate reduction to ammonium (DNRA) rates and decreased the rate of anaerobic ammonium oxidation (ANA), as compared with non-aquaculture sites. DNF was the dominant pathway contributing to the total nitrate reduction, and its contribution to the total nitrate reduction significantly increased from 66.72% at non-aquaculture sites to 78.50% at aquaculture sites. Marine aquaculture can significantly affect the physicochemical characteristics of sediment and the abundances of related functional genes, leading to variations in the nitrate reduction rates. Although nitrate removal rates increased in the marine aquaculture area, ammonification rates and the nitrogen retention index in the aquaculture areas were 2.19 and 1.24 times, respectively, higher than those at non-aquaculture sites. Net reactive nitrogen retention exceeded nitrogen removal in the aquaculture area, and the retained reactive nitrogen could diffuse with the tidal current to the entire bay, thereby aggravating N pollution in the entire study area. These results show that marine aquaculture is the dominant source of nitrogen pollution in semi-enclosed bays. This study can provide insights into nitrogen pollution control in semi-enclosed bays with well-developed marine aquaculture.

Inherent bacterial community response to multiple heavy metals in sediment from river-lake systems in the Poyang Lake, China.

Sediment is the one of most important storage of heavy metal. Microbiotas in sediment can be used as the effective indicators of heavy metals. The goal of this study was to understand the bacterial communities responding to heavy metal enrichment in sediments and prioritize some factors that affected significantly to bacterial community. Sediments were sampled from five river-lake systems in the Poyang Lake in dry season, and the bacterial community was analyzed using Illumina high-throughput sequencing. Relationships between sediment environment and the diversity and structure of bacterial communities were determined by correlation analysis and redundancy analysis (RDA). The result indicated that Cd and Sb were identified as the heavy metals of the great risk in sediments. Sediments from five river-lake systems shared 31.83% core operational taxonomic units (OTUs) of bacterial communities. Proteobacteria (33.54% of total sequences) and Actinobacteria (15.04%) were the dominant phyla across all sites. High enrichment of heavy metals (MRI and mCd) resulted in low diversity of bacterial communities (Simpson index). The RDA revealed pH, OC, mCd, and Efs of As, Pb, Cd were major factors related to bacterial community structure changes. The dominant phylum Actinobacteria was regarded as tolerant bacteria, while the dominant phylum Proteobacteria was named as resistant bacteria in sediment with high anthropogenic Cd enrichment.

Effects of land use types on dissolved trace metal concentrations in the Le'an River Basin, China.

Using land use types in multiple spatial scales (entire basin, buffer zones, and slopes) as well as statistical and spatial analysis, relationships between land use types and concentrations of dissolved trace metals were determined in the Le'an River Basin, China. The result showed that farmland and urban land were determined as the source of the pollutants, while forestland and grassland were identified as the sink of the pollutants. The temporal differences of relationships between land use types and concentrations of dissolved trace metals mainly due to the discrepancy of rainfall characteristics. Land use type close to river was a better indicator for the effectiveness of concentrations of trace metals, especially at scale of 0-200 m. Forestland and grassland on lower slopes greatly affected the water quality, and the former had no significant or weak influences on higher slopes. Urban land had the greater positive correlations with concentrations of dissolved trace metals on higher slopes, which are mainly due to frequent mining activity. Further analysis suggested that the buffer zones with low slope needed to be seriously taken into consideration for effective land use management in similar basin.

Strain-level visualized analysis of cold-stressed Vibrio parahaemolyticus based on MALDI-TOF mass fingerprinting.

In this study, strain-level visualized analysis of cold-stressed Vibrio parahaemolyticus based on matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass fingerprinting was investigated. All the peptide mass fingerprinting profiles obtained were analyzed by self-organized map (SOM) and cluster analysis. Our results showed that the peptide mass fingerprinting profiles of V. parahaemolyticus substantially changed under cold stress at strain level. The cold-stressed V. parahaemolyticus strains were distributed to 14 neurons by SOM classification, almost totally different from the controls. This is the first time that so many strains had been chosen to study bacterial cold stress responses, which can help promote an overall understanding to stress responses of cold-stressed strains.

Simultaneous quantification of Escherichia coli O157:H7 and Shigella boydii using a visual-antibody-macroarray.

Being high throughput, rapid, automated, economical, convenient to operate and highly sensitive, protein arrays have been widely used in the analysis of tumor markers and veterinary drug residues. Pathogenic microbes also can be detected qualitatively by DNA array or protein array; however, their high throughput detection and quantification remains a big obstacle. To evaluate the potentiality of protein arrays for multiple quantitative detection of microorganisms with naked eye examination without the help of any equipment, here we developed a visual-antibody-macroarray (VAMA) aiming at rapid and simultaneous quantification of Escherichia coli O157:H7 and Shigella boydii. The results show that this VAMA is highly specific and is able to distinguish mixed Escherichia coli O157:H7 and Shigella boydii synchronously. The detection limits are equivalent to 3.4 × 10(5) CFU mL(-1) and 3.2 × 10(5) CFU mL(-1), respectively, which conform to the results of plate counting and ELISA. Importantly, the examination can be solely performed with the naked eye. Therefore, we provide an easy, reliable and rapid method for quantitative analysis of microorganisms.

Automated molecular structure segmentation from documents using ChemSAM.

Chemical structure segmentation constitutes a pivotal task in cheminformatics, involving the extraction and abstraction of structural information of chemical compounds from text-based sources, including patents and scientific articles. This study introduces a deep learning approach to chemical structure segmentation, employing a Vision Transformer (ViT) to discern the structural patterns of chemical compounds from their graphical representations. The Chemistry-Segment Anything Model (ChemSAM) achieves state-of-the-art results on publicly available benchmark datasets and real-world tasks, underscoring its effectiveness in accurately segmenting chemical structures from text-based sources. Moreover, this deep learning-based approach obviates the need for handcrafted features and demonstrates robustness against variations in image quality and style. During the detection phase, a ViT-based encoder-decoder model is used to identify and locate chemical structure depictions on the input page. This model generates masks to ascertain whether each pixel belongs to a chemical structure, thereby offering a pixel-level classification and indicating the presence or absence of chemical structures at each position. Subsequently, the generated masks are clustered based on their connectivity, and each mask cluster is updated to encapsulate a single structure in the post-processing workflow. This two-step process facilitates the effective automatic extraction of chemical structure depictions from documents. By utilizing the deep learning approach described herein, it is demonstrated that effective performance on low-resolution and densely arranged molecular structural layouts in journal articles and patents is achievable.

Proteomic analysis of protein expression in the induction of the viable but Nonculturable State of Vibrio harveyi SF1.

Vibrio harveyi has been reported to enter into a viable but nonculturable (VBNC) state. One marine V. harveyi strain, SF1 became nonculturable when incubated in seawater microcosm at 4 °C within 60 days. We investigated protein expression in the exponential phase of V. harveyi SF1 and compared it to the VBNC state. Cytosolic proteins were resolved by two-dimensional polyacrylamide gel electrophoresis using pH 4-7 linear gradients. Among these proteins, sixteen proteins which were strongly downregulated or upregulated in the VBNC cells were identified by MALDI-TOF-TOF mass spectrometry. The results indicated that the differentially expressed proteins were mainly focused on stress response proteins and key components of central and intermediary metabolism, like carbohydrate metabolism, transport, and translation. This study provided clues for understanding the mechanism of adaptation to the VBNC state.

[Identification of Vibrio metschnikovii from Homarus americanus and its changes in membrane fatty acid composition in response to low temperature].

OBJECTIVE: Bacterial strain F5-1 isolated from the Homarus americanus was characterized and its changes in membrane fatty acid composition in response to low temperature were also studied. METHODS: The physiological and biochemical characteristics were carried out by using VITEK 2 compact automated microbiology system. The 16S rRNA gene was sequenced and subjected to phylogenetic analysis. Fatty acids were detected by gas chromatography-mass spectrometry (GC-MS). RESULTS: Strain F5-1 was Gram-negative and susceptible to the vibriostatic agent O/129. Strain F5-1 was resistant to Penicillin. The isolated strain exhibited the highest levels of 99% probability to Vibrio metschnikovii based on the conventional physiological test. The sequence analysis of 16S rRNA gene of F5-1 isolation and comparison with that of other related vibrios showed that F5-1 was very close to V. metschnikovii (GenBank No. HQ658055). The similarity was 99%. The major fatty acids were C12:0, C14:0, C16:0 and C16:1 (n-7). Palmitoleic acid was the dominant unsaturated fatty acids. The major change in fatty acid composition occurred in response to low temperature, with an increase in palmitoleic acid from 34% to 40%. CONCLUSION: Bacterial strain F5-1 isolated from Homarus americanus was identified as V. metschnikovii and was sensitive to multiple drugs. The fatty acid composition of F5-1 was different from V. metschnikovii isolated from a drinking water reservoir near Vladivostok City in the Russia Far East. Results of this study indicated that environmental conditions allowed modulation of the fatty acid composition of V. metschnikovii.

[Analysis of the positive rate of otitis media in 1-12 years old children based on brain MRI].

Objective:To investigate the positive imaging rate of otitis media in children aged 1-12 years by analyzing the positive rate of otitis media suggested by cranial magnetic resonance imaging（MRI） images in children. Methods:By collecting the brain MRI images of children aged 1-12 in Department of Otolaryngology, Jinan children's Hospital from January 2014 to December 2020, the overall incidence of otitis media and mastoiditis was firstly determined, and then it was divided into 12 age groups according to age, each age group was split into boy and girl groups according to gender, each group was divided into left, right and bilateral groups, with the changes of otitis media and mastoiditis in the scanning field as the positive standard statistical analysis of the results. Results:Among 12 439 children in the study, 1321 cases were diagnosed with tympanitis, with a positive rate of 10.62%. Among them, 892 patients were male, with a positive rate of 67.52%, and 429 cases were female, with a positive rate of 32.48%. The positive rate of the male was higher than that of female children, 1.84 times higher than that of female children. The positive momentum in male and female children was negatively correlated with age （P<0.01）. The favorable rates of male children in the left ear, right ear, and both ears were higher than those in female children of the same age（P<0.05）. There was no difference in the positive rate of the left and right ear in children with tympanitis （P=0.76）. Conclusion:Craniocerebral MRI examination in children with tympanitis can clarify the inflammation of the middle ear cavity and the effusion in the mastoid air chamber. The positive rate of children at two years old showed a steep decline, which may be due to the acceleration of mastoid gasification, the change of tympanic environment, the increase of air chamber in the tympanic room, the evolution of air pressure could offset the negative pressure caused by poor Eustachian tube function, to reduce the incidence of tympanitis.

Volatile fatty acids changed the microbial community during feammox in coastal saline-alkaline paddy soil.

In order to indicate the effect of volatile fatty acids (VFAs) on the characteristics of feammox and dissimilatory iron reducing bacteria (DIRB) in paddy soils, different VFAs were selected with paddy soils for anaerobic cultivation. Five treatments were set up, respectively, only adding N and both adding N and C (formate + NH4+ (Fo-N), acetate + NH4+ (Ac-N), propionate + NH4+ (Pr-N), and butyrate + NH4+ (Bu-N)) treatments. The concentration of Fe(II), Fe(III), NH4+, and VFAs was assessed within 45 d, and the bacterial community was determined after cultivation. The oxidation rates of NH4+ were the highest in N treatment, while it was the lowest in Fo-N treatment. Under the four C treatments, the consumption of NH4+ and Fe(III) was the fastest in Pr-N treatment, which was consumed by 31.2% and 76.3%, respectively. Different VFAs selected for distinct DIRB. Compared with N treatment, Ac-N and Bu-N treatment increased the relative abundance of DIRB, such as Geobacter and Clostridia, which increased the consumption of VFAs during incubation. Overall, VFAs, especially formate, could promote Fe(III) reduction and compete with the feammox process for the electron acceptors to decrease the feammox reaction, and prohibited soil NH4+ loss. Therefore, VFAs, which was released from organic fertilizer, could reduce NH4+ loss in feammox process of saline-alkaline paddy soils.

Serum PCSK9 is positively correlated with disease activity and Th17 cells, while its short-term decline during treatment reflects desirable outcomes in ankylosing spondylitis patients.

OBJECTIVE: Proprotein convertase subtilisin/kexin type 9 (PCSK9) participates in the autoimmune disease pathology by regulating T helper (Th) cell differentiation, NF-κB pathway, toll-like receptor 4, etc. This study intended to investigate the association of serum PCSK9 with disease activity, Th cells, and treatment response in ankylosing spondylitis (AS) patients. METHODS: Eighty-nine active AS patients were enrolled in this multicenter, prospective study. Serum was collected from AS patients at week (W)0, W4, W8, and W12, as well as from 20 osteoarthritis patients and 20 healthy controls after enrollment to detect PCSK9 by ELISA. Based on the ASAS40 response at W12, AS patients were classified as responders and non-responders. RESULTS: PCSK9 was increased in AS patients versus healthy controls (P < 0.001) and osteoarthritis patients (P = 0.006). In AS patients, PCSK9 was positively linked with C-reactive protein (CRP) (P = 0.003) and ASDAS-CRP (P = 0.017), but not with other clinical properties (P > 0.05). Besides, PCSK9 was negatively correlated with interleukin-4 (P = 0.034), positively associated with Th17 cells (P = 0.005) and interleukin-17A (P = 0.014), but did not relate to Th1 cells, interferon-γ, or Th2 cells (all P > 0.05). Additionally, PCSK9 was decreased from W0 to W12 in general AS patients (P < 0.001) and responders (P < 0.001) but remained unchanged in non-responders (P = 0.129). Moreover, PCSK9 was lower at W4 (P = 0.045), W8 (P = 0.008), and W12 (P = 0.004) in responders versus non-responders. Furthermore, the treatment options did not affect the PCSK9 level (P > 0.05). CONCLUSION: Serum PCSK9 is positively associated with disease activity and Th17 cells, while its short-term decline reflects desirable treatment response in AS patients.

PCSK9 promotes T helper 1 and T helper 17 cell differentiation by activating the nuclear factor-κB pathway in ankylosing spondylitis.

OBJECTIVE: Our previous study reveals that proprotein convertase subtilisin/kexin type 9 (PCSK9) is positively related to inflammatory markers, T helper (Th)-17 cells, and treatment response in ankylosing spondylitis (AS) patients. Subsequently, this study aimed to explore the effect of PCSK9 on Th cell differentiation and its potential molecular mechanism in AS. METHODS: Serum PCSK9 was determined by enzyme-linked immunosorbent assay in 20 AS patients and 20 healthy controls (HCs). Then naïve CD4+ T cells were isolated from AS patients and infected with PCSK9 overexpression or knockdown adenovirus followed by polarization assay. Afterward, PMA (an NF-κB activator) was administrated. RESULTS: PCSK9 was increased in AS patients compared to HCs (p < .001), and it was positively related to Th1 cells (p = .050) and Th17 cells (p = .039) in AS patients. PCSK9 overexpression increased the CD4+ IFN-γ+ cells (p < .05), CD4+ IL-17A+ cells (p < .01), IFN-γ (p < .01), and IL-17A (p < .01), while it exhibited no effect on CD4+ IL-4+ cells or IL-4 (both p > .05); its knockdown displayed the opposite function on them. Moreover, PCSK9 overexpression upregulated the p-NF-κB p65/NF-κB p65 (p < .01), while it had no effect on p-ERK/ERK or p-JNK/JNK (both p > .05); its knockdown decreased p-NF-κB p65/NF-κB p65 (p < .01) and p-JNK/JNK (p < .05). Then, PMA upregulates p-NF-κB p65/NF-κB p65 (p < .001) and increased CD4+ IFN-γ+ cells, CD4+ IL-17A+ cells, IFN-γ, and IL-17A (all p < .01), also it alleviated the effect of PCSK9 knockdown on NF-κB inhibition and Th cell differentiation (all p < .01). CONCLUSION: PCSK9 enhances Th1 and Th17 cell differentiation in an NF-κB-dependent manner in AS, while further validation is necessary.

Pharmacophoric-constrained heterogeneous graph transformer model for molecular property prediction.

Informative representation of molecules is a crucial prerequisite in AI-driven drug design and discovery. Pharmacophore information including functional groups and chemical reactions can indicate molecular properties, which have not been fully exploited by prior atom-based molecular graph representation. To obtain a more informative representation of molecules for better molecule property prediction, we propose the Pharmacophoric-constrained Heterogeneous Graph Transformer (PharmHGT). We design a pharmacophoric-constrained multi-views molecular representation graph, enabling PharmHGT to extract vital chemical information from functional substructures and chemical reactions. With a carefully designed pharmacophoric-constrained multi-view molecular representation graph, PharmHGT can learn more chemical information from molecular functional substructures and chemical reaction information. Extensive downstream experiments prove that PharmHGT achieves remarkably superior performance over the state-of-the-art models the performance of our model is up to 1.55% in ROC-AUC and 0.272 in RMSE higher than the best baseline model) on molecular properties prediction. The ablation study and case study show that our proposed molecular graph representation method and heterogeneous graph transformer model can better capture the pharmacophoric structure and chemical information features. Further visualization studies also indicated a better representation capacity achieved by our model.