Core Module Research Articles

AutoCore: A network-based definition of the core module of human autoimmunity and autoinflammation.

Although research on rare autoimmune and autoinflammatory diseases has enabled definition of nonredundant regulators of homeostasis in human immunity, because of the single gene-single disease nature of many of these diseases, contributing factors were mostly unveiled in sequential and noncoordinated individual studies. We used a network-based approach for integrating a set of 186 inborn errors of immunity with predominant autoimmunity/autoinflammation into a comprehensive map of human immune dysregulation, which we termed "AutoCore." The AutoCore is located centrally within the interactome of all protein-protein interactions, connecting and pinpointing multidisease markers for a range of common, polygenic autoimmune/autoinflammatory diseases. The AutoCore can be subdivided into 19 endotypes that correspond to molecularly and phenotypically cohesive disease subgroups, providing a molecular mechanism-based disease classification and rationale toward systematic targeting for therapeutic purposes. Our study provides a proof of concept for using network-based methods to systematically investigate the molecular relationships between individual rare diseases and address a range of conceptual, diagnostic, and therapeutic challenges.

ATAC and SAGA co-activator complexes utilize co-translational assembly, but their cellular localization properties and functions are distinct

To understand the function of multisubunit complexes, it is of key importance to uncover the precise mechanisms that guide their assembly. Nascent proteins can find and bind their interaction partners during their translation, leading to co-translational assembly. Here, we demonstrate that the core modules of ATAC (ADA-two-A-containing) and SAGA (Spt-Ada-Gcn5-acetyltransferase), two lysine acetyl transferase-containing transcription co-activator complexes, assemble co-translationally in the cytoplasm of mammalian cells. In addition, a SAGA complex containing all of its modules forms in the cytoplasm and acetylates non-histone proteins. In contrast, ATAC complex subunits cannot be detected in the cytoplasm of mammalian cells. However, an endogenous ATAC complex containing two functional modules forms and functions in the nucleus. Thus, the two related co-activators, ATAC and SAGA, assemble using co-translational pathways, but their subcellular localization, cytoplasmic abundance, and functions are distinct.

Human-technology integration with industrial conversational agents: A conceptual architecture and a taxonomy for manufacturing

Conversational agents are systems with great potential to enhance human-computer interaction in industrial settings. Although the number of applications of conversational agents in many fields is growing, there is no shared view of the elements to design and implement for chatbots in the industrial field. The paper presents the combination of many research contributions into an integrated conceptual architecture, for developing industrial conversational agents using Nickerson's methodology. The conceptual architecture consists of five core modules; every module consists of specific elements and approaches. Furthermore, the paper defines a taxonomy from the study of empirical applications of manufacturing conversational agents. Indeed, some applications of chatbots in manufacturing are available but those have never been collected in single research. The paper fills this gap by analyzing the empirical cases and presenting a qualitative analysis, with verification of the proposed taxonomy. The contribution of the article is mainly to illustrate the elements needed for the development of a conversational agent in manufacturing: researchers and practitioners can use the proposed conceptual architecture and taxonomy to more easily investigate, define, and develop all the elements for chatbot implementation.

Self-Reporting Theranostic: Nano Tool for Arterial Thrombosis.

Arterial thrombosis (AT) originates through platelet-mediated thrombus formation in the blood vessel and can lead to heart attack, stroke, and peripheral vascular diseases. Restricting the thrombus growth and its simultaneous monitoring by visualisation is an unmet clinical need for a better AT prognosis. As a proof-of-concept, we have engineered a nanoparticle-based theranostic (combined therapy and monitoring) platform that has the potential to monitor and restrain the growth of a thrombus concurrently. The theranostic nanotool is fabricated using biocompatible super-paramagnetic iron oxide nanoparticles (SPIONs) as a core module tethered with the anti-platelet agent Abciximab (ReoPro) on its surface. Our in vitro feasibility results indicate that ReoPro-conjugated SPIONS (Tx@ReoPro) can effectively prevent thrombus growth by inhibiting fibrinogen receptors (GPIIbIIIa) on the platelet surface, and simultaneously, it can also be visible through non-invasive magnetic resonance imaging (MRI) for potential reporting of the real-time thrombus status.

Identification of molecular subtypes and immune infiltration in endometriosis: a novel bioinformatics analysis and In vitro validation.

Endometriosis is a worldwide gynacological diseases, affecting in 6-10% of women of reproductive age. The aim of this study was to investigate the gene network and potential signatures of immune infiltration in endometriosis. The expression profiles of GSE51981, GSE6364, and GSE7305 were obtained from the Gene Expression Omnibus (GEO) database. Core modules and central genes related to immune characteristics were identified using a weighted gene coexpression network analysis. Bioinformatics analysis was performed to identify central genes in immune infiltration. Protein-protein interaction (PPI) network was used to identify the hub genes. We then constructed subtypes of endometriosis samples and calculated their correlation with hub genes. qRTPCR and Western blotting were used to verify our findings. We identified 10 candidate hub genes (GZMB, PRF1, KIR2DL1, KIR2DL3, KIR3DL1, KIR2DL4, FGB, IGFBP1, RBP4, and PROK1) that were significantly correlated with immune infiltration. Our study established a detailed immune network and systematically elucidated the molecular mechanism underlying endometriosis from the aspect of immune infiltration. Our study provides comprehensive insights into the immunology involved in endometriosis and might contribute to the development of immunotherapy for endometriosis. Furthermore, our study sheds light on the underlying molecular mechanism of endometriosis and might help improve the diagnosis and treatment of this condition.

AGPN: Action Granularity Pyramid Network for Video Action Recognition

Video action recognition is a fundamental task for video understanding. Action recognition in complex spatio-temporal contexts generally requires fusing of different multi-granularity action information. However, existing works do not consider spatio-temporal information modeling and fusion from the perspective of action granularity. To address this problem, this paper proposes an Action Granularity Pyramid Network (AGPN) for action recognition, which can be flexibly integrated into 2D backbone networks. The core module is the Action Granularity Pyramid Module (AGPM), a hierarchical pyramid structure with residual connections, which is established to fuse multi-granularity action spatio-temporal information. From top to bottom level in the designed pyramid structure, the receptive field decreases and action granularity becomes more refined. To enrich temporal information of the inputs, a Multiple Frame Rate Module (MFM) is proposed to mix different frame rates at a fine-grained pixel-wise level. Moreover, a Spatio-temporal Anchor Module (SAM) is employed to fix spatio-temporal feature anchors to promote the effectiveness of feature extraction. We conduct extensive experiments on three large-scale action recognition datasets, Something-Something V1 & V2 and Kinetics-400. The results demonstrate that our proposed AGPN outperforms the state-of-the-art methods for the tasks of video action recognition.

Comprehensive Analysis to Identify Rh Family C Glycoprotein (RHCG) as the Causative Gene for Psoriasis and Search for Alternative Treatment Modalities.

Psoriasis is a complex autoimmune disease. Frequent interactions between epidermal and immune cells are likely to be responsible for the strong heterogeneity of psoriasis. Therefore, our work aims to build on current knowledge and further search for new molecular mechanisms related to psoriasis pathogenesis in order to develop new targeted drugs. Data from psoriasis samples were obtained from the Gene Expression Omnibus (GEO) database, and batch effects were corrected using the "Combat" algorithm in the "SVA" package. Functional annotation of differential genes in psoriasis was performed by Gene set enrichment analysis (GSEA). Core functional modules were identified using the Multiscale Embedded Gene Co-Expression Network Analysis (MEGENA) algorithm for selection from the differential gene interaction network. The expression and potential function of Rh Family C Glycoprotein (RHCG) was predicted in single cell data by the "Seurat" package and validated in psoriasis samples by multiplex immunofluorescence. In addition, the regulatory function of HOP Homeobox (HOPX) on RHCG in keratinocytes was confirmed using RNA interference. Using immune infiltration analysis, RHCG and DC cells were analyzed for their association. Finally, the molecular mechanisms of treatment of psoriasis using Tripterygii Radix (TR) and Cinnamomi Ramulus (CR) were explored through network pharmacology and experimental validation. Immune response (represented by C1_2) and collagen matrix formation (represented by C1_3) were identified as two important pathogenic factors in psoriasis and helped to define new biological subtypes of psoriasis. One important psoriasis hub gene, RHCG, was obtained and found to be closely associated with keratinocyte differentiation as well as DC cell maturation. And RHCG was regulated by HOPX in keratinocytes. In addition, the mechanism of action of CR and TR in the treatment of psoriasis was tentatively confirmed to be related to TRPV3, NFKB2, and YAP1. Our study identifies a new causal disease gene (RHCG) and offers potential alternatives for the treatment of psoriasis.

Multi-modality multi-scale cardiovascular disease subtypes classification using Raman image and medical history

Raman spectroscopy (RS) has been widely used for disease diagnosis, e.g., cardiovascular disease (CVD), owing to its efficiency and component-specific testing capabilities. A series of popular deep learning methods have recently been introduced to learn nuance features from RS for binary classifications and achieved outstanding performance than conventional machine learning methods. However, these existing deep learning methods still confront some challenges in classifying subtypes of CVD. For example, the nuance between subtypes is quite hard to capture and represent by intelligent models due to the chillingly similar shape of RS sequences. Moreover, medical history information is an essential resource for distinguishing subtypes, but they are underutilized. In light of this, we propose a multi-modality multi-scale model called M3S, which is a novel deep learning method with two core modules to address these issues. First, we convert RS data to various resolution images by the Gramian angular field (GAF) to enlarge nuance, and a two-branch structure is leveraged to get embeddings for distinction in the multi-scale feature extraction module. Second, a probability matrix and a weight matrix are used to enhance the classification capacity by combining the RS and medical history data in the multi-modality data fusion module. We perform extensive evaluations of M3S and found its outstanding performance on our in-house dataset, with accuracy, precision, recall, specificity, and F1 score of 0.9330, 0.9379, 0.9291, 0.9752, and 0.9334, respectively. These results demonstrate that the M3S has high performance and robustness compared with popular methods in diagnosing CVD subtypes.

Identification and validation of genes associated with copper death in oral squamous cell carcinoma based on machine learning and weighted gene co-expression network analysis

ObjectiveTo identify copper-induced death-associated hub genes in oral squamous cell carcinoma (OSCC) and understand their functional and biological significance using machine learning and Weighted Gene Co-expression Network Analysis (WGCNA). MethodsOSCC transcriptomic data from GEO and TCGA databases were subjected to data integration, batch effect removal, background correction, and quantile normalization to select cuproptosis-associated genes using Spearman's correlation analysis. The 'limma' R package was used to filter differentially expressed genes (DEGs). Core module genes selected using gene co-expression network analysis with R package 'WGCNA' were screened using Support Vector Machine (SVM), LASSO regression, and Random Forest (RF) machine learning algorithms and validated using TCGA database samples. Core gene expression variations between OSCC and adjacent normal tissues were validated using immunohistochemistry. Immune infiltration analysis using package 'CIBERSORT' correlated hub genes with immune cells. ResultsFrom 19 cuproptosis-related genes (identified from literature), 2382 cuproptosis-related mRNA were obtained through Spearman's correlation analysis; 112 DEGs using 'limma' R package and 32 hub genes using WGCNA were obtained. Hub genes TMPRSS11B, SERPINH1, and CDH3 were identified using machine learning algorithms. TCGA validation showed that TMPRSS11B significantly underexpressed (P < 0.001) but SERPINH1 and CDH3 significantly overexpressed (P < 0.001) in tumor samples. The AUC for TMPRSS11B, SERPINH1, and CDH3 in ROC curve analysis were 78.1%, 95.6%, and 87.5%, respectively. ConclusionTMPRSS11B, SERPINH1, and CDH3 may be pivotal for OSCC development and progression and potential targets for new therapeutic and predictive strategies. However, their specific functions and mechanisms underlying OSCC remain to be elucidated.

Avoiding Plagiarism: Perception of Undergraduates on Referencing When Learning English for Academic Purposes (EAP) at a Private University in Sri Lanka

This study examines undergraduates’ perception on referencing which is taught to them in the course English for Academic Purposes (EAP). The objective of the research is to identify the extent to which awareness of academic referencing is perceived as significant by undergraduates at a private university where they follow EAP as a core module in their first year. The study examines the extent to which these undergraduates are adhering to their institutional academic integrity policy. An online questionnaire was administered among 400 undergraduates, inquiring about their experience in avoiding plagiarism, a form of academic dishonesty, and how much they are aware on overcoming this type of academic deceit. Additionally, attention was focused on examining how plagiarism is avoided with the advancement of technology where online based plagiarism is found to be common among undergraduates. According to the findings of the study, many undergraduates refer online resources when submitting their work and they mention that they follow proper referencing strategies when submitting their work; additionally, they are aware of the plagiarism policy maintained by their institution as well as by their lecturers and the penalties that follow. However, one of the key findings which is significant and should be noted is that, many undergraduates mention that they do not get adequate training or practice on referencing other than the exposure they get during their first year at the university.

A renewed argument for the designation of Social Security Law as a module in the LLB curriculum

The Covid-19 pandemic and the government’s response thereto has shifted socio-economic rights to the forefront of the nation’s attention. Social security measures formed the vanguard of measures to address the dire economic consequences of the lockdown and subsequent restrictions on breadwinners and their dependants. More than two years after the government’s temporary interventions, the discourse has morphed into a debate on the need and viability of more comprehensive and long-term measures. All the proposed measures – whether it be a national social security fund, national health insurance or even a basic income grant – point towards social security becoming ubiquitous in South Africa. However, very few LLB graduates have been exposed to the Social Security Law module as part of their law degree. This underscores the need for equipping LLB graduates with the fundamentals of this branch of law. This article makes three arguments. First, it argues for the inclusion of the Social Security Law module in the LLB curriculum of all law faculties in the country, preferably as a core module. Second, it posits that it is time, almost 30 years into the country’s democracy, to shift our attention towards an aggressive pursuit of socio-economic rights, and in doing so, there is a corresponding need to equip LLB graduates with the tools to play a meaningful role in the country’s transformation agenda. Finally, the article makes a case for the creation of an association of Social Security Law teachers to ensure, among others, the sharing of expertise and the provision of support and guidance to faculties and colleagues who wish to introduce Social Security Law as an LLB module.

Humoral immune disorders affect clinical outcomes of oral lichen planus.

The molecular characteristics of oral lichen planus (OLP) are still unclear, and it is not possible to distinguish the clinical outcome of OLP patients in a short period of time for follow-up. Here, we investigate the molecular characteristics of lesions in patients with stable lichen planus (SOLP) and recalcitrant erosive oral lichen planus (REOLP). Our clinical follow-up cohort was split into SOLP and REOLP groups based on the follow-up clinical data. The core modules associated with the clinical information were identified by weighted gene co-expression network analysis (WGCNA). The OLP cohort samples were divided into two groups by molecular typing, and a prediction model for OLP was created by training neural networks with the neuralnet package. We screened 546 genes in five modules. After doing a molecular type of OLP, it was determined that B cells might have a significant impact on the clinical outcome of OLP. In addition, by means of machine learning, a prediction model was developed to predict the clinical regression of OLP with greater accuracy than the existing clinical diagnostic. Our study revealed humoral immune disorders may make an important contribution to the clinical outcome of OLP.

Deciphering Nucleic Acid Binding Proteome of Mouse Immune Organs Reveals Hub Proteins for Aging

Profiling the nucleic acid-binding proteins (NABPs) during aging process is critical to elucidate its roles in biological systems as well as transcriptional and translational regulation. Here, we developed a comprehensive strategy to survey the NABPs of mouse immune organs by using single cell preparation and selective capture technology-based proteomics. Our approach provided a global view of tissue NABPs from different organs under normal physiological conditions with extraction specificity of 70 to 90%. Through quantitative proteomics analysis of mouse spleen and thymus at 1, 4, 12, 24, 48, and 72weeks, we investigated the molecular features of aging-related NABPs. A total of 2674 proteins were quantified in all six stages, demonstrating distinct and time-specific expression pattern of NABPs. Thymus and spleen exhibited unique aging signatures, and differential proteins and pathways were enriched across the mouse lifespan. Three core modules and 16 hub proteins associated with aging were revealed through weighted gene correlation network analysis. Significant candidates were screened for immunoassay verification, and six hub proteins were confirmed. The integrated strategy pertains the capability to decipher the dynamic functions of NABPs in aging physiology and benefit further mechanism research.

Towards Complex Scenarios: Building End-to-End Task-Oriented Dialogue System across Multiple Knowledge Bases

With the success of the sequence-to-sequence model, end-to-end task-oriented dialogue systems (EToDs) have obtained remarkable progress. However, most existing EToDs are limited to single KB settings where dialogues can be supported by a single KB, which is still far from satisfying the requirements of some complex applications (multi-KBs setting). In this work, we first empirically show that the existing single-KB EToDs fail to work on multi-KB settings that require models to reason across various KBs. To solve this issue, we take the first step to consider the multi-KBs scenario in EToDs and introduce a KB-over-KB Heterogeneous Graph Attention Network (KoK-HAN) to facilitate model to reason over multiple KBs. The core module is a triple-connection graph interaction layer that can model different granularity levels of interaction information across different KBs (i.e., intra-KB connection, inter-KB connection and dialogue-KB connection). Experimental results confirm the superiority of our model for multiple KBs reasoning.

Solving Large-Scale Pursuit-Evasion Games Using Pre-trained Strategies

Pursuit-evasion games on graphs model the coordination of police forces chasing a fleeing felon in real-world urban settings, using the standard framework of imperfect-information extensive-form games (EFGs). In recent years, solving EFGs has been largely dominated by the Policy-Space Response Oracle (PSRO) methods due to their modularity, scalability, and favorable convergence properties. However, even these methods quickly reach their limits when facing large combinatorial strategy spaces of the pursuit-evasion games. To improve their efficiency, we integrate the pre-training and fine-tuning paradigm into the core module of PSRO -- the repeated computation of the best response. First, we pre-train the pursuer's policy base model against many different strategies of the evader. Then we proceed with the PSRO loop and fine-tune the pre-trained policy to attain the pursuer's best responses. The empirical evaluation shows that our approach significantly outperforms the baselines in terms of speed and scalability, and can solve even games on street maps of megalopolises with tens of thousands of crossroads -- a scale beyond the effective reach of previous methods.

Improved prototypical network for active few-shot learning

Different from deep learning with large scale supervision, few-shot learning aims to learn the samples’ characteristics from few labeled examples. Apparently, few-shot learning is more in line with the visual cognitive mechanism of the human brain. In recent years, few-shot learning has attracted more researchers’ attention. They all assume that each category only contains a few labeled samples and without unlabeled samples during the few-shot training. However, in reality, each visual categories also includes some unlabeled samples which store rich semantic information. Labeling all unlabeled samples is time-consuming and laborious. Therefore, we combine the few-shot learning and active learning together, and propose an active few-shot learning model (AC-FSL) based on prototype network. This model includes two core modules: few-shot classification module and loss prediction module. The former is used for classification task under few labeled samples, while the latter is applied to choose some unlabeled samples of high value for labeling, and then assist for the few-shot classification task. We conduct extensive experiments on two image benchmark datasets. The results show that the proposed model AC-FSL can effectively improve the classification performance of current methods on few-shot setting.

Computerized Cognitive Behavioral Therapy for Anxiety and Depression in Farming Communities: Mixed Methods Feasibility Study of Participant Use and Acceptability.

Farmers have higher rates of depression than nonfarmers and higher rates of suicide than the general population. Several barriers to help seeking have been identified in farmers, which may be overcome by offering web-based mental health support. Computerized cognitive behavioral therapy (cCBT) is an effective intervention used to prevent and treat mild to moderate depression but has not been evaluated in the farming community. This study explored the feasibility of delivering a cCBT course tailored to farmers using a mixed methods approach. Farmers (aged ≥18 years) with no, minimal, or moderately severe depressive symptoms (Patient Health Questionnaire-9 [PHQ-9] score <20) were recruited using web-based and offline advertisements and given access to a cCBT course consisting of 5 core modules and automated and personalized email support. Depression (PHQ-9), anxiety (General Anxiety Disorder-7), and social functioning (Work and Social Adjustment Scale) were measured at baseline and the 8-week follow-up. Wilcoxon signed rank tests assessed changes in scores for all outcome measures over time. Telephone interviews focusing on participant use and satisfaction with the course were analyzed using thematic analysis. Overall, 56 participants were recruited; 27 (48%) through social media. Overall, 62% (35/56) of participants logged into the course. At baseline, almost half of the participants reported experiencing minimal depressive symptoms (25/56, 45%) and mild anxiety (25/56, 45%), and just over half (30/56, 54%) reported mild to moderate functional impairment. Posttreatment data were available for 27% (15/56) of participants (41/56, 73% attrition rate). On average, participants experienced fewer depressive symptoms (P=.38) and less functional impairment (P=.26) at the 8-week follow-up; these results were not statistically significant. Participants experienced significantly fewer symptoms of anxiety at the 8-week follow-up (P=.02). Most participants (13/14, 93%) found the course helpful and easy to access (10/13, 77%) and the email support helpful (12/14, 86%). Qualitative interviews identified heavy workloads and mental health stigma within the farming community as barriers to help seeking. Participants thought that web-based support would be helpful, being convenient and anonymous. There were concerns that older farmers and those with limited internet connections may have difficulty accessing the course. Improvements regarding the layout and content of the course were suggested. Dedicated support from someone with farming knowledge was recommended to improve retention. cCBT may be a convenient way of supporting mental health within farming communities. However, challenges in recruiting and retaining farmers may indicate that cCBT supported only by email may not be an acceptable mode of mental health care delivery for many; however, it was valued by respondents. Involving farming organizations in planning, recruitment, and support may address these issues. Mental health awareness campaigns targeting farming communities may also help reduce stigma and improve recruitment and retention.

SPS-LCNN: A Significant Point Sampling-based Lightweight Convolutional Neural Network for point cloud processing

Point cloud data have very promising applications, but the irregularity and disorder make it a challenging problem how to use them. In recent years, an increasing number of new and excellent research solutions have been proposed, which focus on exploring local feature extractors. Over-engineered feature extractors lead to saturating the performance of current methods and often introduce unfavorable latency and additional overhead. This defeats the original purpose of using point cloud data, which is simplicity and efficiency. In this paper, we construct a learnable pipeline by designing two core modules with a small number of parameters – significant point sampling (SPS) and multiscale significant feature extraction (MS-SFE) – to balance accuracy and overhead. Our pipeline demonstrates comparable performance to state-of-the-art methods while requiring fewer parameters, making it well-suited for real-time applications.

Pyramid multi-loss vision transformer for thyroid cancer classification using cytological smear

Multi-instance learning, a commonly used technique in artificial intelligence for analyzing slides, can be applied to diagnose thyroid cancer based on cytological smears. Since smears do not have multidimensional histological features similar to histopathology, mining potential contextual information and diversity of features is crucial for better classification performance. In this paper, we propose a pyramid multi-loss vision transformer model called PyMLViT, a novel algorithm with two core modules to address these issues. Specifically, we design a pyramid token extraction module to acquire potential contextual information on smears. The pyramid token structure extracts multi-scale local features, and the vision transformer structure further obtains global information through the self-attention mechanism. Furthermore, we construct multi-loss fusion module based on the conventional multi-instance learning framework. With carefully designed bag and patch weight allocation strategies, we incorporate slide-level annotations as pseudo-labels for patches to participate in training, thus enhancing the diversity of supervised information. Extensive experimental results on the real-world dataset show that PyMLViT has a high performance and a competitive number of parameters compared to popular methods for diagnosing thyroid cancer in cytological smears.

Identification of novel prognostic circRNA biomarkers in circRNA-miRNA-mRNA regulatory network in gastric cancer and immune infiltration analysis

BackgroundGastric cancer (GC) carries significant morbidity and mortality globally. An increasing number of studies have confirmed that circular RNA (circRNA) is tightly associated with the carcinogenesis and development of GC, especially acting as a competing endogenous RNA for miRNAs.ObjectiveOur study aimed to construct the circRNA-miRNA-mRNA regulatory network and analyze the function and prognostic significance of the network using bioinformatics tools.MethodsWe first downloaded the GC expression profile from the Gene Expression Omnibus database and identified differentially expressed genes and differentially expressed circRNAs. Then, we predicted the miRNA-mRNA interaction pairs and constructed the circRNA-miRNA-mRNA regulatory network. Next, we established a protein-protein interaction network and analyzed the function of these networks. Finally, we primarily validated our results by comparison with The Cancer Genome Atlas cohort and by performing qRT-PCR.ResultsWe screened the top 15 hub genes and 3 core modules. Functional analysis showed that in the upregulated circRNA network, 15 hub genes were correlated with extracellular matrix organization and interaction. The function of downregulated circRNAs converged on physiological functions, such as protein processing, energy metabolism and gastric acid secretion. We ascertained 3 prognostic and immune infiltration-related genes, COL12A1, COL5A2, and THBS1, and built a nomogram for clinical application. We validated the expression level and diagnostic performance of key prognostic differentially expressed genes.ConclusionsIn conclusion, we constructed two circRNA-miRNA-mRNA regulatory networks and identified 3 prognostic and screening biomarkers, COL12A1, COL5A2, and THBS1. The ceRNA network and these genes could play important roles in GC development, diagnosis and prognosis.