Construction of a molecular diagnostic system for neurogenic rosacea by combining transcriptome sequencing and machine learning

Construction of a molecular diagnostic system for neurogenic rosacea by combining transcriptome sequencing and machine learning | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Construction of a molecular diagnostic system for neurogenic rosacea by combining transcriptome sequencing and machine learning Rui Mao, Ji Li This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3791877/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 13 Sep, 2024 Read the published version in BMC Medical Genomics → Version 1 posted 12 You are reading this latest preprint version Abstract Patients with neurogenic rosacea (NR) frequently demonstrate pronounced neurological manifestations, often unresponsive to conventional therapeutic approaches. A molecular-level understanding and diagnosis of this patient cohort could significantly guide clinical interventions. In this study, we amalgamated our sequencing data (n = 46) with a publicly accessible database (n = 38) to perform an unsupervised cluster analysis of the integrated dataset. The eighty-four rosacea patients were partitioned into two distinct clusters. Neurovascular biomarkers were found to be elevated in cluster 1 compared to cluster 2. Pathways in cluster 1 were predominantly involved in neurotransmitter synthesis, transmission, and functionality, whereas cluster 2 pathways were centered on inflammation-related processes. Differential gene expression analysis and WGCNA were employed to delineate the characteristic gene sets of the two clusters. Subsequently, a diagnostic model was constructed from the identified gene sets using linear regression methodologies. The model's C index, comprising genes PNPLA3, CUX2, PLIN2, and HMGCR, achieved a remarkable value of 0.9683, with an area under the curve (AUC) for the training cohort's nomogram of 0.9376. Clinical characteristics from our dataset (n = 46) were assessed by three seasoned dermatologists, forming the NR validation cohort (NR, n = 18; non-neurogenic rosacea, n = 28). Upon application of our model to NR diagnosis, the model's AUC value reached 0.9023. Finally, potential therapeutic candidates for both patient groups were predicted via the Connectivity Map. In summation, this study unveiled two clusters with unique molecular phenotypes within rosacea, leading to the development of a precise diagnostic model instrumental in NR diagnosis. neurogenic rosacea diagnostic system transcriptome Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Rosacea is a prevalent chronic inflammatory cutaneous disorder of the face. Epidemiological surveys report its prevalence to range between 0% and 22%. ( 1 , 2 ) Often manifested as erythema likened to a “drunken” face, accompanied by sensations of burning and tingling, rosacea markedly impacts the patient's appearance, quality of life, and psychological well-being.( 3 , 4 ) A subset of rosacea patients exhibits distinct clinical features, characterized by pronounced facial redness, burning, tingling, and sensory disturbances extending beyond mere inflammation. In 2011, Scharschmidt designated this subgroup as neurogenic rosacea (NR),( 6 ) based on sensations rather than cutaneous manifestations. NR patients commonly present with neurological or neuropsychiatric conditions such as complex regional pain syndrome, idiopathic tremor, depression, and obsessive-compulsive disorder.( 6 – 8 ) Conventional rosacea treatments like metronidazole,( 9 ) isotretinoin,( 10 ) and oral tetracycline,( 11 ) are largely ineffective for NR, while neurological treatments, including oral gabapentin, pregabalin, and tricyclic antidepressants,( 6 , 8 ) may yield superior therapeutic outcomes. Despite this, NR is not internationally recognized, and diagnosis, largely reliant on subjective clinical judgment, lacks an objective molecular diagnostic model. This oversight, particularly among primary care dermatologists, often leads to inappropriate treatments and exacerbation of NR symptoms, underscoring the pressing need for precise NR diagnosis. Although the pathogenesis of rosacea remains elusive, factors including innate immune system imbalance,( 12 ) gut microbiota,( 13 ) physical triggers like ultraviolet radiation,( 14 ) skin barrier dysfunction,( 15 ) and aberrant neurovascular signaling ( 16 ) are implicated in rosacea's onset and progression. The role of neurovascular homeostasis in rosacea has gained recognition, particularly regarding the hypersensitivity of neural responses in NR patients and prevalent neurological complications( 6 ). This highlights the significance of neurovascular dysfunction in NR, warranting exploration of the molecular mediators involved. Two major molecular categories are involved in mediating neural functions in rosacea: ion channel-associated proteins (including the Mas-related G protein-coupled receptors (Mrgpr) family and TRP channels) and neuropeptides (such as substance P and calcitonin gene-related peptide (CGRP)).( 17 ) The activation of neurogenic TRP channels may stimulate skin vascular systems, leading to flushing, a hallmark rosacea feature. Specific G-protein–coupled receptors in the Mrgpr family are principally engaged in cutaneous neurogenic inflammation, facilitating mast cell communication with sensory nerves and cutaneous cells. ( 18 – 20 ) These molecules might be instrumental in NR development. Biomarker-based diagnostics, owing to its objectivity and accuracy, is employed across a range of diseases. ( 21 – 23 ) Construction of disease-specific biomarkers and auxiliary diagnostic models have become the prevailing diagnostic methodology. In this context, our study assembled 43 biomarkers, encompassing mRNAs encoding neuropeptides like substance P, PACAP, VIP, and CRGP, as well as mRNAs for ion channel-related proteins like TRP channels and the Mrgpr family. Through clustering, we differentiated two patient groups and constructed an auxiliary diagnostic model to facilitate NR diagnosis. Materials and methods Clinical diagnosis Three seasoned dermatologists from the Xiangya Hospital's dermatology department, well-versed in diagnosing and treating rosacea, collaborated with three highly experienced psychiatric clinicians from the same hospital to categorize 46 rosacea patients in our cohort. The diagnostic criteria for NR encompassed:: First, rosacea could be diagnosed according to the 2017 edition of the International Diagnostic guidelines ( 24 ) for rosacea and the patient’s facial symptoms and other clinical data. Second, there were no obvious papules and pustules in the patient's facial photos. Third, the patient had obvious facial redness, burning, tingling, and sensory disorders beyond blushing or inflammation. Fourth, patients might have had neurological or neuropsychiatric disorders, including complex regional pain syndrome, idiopathic tremor, depression, and obsessive-compulsive disorder.( 25 ) Fifth, topical or oral metronidazole, isotretinoin, and other first-line treatment drugs were ineffective, but the symptoms significantly improved after using gabapentin or tricyclic antidepressants.( 6 , 7 ) Finally, demodex acne, contact dermatitis, and other differential diagnoses were excluded. Participants had to meet the above six diagnostic criteria at the same time to be diagnosed as NR. We used a questionnaire survey and physician evaluation to evaluate the clinical phenotype of patients with rosacea. Specifically, we used the Global flush severity score (GFSS) scale to evaluate paroxysmal flashes.( 26 ) The researcher erythema rating scale (CEA)( 26 ) and the patient erythema self-rating scale (PSA)( 26 ) were used to evaluate persistent erythema. The evaluation of papules and pustules was carried out by the combination of “Statistics of the number of papules and pustules” and “researcher Global Evaluation (IGA) scale.” The psychological burden of patients with rosacea was mainly assessed by the Chinese version of the Rosacea Quality of Life (RosaQoL) scale, the Dermatology Life Quality Index (DLQI),( 26 ) and the Penn State Worry Questionnaire (PSWQ).( 26 ) All evaluation forms are available in Table S1 . The clinical data ( Table S2 ) of 46 patients with rosacea are provided as supplementary data . Data acquisition and integration Our team's previous research collected skin biopsy samples from the central facial area of patients diagnosed with rosacea, aged between 20 and 60, as well as from healthy females undergoing cosmetic procedures (referred to as the HS group), at the Dermatology Department of Xiangya Hospital, Central South University. The collection period spanned from June 1, 2014, to April 4, 2020. This study involved 46 clinically and pathologically diagnosed rosacea patients and 19 age-similar HS participants. Post-collection, all samples were immediately stored at − 80°C pending analysis. Every sample and its associated clinical information were gathered with informed consent. The study's protocol underwent rigorous review and received approval from the Xiangya Hospital Central South University's ethics review board. all methods were performed in accordance with the relevant guidelines and regulations. Our sequencing data from rosacea patients were deposited in the genome sequence archive under accession number HRA000379 ( http://bigd.big.ac.cn/gsa-human/ ). The GSE65914 datasets (microarray chip ), which contain 38 rosacea and 20 normal samples, were downloaded from the GEO ( https://www.ncbi.nlm.nih.gov/gds/ ) database with a log2 transform and quantile normalization matrix. We used the combat function of R package “sva” to combine the GSE65914 dataset and our local dataset and remove the batch effect. Unsupervised clustering Utilizing the expression patterns of the 43 identified biomarkers (Table 1 ), we embarked on an unsupervised cluster analysis for the 84 rosacea patients via the ConsensusClusterPlus software package ( 27 ). Detailed information about these 43 biomarkers is shown in Table 1 . Using agglomerative km clustering with a Pearson correlation distance of 1 and resampling 80% of the samples for 10 repetitions, the number and stability of clusters were determined according to the consensus clustering algorithm.( 28 ) To enhance the stability of classification, this process was repeated 1,000 times. Table 1 Detailed information of 43 neurogenic rosacea related genes SYMBOL ENSEMBL ENTREZID GENENAME ADCYAP1 ENSG00000141433 116 adenylate cyclase activating polypeptide 1 ADCYAP1R1 ENSG00000078549 117 ADCYAP receptor type I ADM ENSG00000148926 133 adrenomedullin ADM2 ENSG00000128165 79924 adrenomedullin 2 ADRA1A ENSG00000120907 148 adrenoceptor alpha 1A ADRA1B ENSG00000170214 147 adrenoceptor alpha 1B ADRA1D ENSG00000171873 146 adrenoceptor alpha 1D ADRA2A ENSG00000150594 150 adrenoceptor alpha 2A ADRA2B ENSG00000274286 151 adrenoceptor alpha 2B ADRA2C ENSG00000184160 152 adrenoceptor alpha 2C ADRB1 ENSG00000043591 153 adrenoceptor beta 1 ADRB2 ENSG00000169252 154 adrenoceptor beta 2 ADRB3 ENSG00000188778 155 adrenoceptor beta 3 CALCA ENSG00000110680 796 calcitonin related polypeptide alpha CALCR ENSG00000004948 799 calcitonin receptor CALCRL ENSG00000064989 10203 calcitonin receptor like receptor DRD1 ENSG00000184845 1812 dopamine receptor D1 F2RL1 ENSG00000164251 2150 F2R like trypsin receptor 1 F2RL3 ENSG00000127533 9002 F2R like thrombin or trypsin receptor 3 HRH1 ENSG00000196639 3269 histamine receptor H1 HRH2 ENSG00000113749 3274 histamine receptor H2 HRH3 ENSG00000101180 11255 histamine receptor H3 HRH4 ENSG00000134489 59340 histamine receptor H4 HTR2A ENSG00000102468 3356 5-hydroxytryptamine receptor 2A HTR3A ENSG00000166736 3359 5-hydroxytryptamine receptor 3A MMEL1 ENSG00000277131 79258 membrane metalloendopeptidase like 1 MRGPRX1 ENSG00000170255 259249 MAS related GPR family member X1 MRGPRX2 ENSG00000183695 117194 MAS related GPR family member X2 MRGPRX4 ENSG00000179817 117196 MAS related GPR family member X4 NGFR ENSG00000064300 4804 nerve growth factor receptor NPFFR2 ENSG00000056291 10886 neuropeptide FF receptor 2 NPY ENSG00000122585 4852 neuropeptide Y TAC1 ENSG00000006128 6863 tachykinin precursor 1 TACR1 ENSG00000115353 6869 tachykinin receptor 1 TRPA1 ENSG00000104321 8989 transient receptor potential cation channel subfamily A member 1 TRPV2 ENSG00000187688 51393 transient receptor potential cation channel subfamily V member 2 TRPV3 ENSG00000167723 162514 transient receptor potential cation channel subfamily V member 3 TRPV4 ENSG00000111199 59341 transient receptor potential cation channel subfamily V member 4 TRPV5 ENSG00000127412 56302 transient receptor potential cation channel subfamily V member 5 TRPV6 ENSG00000276971 55503 transient receptor potential cation channel subfamily V member 6 VIP ENSG00000146469 7432 vasoactive intestinal peptide VIPR1 ENSG00000114812 7433 vasoactive intestinal peptide receptor 1 VIPR2 ENSG00000106018 7434 vasoactive intestinal peptide receptor 2 Gene differential expression analysis The linear model for microarray analysis (Limma) ( 29 ) is a differential expression screening method based on the generalised linear model. Here, we used the R software package limma (version 3.40.6) for differential analysis to obtain differential genes between different comparison and control groups. We used the criterion that the absolute value of the fold change (FC) is greater than 1.5 and the adjusted-P value is less than 0.05 to screen for significant differentially expressed genes. Principal component analysis (PCA) We executed PCA through the "FactoMineR" package of R software. Briefly, we first standardized the data, then calculated the eigenvalues and variances, and simultaneously calculated the contribution of each variable to PC. Finally, the PCA results were visualised through the "factoextra" package. Analysis of immune cell infiltration Based on the merged data matrix, we used the ssGSEA method of the GSVA package ( 30 ), MCPcounter ( 31 ), and Xcell ( https://xcell.ucsf.edu/ ) ( 32 ) to calculate the abundance of immune cells and excluded the samples with P > 0.05. Finally, the Mann–Whitney U test was used to analyse the differences in immune cell subtypes among different groups. Gene enrichment analysis and gene set enrichment analysis We used the clusterProfiler ( 33 ) package for GO-BP and KEGG enrichment analysis, and the cutoff values of the P value and adjusted-P value were 0.05. There were fewer than three mRNAs covered in the enrichment path. The clusterProfiler ( 33 ) package was used for gene set enrichment analysis (GSEA). The input files were gene names, and their values of log 2 (fold change) were between the high- and low-risk groups. Weighted gene co-expression network analysis (WGCNA) First, based on the gene expression profile, we calculated the Median Absolute Deviation (MAD) of each gene, eliminated the top 50% of the genes with the smallest MAD, removed the outlier genes and samples by using the good Sample Genes method of the R software package WGCNA ( 34 ), and further used WGCNA to construct a scale-free co-expression network. Specifically, first, Pearson's correlation matrices and average linkage method were performed for all pair-wise genes. Then, a weighted adjacency matrix was constructed using the power function A_mn=|C_mn|^β (C_mn = Pearson's correlation between Gene_m and Gene_n; A_mn = adjacency between Gene m and Gene n). Here, β is a soft-thresholding parameter that emphasises strong correlations between genes and penalises weak correlations. After choosing the power of 4, we transformed the adjacency into a topological overlap matrix (TOM), which could measure the network connectivity of a gene defined as the sum of its adjacency with all other genes for the network gene ratio, and the corresponding dissimilarity (1-TOM) was calculated. To classify genes with similar expression profiles into gene modules, average linkage hierarchical clustering was conducted according to the TOM-based dissimilarity measure, with a minimum size (Gene group) of 30 for the genes’ dendrogram. We set the sensitivity to 3. To further analyse the module, we calculated the dissimilarity of module eigen genes, chose a cut line for the module dendrogram and merged some modules. In addition, we merged the modules with a distance of less than 0.25 and finally obtained three co-expression modules. Note that the grey module is considered a collection of genes that cannot be assigned to any module. We calculated the correlation between the module and the sample characters and drew the correlation heatmap. The modules related to traits were mined according to the correlation between traits, module feature vector genes, and the p-value. According to the significance of the gene network, that is, the mean value of the correlation between the trait and each gene expression in each module as the significance of the trait in this module, the module with the greatest significance was most related to the trait. Looking for the hub gene (hub genes) related to this trait, we first calculated the internal connectivity and module identity of the gene. The internal connectivity measures the status of the gene within the module, while the module identity indicates to which module the gene belongs. Protein-protein interaction analysis We used the Metascape ( https://metascape.org/gp/index.html#/main/step1 ) website ( 35 ) for protein-protein (PPI) analysis of the modular hub gene, and the PPI analysis of the website was based on the Molecular Complex Detection (MCODE ( 36 )) tool. Finally, we used Cytoscape (version 3.8.2) to visualise the PPI network. Drug prediction The Connectivity Map is a growing resource of over 3 million perturbational profiles available to the wider scientific community. The clue.io platform ( https://clue.io .) offers a range of apps to facilitate the analysis of these data. Based on the QUERY CMap module on the CLUE website, we uploaded the hub gene in the WGCNA module and set the query parameters to "Gene expression (L1000)." Random forest analysis We used the expression matrix of 696 differential genes and 84 rosacea samples as input files and standardized the data. After setting the random number seed, we analyzed the data with the “randomForest” R software package. When building 500 trees, the error rate estimated by OOB was 5.3%. The experimental and control group's classification error rates were 0.02 and 0.06, respectively. Then, we used the “caret” package to split the training and test sets and the “Boruta” package to select and identify the key classification variables. A total of 51 important variables, 40 potentially important variables (tentative variable, no statistical difference between the importance score and the best shadow variable score), and 605 unimportant variables were identified. Next, we used the “dplyr” package to define a function to extract the important values corresponding to each variable and used the “ImageGP” package to draw important variable results. Finally, we used the “Caret” package cross-validation to select parameters and fit the model; specifically, we defined a function to generate some columns of mtry for testing (a series of values not greater than the total number of variables), select data related to key feature variables, and select data related to key feature variables. Construction of the diagnostic model The least absolute shrinkage and selection operator (LASSO) method, which is suitable for a reduction in high-dimensional data,( 37 , 38 ) was used to select the best predictive characteristics. Features with nonzero coefficients in the LASSO regression model were selected.( 39 ) All of the potential predictors selected by lasso analysis were applied to develop a predicting model for the diagnosis of NR (R packages “glmnet” and “rms”).( 40 ) Calibration curves were plotted to assess the calibration of the diagnosis nomogram (R package “rms”). We employed the bootstrap method to assess both the stability and performance of a given model. To quantify the discrimination performance of the diagnosis nomogram, Harrell’s C-index was measured (R package “rms”).( 41 ) Decision curve analysis was conducted to determine the clinical usefulness of the diagnosis nomogram by quantifying the net benefits at different threshold probabilities in the cohort.( 42 ) The net benefit was calculated by subtracting the proportion of all of the patients with false-positive results from the proportion of patients with true-positive results and by weighing the relative harm of forgoing interventions compared with the negative consequences of an unnecessary intervention (R packages “rms” and “rmda”). Receiver operating characteristic (ROC) curve analysis was employed to compare predictions concerning the sensitivity and specificity (R packages “pROC” and “rms”). Results Data processing We integrated GSE65914 with our own sequencing cohort after removing the batch effect through the sva package as a training cohort. As delineated in Figure S1 A and S1B 's bar chart, post-elimination of the batch effect, the data distribution between the two datasets demonstrated a trend towards uniformity with the median lying on the same axis Analysis through UPSET( 43 ) revealed the intersection of the two cohorts yielding a total of 18,384 common protein-coding genes ( Figure S1 C ). The density maps ( Figure S1 D and S1E ) reveal significant discrepancies in the sample distribution across each dataset prior to batch effect removal, indicative of a batch effect. Subsequent to this removal, data distribution within each dataset appeared congruent, with analogous mean and variance. Furthermore, the PCA diagrams ( Figure S1 F and S1G ) illustrated that prior to batch effect removal, the dataset samples were homogeneously clustered, suggestive of a batch effect. However, post-removal, samples from each dataset were both clustered and interspersed, a strong indication of effective batch effect elimination. The final product was a data matrix comprised of 123 samples and 18,384 protein-coding genes. Identification of Two Distinctive Clusters through Unsupervised Clustering Cluster analysis was carried out utilizing using ConsensusClusterPlus (Fig. 1 A). The optimal number of clusters was discerned via the empirical cumulative distribution function (CDF) plot (Fig. 1 B and 1 C) and the evaluation of average intra-group consistency (Fig. 1 D and 1 E). The analysis revealed that at K = 2, intra-group consistency was at its zenith, thereby resulting in the most efficacious clustering. Consequently, 84 rosacea patients were categorized into two clusters: Cluster 1 (n = 46) and Cluster 2 (n = 38). The expression differences of 43 established neurovascular loop biomarkers in rosacea between clusters 1 and 2 were then assessed. A total of 17 biomarkers were discovered to be highly expressed in Cluster 1 (Fig. 1 F). These included nerve growth factor receptor (NGFR), tachykinin precursor 1 (TAC1), adrenomedullin (ADM and ADM2), adrenoceptor (ADRA1A, ADRB1, and ADRA2A), calcitonin (CALCA and CALCRL), histamine (HRH1, HRH2, and HTR3A), and TRPV family genes (TRPA1, TRPV2, TRPV3, and TRPV6). TAC1 encodes four products of the tachykinin family: substance P and neurokinin A, and neuropeptide K and neuropeptide γ. These hormones are thought to be neurotransmitters that interact with neuroreceptors and smooth muscle cells. It is well known that they induce behavioral responses and act as vasodilators and secretagogues.( 44 ) CGRP induces vasodilation. It dilates various blood vessels, including coronary arteries, brain arteries, and systemic vascular system. The anchor-like receptor family (TRPA1) is a member of the ion channel transient receptor potential (TRP) superfamily. Receptor-activated non-selective cationic channels are involved in pain detection and may also be involved in cold perception ( 45 ) and itching.( 46 ) The abundance of inflammatory cell infiltration in cluster 2 was significantly higher than that in cluster 1 Initially, PCA was performed on 84 rosacea patients using 43 biomarkers (Fig. 2 A), followed by analysis on a complete cohort (Fig. 2 B) that included the aforementioned patients and 39 normal samples. The results showed that these 43 biomarkers were able to distinguish between cluster 1, cluster 2, and normal people in the two cohorts. According to the predicted results of Xcell ( https://xcell.ucsf.edu/ ) ( 32 ), the abundance of nerve-related biomarkers in cluster 1 was significantly higher than in cluster 2 (Fig. 2 C). Subsequently, ssGSEA( 30 ) was employed to assess differences in immune cell abundance between the clusters. The results indicated that the immuno-inflammatory cell abundance in Cluster 1 was significantly reduced compared to Cluster 2 (Fig. 2 E). Specifically, most immune cells, including B cells, CD4-positive T cells, CD8-positive T cells, macrophages, mast cells, natural killer cells, and neutrophils were considerably less abundant in cluster1 than in cluster2 (Fig. 2 D and 2 F). Consistency in immunocyte infiltration abundance was observed in the MCP ( Figure S2 ) and Xcell ( Figure S3 ) analyses, with Cluster 1 significantly lower than Cluster 2. The pathways in cluster 1 are mostly concentrated in the synthesis, transmission, and function of neurotransmitters In an endeavor to delineate the disparities in functionality and pathway engagement between clusters 1 and 2, Gene Set Enrichment Analysis (GSEA) was executed for both groups. Utilizing the KEGG database, the GSEA findings revealed that the functional pathways within cluster 1 predominantly encompass the synaptic vesicle cycle, cholesterol metabolism, valine, leucine and isoleucine degradation, in addition to the biosynthesis of the amino acid pathway ( Figure S4 A ). Moreover, the GSEA analysis grounded in the GO database ascertained that the pathways of synaptic signaling, cellular amino acid metabolic process, and cholesterol biosynthesis are predominantly augmented in cluster 1 ( Figure S4 B ). Conversely, the GSEA-KEGG and GSEA-GO results of cluster 2 unveiled that the functional pathways within this cluster are primarily associated with immune- and inflammation-related processes, including but not limited to the IL–17 signaling pathway, NF–kappa B signaling pathway, TNF signaling pathway, positive regulation of inflammatory response to an antigenic stimulus, regulation of CD8-positive, alpha–beta T cell activation, and Th1 and Th2 cell differentiation ( Figure S4 C and S4D ). The foregoing results underscore that the pathways in cluster 1 are overwhelmingly concentrated on the synthesis, transmission, and functionalization of neurotransmitters, whereas the functionality of cluster 2 is largely oriented toward inflammation and immune responsiveness. Subsequently, we categorized cluster 1 as the rosacea group demonstrating a conspicuous neuromolecular phenotype (RHNAG) and cluster 2 as the rosacea group lacking a neuromolecular phenotype (RLNAG). As delineated in Table S2 , the onset symptoms of RHNAG patients predominantly manifested as paroxysmal flushing (n = 16; 66.7%), while a minority exhibited persistent erythema (n = 1; 4.2%) or papular pustules (n = 6; 25.0%). Conversely, the RLNAG patients chiefly presented with papular pustules (n = 15; 68.2%), with a smaller proportion experiencing paroxysmal flushing (n = 7; 31.8%). Furthermore, an evaluation of the Global Flush Severity Score (GFSS) scale revealed significantly elevated scores in RHNAG patients relative to their RLNAG counterparts. Assessments employing the CEA, PSA, and IGA scales intimated that erythema was more pronounced in RHNAG patients, whereas papular pustules were markedly less prevalent compared to RLNAG patients. Noteworthy is the fact that according to the Dermatology Life Quality Index (DLQI), RHNAG patients suffered from a diminished quality of life compared to RLNAG patients, and the Penn State Worry Questionnaire (PSWQ) indicated exacerbated levels of anxiety and depression within the RHNAG group. Collectively, these findings elucidate that the phenotype of RHNAG, characterized by elevated expression of neurotransmitters and receptor-associated markers, aggravated mental health conditions, and comparatively subdued inflammation, bears significant resemblance to that of NR in a clinical context. Screening the gene modules most related to RHNAG or RLNAG by WGCNA In the quest to isolate the gene modules most pertinent to RHNAG or RLNAG, we scrutinized the genetic variances between these patient cohorts. Our examination unveiled a total of 1071 differentially expressed genes, comprising 411 genes that were upregulated in the RHNAG group and 660 genes that were upregulated in the RLNAG group ( Figure S5 A and S5B ). Pursuant to this, we employed Weighted Gene Co-expression Network Analysis (WGCNA) on 3379 genes, all of which exhibited a variance exceeding 0.2 across the samples, and subsequently segmented these genes into four distinct modules ( Figure S6A-S6D ). Correlational analysis between the modules and clinical attributes revealed that the blue module exhibited positive correlations with the RHNAG group and negative correlations with immune cell abundance, including T cells, cytotoxic lymphocytes, and natural killer cells ( Figure S6E ). In contrast, the turquoise module manifested a pronounced positive correlation with the RLNAG group and an abundance of immune cell infiltration. The eigengene distance heatmap showcased the maximum distance between the blue and turquoise modules, indicative of a substantial divergence in gene expression patterns ( Figure S6F ). Furthermore, the scatter plot correlating Gene Significance (GS) with Module Membership (MM) confirmed that the genes most intimately associated with the RHNAG group were also pivotal within the blue module ( Figure S6G ). The function of the hub gene in the blue module is mainly concentrated in the synthesis and function of neurotransmitters To further explore the function of hub genes in the blue and turquoise modules, we carried out PPI analysis and functional enrichment analysis of hub genes in the two modules. The analysis results of MCODE software show that the hub genes of the blue module were mainly focused on the synthesis and function of neurotransmitter materials, such as cholesterol synthesis, amino acid synthesis, and steroid metabolism ( Figure S7A) . The results of KEGG and GO-BP enrichment analysis of the hub genes of the blue module also indicate that the functions of these genes are mainly focused on the pathways related to the synthesis and function of neurotransmitters such as amino acid metabolism, cholesterol metabolism, and steroid metabolism ( Figures S7B and S7C) . On the contrary, the functions of the core genes of the turquoise module are mainly concentrated in immune and inflammatory pathways, such as activation of the immune response, T cell receptor signaling pathway, JAK–STAT signaling pathway, NF–kappa B signaling pathway, and positive regulation of T cell activation ( Figures S8A–S8C ). Construction of the diagnostic model of RHNAG and RLNAG Initially, a random forest analysis was executed on a set of 696 differentially expressed genes contrasting RHNAG and RLNAG, from which 51 pivotal genes were identified as key differentiators between the phenotypes of RHNAG and RLNAG (Fig. 3 A). Subsequently, a scatterplot delineating the top 20 genes based on importance scores was generated (Fig. 3 B). From this pool, 16 molecular entities boasting importance scores surpassing 50 were earmarked for ROC analysis. Out of these, 10 molecules demonstrating AUC values exceeding 0.9 were further shortlisted for lasso regression scrutiny (Fig. 3 C) The subsequent lasso regression pinpointed four genes—HMGCR, CUX2, PLIN2, and PNPLA3—from the aforementioned 10 for model construction, with a lambda value set at 4 ( Figs. 3 D & 3 E ) Utilizing the nomogram approach, a linear diagnostic prediction model for RHNAG was formulated based on these four genes (Fig. 3 G). According to the algorithm in the package, the scores corresponding to the expression of the four molecules are obtained from the following linear formulas, respectively. ( \(\text{s}\text{c}\text{o}\text{r}\text{e} \text{o}\text{f} \text{P}\text{N}\text{P}\text{L}\text{A}3=-20.154(\text{exp }of PNPLA3)+147.73\) ; $$\text{s}\text{c}\text{o}\text{r}\text{e} \text{o}\text{f} \text{C}\text{U}\text{X}2=19.133(\text{exp }of CUX2)-19.4$$ ; $$\text{s}\text{c}\text{o}\text{r}\text{e} \text{o}\text{f} \text{P}\text{L}\text{I}\text{N}2=-21.473(\text{exp }of PLIN2)+181.47$$ ; \(\text{s}\text{c}\text{o}\text{r}\text{e} \text{o}\text{f} \text{H}\text{M}\text{G}\text{C}\text{R}=-10.7(\text{exp }of HMGCR)+120.85\) ). The formula for calculating the RHNAG diagnostic probability corresponding to the total score of the four molecules is as follows: $$probability=1.06777*ln\left(total score\right)-19.4$$ For example, the expression of HMGCR, CUX2, PLIN2, and PNPLA3 in patient numbered GSM1611085 was 8.86, 4.4, 7.29, and 6.78, corresponding respectively to 73, 36, 74, and 91 points according to the formula. Therefore, the patient's overall score was 274, corresponding to the probability of 0.984 according to the formula that they had RHNAG. The final result shows that the patient was indeed an RHNAG patient. Harrell’s C-index of this model was 0.9683. The results of BOOTSTRAT indicated a robust C-index of 0.908 (95% CI [0.901, 0.915]), suggesting high predictive accuracy and reliability of our model. The calibration curve analysis results show that the model's predicted value was close to the ideal (Fig. 3 H). The above results indicate that the diagnostic model can predict whether a patient has the RHNAG type of rosacea. Finally, the results of DCA show that the benefits order of each of the four genes as a model to predict the clinical rate of return of RHNAG was HMGCR > CUX2 > PLIN2 > PNPLA3. The clinical benefit rate of the predictive model of the four genes was higher than that of each gene alone (Fig. 3 F). The model has high sensitivity and specificity in the diagnosis of NR In line with the diagnostic categorization by three experienced dermatologists from Xiangya Hospital, 18 of the 46 rosacea patients were diagnosed with Neurogenic Rosacea (NR) while the remaining 28 were classified as Non-Neurogenic Rosacea (NNR) patients ( Table S3 ). Visual documentation, comprising images of five NR and five NNR patients, revealed a notable absence of papules and pustules on the facial region of NR patients, replaced instead by pervasive flushing (Fig. 4 A). Conversely, NNR patients exhibited pronounced papules and pustules either locally or encompassing the entire visible region (Fig. 4 B). Finally, the total scores of 84 patients with rosacea were calculated according to the diagnostic model, and the ROC analysis was carried out according to the total scores. The results show that the AUC for the nomogram was 0.9376 (95% confidence interval (CI): 0.8898–0.9855; P < 0.001), the best cutoff value was 156, and the sensitivity and specificity were 0.974 and 0.861, respectively, in the total cohort of 84 patients with rosacea (Fig. 4 C). Remarkably, an isolated ROC analysis on the cohort of 46 rosacea patients (NR: n = 18; NNR: n = 28), as assessed by the trio of dermatologists, showcased an AUC of 0.9023 for NR diagnosis (95% CI: 0.826–0.9925; P < 0.001). The prime cutoff value was determined to be 169, with a corresponding sensitivity and specificity of 0.926 and 0.870, respectively (Fig. 4 D). Cumulatively, these findings underscore the robustness and elevated precision of our diagnostic model in clinical applications. Prediction of Therapeutic Agents Leveraging the cardinal genes identified within the blue and turquoise modules, potential therapeutic agents were prognosticated utilizing the Connectivity Map function on the clue.io platform. Agents with predictive scores beneath − 0.6 were earmarked as potential treatment options. Our analysis indicated that the prospective therapeutic agents for RHNAG, as gleaned from the blue module, predominantly belonged to classes such as dopamine receptor antagonists, norepinephrine inhibitors, tricyclic antidepressants, opioid receptor antagonists, and sigma receptor antagonists (Refer to Figure S9A ). Contrarily, the turquoise module's hub gene projections for RLNAG treatments were primarily aligned with mTOR inhibitors, JAK inhibitors, protein kinase inhibitors, MEK inhibitors, and STAT inhibitors ( Figure S9B ). Discussion In this study, leveraging neurovascular biomarkers, we stratified rosacea patients into two distinct groups: RHNAG and RLNAG. Intriguingly, the RLNAG group demonstrated a markedly heightened abundance of immune cells compared to the RHNAG cohort. Comprehensive GSEA, KEGG, and GO-BP enrichment analyses revealed that genes specific to RHNAG are predominantly enriched in pathways involving synaptic neurotransmitters, phospholipids, unsaturated fatty acids, amino acids, and cholesterol metabolism. Conversely, RLNAG-specific genes predominantly aligned with immune and inflammatory pathways. This molecular distinction was mirrored clinically; RHNAG patients typically presented absent of facial papules and pustules, but with pronounced flushing, burning, and tingling sensations. Additionally, a significant portion of the RHNAG group manifested neurotic symptoms including anxiety, depression, and neuroticism, and standard treatments often proved ineffectual. In stark contrast, the RLNAG patients, as observed in the clinical validation cohort, predominantly exhibited facial papules and pustules with a scant occurrence of neurological symptoms such as sensory aberrations and depression. Notably, their symptoms were ameliorated by conventional treatments. A salient observation from the clinical validation cohort, as assessed by a trio of seasoned dermatologists, was the unanimous categorization of all NR patients into the RHNAG group, while the majority of NNR patients predominantly clustered within the RLNAG cohort. Thus, we postulate that RHNAG patients share striking clinical parallels with NR patients, primarily presenting with neurological symptoms. In juxtaposition, RLNAG patients align closely with NNR patients, typified by pronounced inflammation and the presence of facial papules and pustules. This concordance underscores why the RHNAG diagnostic model boasts such elevated sensitivity and specificity in NR diagnosis, reinforcing its potential as a diagnostic surrogate for NR. Cholesterol stands as the primary lipid collaborator with sphingolipids within membrane microdomains. Within the nervous system, cholesterol predominantly constitutes the primary lipid component of myelin (28%).( 47 ) Importantly, pivotal synaptic transmission processes, such as endocytosis, exocytosis, and the lateral diffusion of neurotransmitter receptors within the membrane, are profoundly modulated by cholesterol concentrations.( 48 ) The functionality of neurotransmitter receptors is modulated by lipid domains,( 49 ) cholesterol,( 50 ) and sphingolipids.( 51 ) Cholesterol's modulatory effects on membrane receptor function are largely attributed to either its direct receptor interaction or its overarching influence on the biophysical attributes of the lipid bilayer of membranes.( 50 , 51 ) In the nervous milieu, lipids emerge as the predominant organic compounds. Glycerol phospholipids, sphingolipids, and cholesterol primarily constitute the neural lipid repertoire within both central and peripheral domains.( 47 , 52 ) Sphingolipids, glycerolipids, and their derivates (namely glycosphingolipids, GSLs) are intricately associated with neurogenesis, synaptic transmission, and the synthesis, functionality, and transport of neurotransmitter receptors ( 53 ). Within the architecture of our NR diagnostic model, 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) occupies a crucial position. HMGCR facilitates the conversion of (3S)-hydroxy-3-methylglutaryl-CoA (HMG-CoA) to mevalonic acid, representing the pivotal regulatory step in cholesterol synthesis and other isoprenoid formations, thereby orchestrating cellular cholesterol equilibrium.( 54 , 55 ) Consequently, we postulate HMGCR's involvement in the synthesis of neurotransmitters and their receptor engagement, potentially mediated through cholesterol synthesis regulation, which may underscore its role in NR's pathogenesis. Further, the protein encoded by perilipin 2 (PLIN2) is aligned with the perilipin family, known for coating intracellular lipid storage vesicles, suggesting its potential as a lipid accumulation marker across diverse cellular environments and pathologies.( 56 ) Another protein of interest, encoded by the patatin-like phospholipase domain containing 3 (PNPLA3), functions as a triacylglycerol lipase, mediating triacylglycerol hydrolysis in adipocytes and intricately involved in glycerol phospholipid biosynthesis. ( 57 ) In our investigations, these phospholipid biosynthesis-centric genes exhibited a positive correlation with the composite score of the NR diagnostic model, suggesting their potential to amplify the synthesis, transport, and functional attributes of neurotransmitters and their receptors, possibly through phospholipid and sphingomyelin synthesis, thereby elucidating their role in NR's etiology.The Cut-like homeobox 2 (CUX2) molecule, uniquely characterized by its negative correlation with the model score, encodes a protein housing three CUT domains and a homeodomain, both serving as DNA-binding moieties. This transcription factor critically governs neuronal proliferation and differentiation within the cerebral landscape, specifically modulating dendrite evolution, branching, dendritic spine genesis, and synaptogenesis within cortical strata II-III while demonstrating sequence-specific DNA binding. Markedly, CUX2 epitomizes a differentiation hallmark of glutamatergic pathways.( 58 ) Suzuki et al.( 59 ) established that a deficit in CUX2 significantly augments glutamatergic synaptic transmission within the hippocampal domain. Our findings resonate with this observation, underscoring that reduced CUX2 expression correlates with an elevated predisposition towards RHNAG diagnosis in rosacea-afflicted patients. This suggests that rosacea patients, due to the attenuated expression of CUX2, might experience amplified glutamate-mediated synaptic transmission, potentially triggering neurological manifestations that evolve into NR. Our investigation represents a pioneering endeavor in the formulation of an auxiliary diagnostic model for NR predicated on biomarker profiling. Within a clinical milieu, it becomes feasible to ascertain the expression levels of these quartet molecules in peripheral blood. This allows for computation of the susceptibility to NR using the provided model equation. Subsequently, the derived prognostication can inform the prescription of appropriate therapeutic agents, an advancement that holds profound implications for the clinical diagnostic and therapeutic paradigms of NR. To elucidate, RHNAG-afflicted individuals might benefit from the administration of tricyclic antidepressants such as nortriptyline and protriptyline (traditionally prescribed for depressive disorders), sigma receptor antagonists like rimcazole (prevalently utilized in schizophrenia management), and dopamine receptor antagonists, including trifluoperazine and fluphenazine (typically prescribed for schizophrenia). Conversely, for those diagnosed with RLNAG, therapeutic options might encompass JAK inhibitors like Ruxolitinib (predominantly used for primary myelodysplasia and post-thrombocythemic myelodysplasia, with recent trials for autoimmune pathologies such as psoriasis) and MTOR inhibitors, e.g., everolimus (commonly incorporated in immunosuppressive regimens post-transplantation).Abnormal neurovascular regulation and imbalance of the inflammatory immune system are two intertwined pathogenetic factors of rosacea. Depending on the different clinical phenotypes of rosacea, the emphasis on these two mechanisms may differ. An intriguing query emerges: Why does NR predominantly exhibit neurological symptoms while manifesting attenuated inflammation? Studies by Kronfol et al. and Rothermundt et al. revealed that individuals without melancholic tendencies or depression often display proinflammatory states. Conversely, those with melancholic characteristics or diagnosed depression typically demonstrate diminished proinflammatory cytokine production.( 60 – 62 ) Significantly, anti-inflammatory cytokines such as transforming growth factor (TGF)-β and IL-10 often present at elevated levels in major depression (MD).( 63 – 65 ) Melancholic and non-melancholic patients show different immune patterns. It's plausible that neurosystemic aberrations in RHNAG patients could be inhibiting systemic inflammation. On the other hand, RLNAG rosacea patients exhibit inflammatory levels that align with their clinical manifestations. Nonetheless, our study is not devoid of limitations. A primary limitation is the paucity of valid samples, amounting to only 84, used in constructing the model. While we integrated all publicly available rosacea transcriptome data with our sequencing cohort, current constraints preclude further expansion of the sequencing sample size. Should additional rosacea data emerge in public repositories, we are poised to reassess our model. Another shortcoming stems from our model's reliance solely on mRNA expression levels within lesions. In a clinical setting, biomarkers sourced from blood or urine might be more feasible and palatable to patients. It's also noteworthy that our study observed five RHNAG patients manifesting pronounced papules and pustules, which are uncharacteristic for NR. This highlights the inherent challenge in perfectly aligning the NR phenotype through molecular modeling. Nevertheless, our RHNAG diagnostic model retains commendable accuracy and sensitivity in diagnosing NR. Future endeavors encompassing more extensive sequencing of both NR and NNR patients might pave the way for a refined and more accurate NR diagnostic model. In general, our pioneering approach combined machine learning with linear regression to devise a molecular diagnostic model for NR. This model, characterized by its high sensitivity and specificity, heralds a promising advancement in the diagnosis and treatment of NR. This could be particularly transformative in regions or among clinicians who might have previously overlooked NR. Abbreviations GEO, Gene Expression Omnibus; PCA, Principal Component Analysis; PLIN2, Perilipin 2; PNPLA3, Patatin Like Phospholipase Domain Containing 3; CUX2, Cut Like Homeobox 2; HMGCR, 3-Hydroxy-3-Methylglutaryl-CoA Reductase. Declarations Consent for publication All the authors in this study agreed to the publication of the manuscript. Informed consent for the publication of identifying images has been obtained from all subjects. It is important to note that no minors were included in the study cohort. Ethics approval and consent to participate The experiments conducted in this study were approved by the Clinical Medical Ethics Committee of Xiangya Hospital, Central South University (approval number: 201404361). Written informed consent was obtained from all participants prior to their participation in the study. Availability of data and materials Sequencing data from rosacea patients have been deposited in the genome sequence archive under accession number HRA000379 (http://bigd.big.ac.cn/gsa-human/). Clinical mugshots of 46 rosacea patients can be obtained by contacting the corresponding author. Competing interests The authors declare that they have no conflict of interest. Author contribution Rui Mao : Conceptualization, Methodology, Software, Investigation, Visualization, Writing an original draft, Investigation. Ji Li : Conceptualization, Writing - original draft, Funding acquisition, Supervision. Funding This work was supported by the National Natural Science Funds for Distinguished Young Scholars (No. 82225039)，the National Natural Science Foundation of China (No.82273557 and 8217344) Acknowledgments The authors would like to thank the National Centre for Biotechnology Information staff. We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript. References Rainer BM, Fischer AH, Luz Felipe da Silva D, Kang S, Chien AL. Rosacea is associated with chronic systemic diseases in a skin severity-dependent manner: results of a case-control study. J Am Acad Dermatol. 2015;73:604–8. Thiboutot D, Anderson R, Cook-Bolden F, Draelos Z, Gallo RL, Granstein RD, et al. Standard management options for rosacea: The 2019 update by the National Rosacea Society Expert Committee. J Am Acad Dermatol. 2020;82:1501–10. Egeberg A, Fowler JF Jr, Gislason GH, Thyssen JP. Nationwide Assessment of Cause-Specific Mortality in Patients with Rosacea: A Cohort Study in Denmark. Am J Clin Dermatol. 2016;17:673–9. Vera N, Patel NU, Seminario-Vidal L. Rosacea Comorbidities. Dermatol Clin. 2018;36:115–22. Wilkin J, Dahl M, Detmar M, Drake L, Feinstein A, Odom R, et al. Standard classification of rosacea: Report of the National Rosacea Society Expert Committee on the Classification and Staging of Rosacea. J Am Acad Dermatol. 2002;46:584–7. Scharschmidt TC, Yost JM, Truong SV, Steinhoff M, Wang KC, Berger TG. Neurogenic rosacea: a distinct clinical subtype requiring a modified approach to treatment. Arch Dermatol. 2011;147:123–6. Parkins GJ, Maan A, Dawn G. Neurogenic rosacea: an uncommon and poorly recognized entity. Clin Exp Dermatol. 2015;40:930–1. Kim HO, Kang SY, Kim KE, Cho SY, Kim KH, Kim IH. Neurogenic rosacea in Korea. J Dermatol. 2021;48:49–55. Miyachi Y. Potential antioxidant mechanism of action for metronidazole: implications for rosacea management. Adv Ther. 2001;18:237–43. van Zuuren EJ, Fedorowicz Z. Low-Dose Isotretinoin: An Option for Difficult-to-Treat Papulopustular Rosacea. J Invest Dermatol. 2016;136:1081–3. Del Rosso JQ, Webster GF, Jackson M, Rendon M, Rich P, Torok H, et al. Two randomized phase III clinical trials evaluating anti-inflammatory dose doxycycline (40-mg doxycycline, USP capsules) administered once daily for treatment of rosacea. J Am Acad Dermatol. 2007;56:791–802. Yamasaki K, Di Nardo A, Bardan A, Murakami M, Ohtake T, Coda A, et al. Increased serine protease activity and cathelicidin promotes skin inflammation in rosacea. Nat Med. 2007;13:975–80. Lacey N, Delaney S, Kavanagh K, Powell FC. Mite-related bacterial antigens stimulate inflammatory cells in rosacea. Br J Dermatol. 2007;157:474–81. Yamasaki K, Kanada K, Macleod DT, Borkowski AW, Morizane S, Nakatsuji T, et al. TLR2 expression is increased in rosacea and stimulates enhanced serine protease production by keratinocytes. J Invest Dermatol. 2011;131:688–97. Dirschka T, Tronnier H, Fölster-Holst R. Epithelial barrier function and atopic diathesis in rosacea and perioral dermatitis. Br J Dermatol. 2004;150:1136–41. Choi JE, Di Nardo A. Skin neurogenic inflammation. Semin Immunopathol. 2018;40:249–59. Buddenkotte J, Steinhoff M. Recent advances in understanding and managing rosacea. F1000Res 2018; 7. Del Rosso JQ. Management of facial erythema of rosacea: what is the role of topical α-adrenergic receptor agonist therapy. J Am Acad Dermatol. 2013;69:44–56. Metzler-Wilson K, Toma K, Sammons DL, Mann S, Jurovcik AJ, Demidova O, et al. Augmented supraorbital skin sympathetic nerve activity responses to symptom trigger events in rosacea patients. J Neurophysiol. 2015;114:1530–7. Steinhoff M, Schmelz M, Schauber J. Facial Erythema of Rosacea - Aetiology, Different Pathophysiologies and Treatment Options. Acta Derm Venereol. 2016;96:579–86. Gillessen S, Armstrong A, Attard G, Beer TM, Beltran H, Bjartell A et al. Management of Patients with Advanced Prostate Cancer: Report from the Advanced Prostate Cancer Consensus Conference 2021. Eur Urol 2022. Tayob N, Kanwal F, Alsarraj A, Hernaez R, El-Serag HB. The Performance of AFP, AFP-3, DCP as Biomarkers for Detection of Hepatocellular Carcinoma (HCC): A Phase 3 Biomarker Study in the United States. Clin Gastroenterol Hepatol; 2022. Xie Q, Xue W. IgE-Mediated food allergy: Current diagnostic modalities and novel biomarkers with robust potential. Crit Rev Food Sci Nutr 2022: 1–25. Gallo RL, Granstein RD, Kang S, Mannis M, Steinhoff M, Tan J, et al. Standard classification and pathophysiology of rosacea: The 2017 update by the National Rosacea Society Expert Committee. J Am Acad Dermatol. 2018;78:148–55. Schram AM, James WD. Neurogenic rosacea treated with endoscopic thoracic sympathectomy. Arch Dermatol. 2012;148:270–1. Norquist JM, Watson DJ, Yu Q, Paolini JF, McQuarrie K, Santanello NC. Validation of a questionnaire to assess niacin-induced cutaneous flushing. Curr Med Res Opin. 2007;23:1549–60. Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics. 2010;26:1572–3. McLachlan GJ, Bean RW, Ng SK. Clustering Methods Mol Biol. 2017;1526:345–62. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47. Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7. Becht E, Giraldo NA, Lacroix L, Buttard B, Elarouci N, Petitprez F, et al. Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol. 2016;17:218. Aran D, Hu Z, Butte AJ. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol. 2017;18:220. Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16:284–7. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559. Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10:1523. Bader GD, Hogue CW. An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinformatics. 2003;4:2. Sauerbrei W, Royston P, Binder H. Selection of important variables and determination of functional form for continuous predictors in multivariable model building. Stat Med. 2007;26:5512–28. Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw. 2010;33:1–22. Kidd AC, McGettrick M, Tsim S, Halligan DL, Bylesjo M, Blyth KG. Survival prediction in mesothelioma using a scalable Lasso regression model: instructions for use and initial performance using clinical predictors. BMJ Open Respir Res. 2018;5:e000240. Huang YQ, Liang CH, He L, Tian J, Liang CS, Chen X, et al. Development and Validation of a Radiomics Nomogram for Preoperative Prediction of Lymph Node Metastasis in Colorectal Cancer. J Clin Oncol. 2016;34:2157–64. Kramer AA, Zimmerman JE. Assessing the calibration of mortality benchmarks in critical care: The Hosmer-Lemeshow test revisited. Crit Care Med. 2007;35:2052–6. Vickers AJ, Cronin AM, Elkin EB, Gonen M. Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers. BMC Med Inform Decis Mak. 2008;8:53. Lex A, Gehlenborg N, Strobelt H, Vuillemot R, Pfister H. UpSet: Visualization of Intersecting Sets. IEEE Trans Vis Comput Graph. 2014;20:1983–92. González-Santana A, Marrero-Hernández S, Dorta I, Hernández M, Pinto FM, Báez D, et al. Altered expression of the tachykinins substance P/neurokinin A/hemokinin-1 and their preferred neurokinin 1/neurokinin 2 receptors in uterine leiomyomata. Fertil Steril. 2016;106:1521–9. Moparthi L, Survery S, Kreir M, Simonsen C, Kjellbom P, Högestätt ED, et al. Human TRPA1 is intrinsically cold- and chemosensitive with and without its N-terminal ankyrin repeat domain. Proc Natl Acad Sci U S A. 2014;111:16901–6. Paulsen CE, Armache JP, Gao Y, Cheng Y, Julius D. Structure of the TRPA1 ion channel suggests regulatory mechanisms. Nature. 2015;520:511–7. Curatolo W. Glycolipid function. Biochim Biophys Acta. 1987;906:137–60. Paulazo MA, Sodero AO. SIRT-1 Activity Sustains Cholesterol Synthesis in the Brain. Neuroscience. 2021;476:116–24. Criado M, Eibl H, Barrantes FJ. Effects of lipids on acetylcholine receptor. Essential need of cholesterol for maintenance of agonist-induced state transitions in lipid vesicles. Biochemistry. 1982;21:3622–9. Burger K, Gimpl G, Fahrenholz F. Regulation of receptor function by cholesterol. Cell Mol Life Sci. 2000;57:1577–92. Baier CJ, Barrantes FJ. Sphingolipids are necessary for nicotinic acetylcholine receptor export in the early secretory pathway. J Neurochem. 2007;101:1072–84. Fantini J, Garmy N, Mahfoud R, Yahi N. Lipid rafts: structure, function and role in HIV, Alzheimer's and prion diseases. Expert Rev Mol Med. 2002;4:1–22. Sonnino S, Chigorno V. Ganglioside molecular species containing C18- and C20-sphingosine in mammalian nervous tissues and neuronal cell cultures. Biochim Biophys Acta. 2000;1469:63–77. Luskey KL, Stevens B. Human 3-hydroxy-3-methylglutaryl coenzyme A reductase. Conserved domains responsible for catalytic activity and sterol-regulated degradation. J Biol Chem. 1985;260:10271–7. Cuccioloni M, Mozzicafreddo M, Spina M, Tran CN, Falconi M, Eleuteri AM, et al. Epigallocatechin-3-gallate potently inhibits the in vitro activity of hydroxy-3-methyl-glutaryl-CoA reductase. J Lipid Res. 2011;52:897–907. Kaushik S, Cuervo AM. Degradation of lipid droplet-associated proteins by chaperone-mediated autophagy facilitates lipolysis. Nat Cell Biol. 2015;17:759–70. Romeo S, Kozlitina J, Xing C, Pertsemlidis A, Cox D, Pennacchio LA, et al. Genetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease. Nat Genet. 2008;40:1461–5. Teissier A, Griveau A, Vigier L, Piolot T, Borello U, Pierani A. A novel transient glutamatergic population migrating from the pallial-subpallial boundary contributes to neocortical development. J Neurosci. 2010;30:10563–74. Suzuki T, Tatsukawa T, Sudo G, Delandre C, Pai YJ, Miyamoto H, et al. CUX2 deficiency causes facilitation of excitatory synaptic transmission onto hippocampus and increased seizure susceptibility to kainate. Sci Rep. 2022;12:6505. Rothermundt M, Arolt V, Fenker J, Gutbrodt H, Peters M, Kirchner H. Different immune patterns in melancholic and non-melancholic major depression. Eur Arch Psychiatry Clin Neurosci. 2001;251:90–7. Rothermundt M, Arolt V, Peters M, Gutbrodt H, Fenker J, Kersting A, et al. Inflammatory markers in major depression and melancholia. J Affect Disord. 2001;63:93–102. Kronfol Z. Immune dysregulation in major depression: a critical review of existing evidence. Int J Neuropsychopharmacol. 2002;5:333–43. Howren MB, Lamkin DM, Suls J. Associations of depression with C-reactive protein, IL-1, and IL-6: a meta-analysis. Psychosom Med. 2009;71:171–86. Dowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim EK, et al. A meta-analysis of cytokines in major depression. Biol Psychiatry. 2010;67:446–57. Köhler CA, Freitas TH, Maes M, de Andrade NQ, Liu CS, Fernandes BS, et al. Peripheral cytokine and chemokine alterations in depression: a meta-analysis of 82 studies. Acta Psychiatr Scand. 2017;135:373–87. Additional Declarations No competing interests reported. Supplementary Files Supplementaryfiles.docx TableS1.docx TableS2.xlsx TableS3.xlsx Cite Share Download PDF Status: Published Journal Publication published 13 Sep, 2024 Read the published version in BMC Medical Genomics → Version 1 posted Editorial decision: Revision requested 29 Aug, 2024 Reviews received at journal 14 Aug, 2024 Reviewers agreed at journal 28 Jun, 2024 Reviewers agreed at journal 28 Jun, 2024 Reviews received at journal 26 Jun, 2024 Reviewers agreed at journal 24 Jun, 2024 Reviewers agreed at journal 29 Jan, 2024 Reviewers invited by journal 29 Dec, 2023 Editor assigned by journal 29 Dec, 2023 Editor invited by journal 26 Dec, 2023 Submission checks completed at journal 26 Dec, 2023 First submitted to journal 22 Dec, 2023 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3791877","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":263656776,"identity":"5d0da0f8-f532-4666-9c93-5a05e797e7aa","order_by":0,"name":"Rui Mao","email":"","orcid":"","institution":"Central South University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Rui","middleName":"","lastName":"Mao","suffix":""},{"id":263656780,"identity":"3da9c467-2028-43f2-b7fb-fdaf4efbb5fa","order_by":1,"name":"Ji Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAw0lEQVRIiWNgGAWjYBACPmaGhANAWg7CZSNCCxtUizEJWqB0YgPxWtgZHh5gbDuc3t9/xoDhQ9lhBv7ZDcQ47Mzh3Bk3cgwYZ5w7zCBx5wAxWioO526Q4DFg5m07zGAgkUCMFoPD6Qb8ZwyY/xKvpeJwggFDjgEzI/FazqQbzriRVnCw51w6j8QNAlr4+c8kf2Bss5bn7z+88cGPMms5/hkEtDAw8CQw/2FoBjMPgLiE1AMBO0hhHREKR8EoGAWjYMQCAOyvPRjcop7nAAAAAElFTkSuQmCC","orcid":"","institution":"Central South University","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Ji","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2023-12-22 12:14:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3791877/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3791877/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12920-024-02008-0","type":"published","date":"2024-09-13T15:58:31+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":49150507,"identity":"83f26a9d-25be-4f55-a782-ca1ccc242dc0","added_by":"auto","created_at":"2024-01-04 00:08:29","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":439915,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of neurogenic and non-neurogenic rosacea by unsupervised clustering. (A)\u003c/strong\u003e Clustering heatmap; panel (\u003cstrong\u003eB\u003c/strong\u003e) shows the cumulative distribution function with different values of K, which is used to determine the approximate maximum value of CDF when the value of K is taken. At this time, the result of cluster analysis is the most reliable, usually taking the value of K with a small slope of CDF. Panel (\u003cstrong\u003eC\u003c/strong\u003e) shows the relative change of the area under the CDF curve between K and K minus 1; panels (\u003cstrong\u003eD\u003c/strong\u003e) and (\u003cstrong\u003eE\u003c/strong\u003e) show the consistent histogram and heat map of the group, respectively; panel (\u003cstrong\u003eF\u003c/strong\u003e) shows a histogram of NR-related biomarkers differentially expressed between RHNAG and RLNAG patients. *: P \u0026lt; 0.05; **: P \u0026lt; 0.01; ***: P \u0026lt; 0.001; ****: P \u0026lt; 0.0001.\u003c/p\u003e","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/85812a156047e9c427c4a81d.jpg"},{"id":49150508,"identity":"bcabc416-775f-44f5-ba2c-c29922582bcf","added_by":"auto","created_at":"2024-01-04 00:08:29","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":503719,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe abundance of inflammatory cell infiltration in RLNAG was significantly higher than that in RHNAG. \u003c/strong\u003eAccording to the expression of 43 NR biomarkers, PCA was performed in the cohort without normal samples (\u003cstrong\u003eA\u003c/strong\u003e) and the cohort containing 39 normal samples (\u003cstrong\u003eB\u003c/strong\u003e). We compared the abundance of nerve-related biomarkers expressed between the RHNAG and RLNAG, according to the predicted results of Xcell (\u003cstrong\u003eC\u003c/strong\u003e). According to the predicted results of ssGSEA, the differences in 26 immunocytes infiltration abundance between the RHNAG group and the RLNAG group were compared (\u003cstrong\u003eD and F\u003c/strong\u003e). (\u003cstrong\u003eE\u003c/strong\u003e) Columnar accumulation map of 26 types of immune cells’ infiltration abundance predicted by ssGSEA in 84 patients with rosacea. *: P \u0026lt; 0.05; **: P \u0026lt; 0.01; ***: P \u0026lt; 0.001; ****: P \u0026lt; 0.0001.\u003c/p\u003e","description":"","filename":"Figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/652d8e6f4c6f5035f1fc863b.jpg"},{"id":49150513,"identity":"625913e5-b8e5-40f5-9030-de2199c450dd","added_by":"auto","created_at":"2024-01-04 00:08:29","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":579081,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConstruction of the diagnostic model of RHNAG. A.\u003c/strong\u003e Histograms of standardized importance scores of 54 biomarkers that are very important for diagnostic typing obtained by random forest analysis; \u003cstrong\u003eB\u003c/strong\u003e. The histogram of the biomarkers of the top 20 importance scores was obtained by random forest analysis. The red ones are markers with an importance score greater than 50; \u003cstrong\u003eC\u003c/strong\u003e. ROC analysis results of 16 biomarkers. \u003cstrong\u003eD and E\u003c/strong\u003e. The result chart of LASSO regression analysis of 10 biomarkers with AUC value greater than 0.9. The red dotted line in the middle indicates that the lambda value of our selection is 4, so we have four genes to model; \u003cstrong\u003eF\u003c/strong\u003e. Analysis of clinical decision curve of the whole model and single molecular model. The y-axis measures the net benefit. The black line represents the non-remission risk nomogram. The thin solid line represents the assumption that all patients are in RHNAG. The thick solid line represents the assumption that all patients are in RLNAG. The decision curve shows that if the threshold probability of a patient and a doctor is \u0026gt;2% and \u0026lt;100%, respectively, using this RHNAG nomogram in the current study to diagnose RHNAG adds more benefit than the intervention-all scheme or the intervention-none scheme; \u003cstrong\u003eG\u003c/strong\u003e. Diagnostic model based on the expression of four biomarkers; on the right is the score corresponding to the expression of a specific biomarker; \u003cstrong\u003eH\u003c/strong\u003e. Calibration curves of the RHNAG diagnostic nomogram in the primary cohorts.\u003c/p\u003e","description":"","filename":"Figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/ab5fddc456a83ab395253103.jpg"},{"id":49150562,"identity":"a729953f-f54c-49dd-9792-b4970a92c696","added_by":"auto","created_at":"2024-01-04 00:16:29","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":420950,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe model has high sensitivity and specificity in the diagnosis of NR. A. \u003c/strong\u003eTypical facial photos of five patients with neurogenic rosacea; B. Typical facial photos of five patients with non-neurogenic rosacea; \u003cstrong\u003eC\u003c/strong\u003e. We performed ROC analysis on the total score of the model in a cohort of 84 patients with rosacea. \u003cstrong\u003eD\u003c/strong\u003e. We performed ROC analysis on the total score of the model in a clinical validation cohort of 46 patients with rosacea (NR = 18, NNR = 28).\u003c/p\u003e","description":"","filename":"Figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/e59d3f696a6ff4c2248dac9c.jpg"},{"id":64619751,"identity":"23ef96d6-e337-45f7-afbb-8b3a3d92cacc","added_by":"auto","created_at":"2024-09-16 16:17:08","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2957160,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/1ffce686-e044-4b2e-ac3b-0541fd229b51.pdf"},{"id":49150514,"identity":"cb43c3a7-3273-4755-8575-1c734fd00959","added_by":"auto","created_at":"2024-01-04 00:08:29","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":5321254,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryfiles.docx","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/cf45c725c1473e19961436be.docx"},{"id":49150511,"identity":"4efce91f-c8e4-410d-9913-5bb04a0e9241","added_by":"auto","created_at":"2024-01-04 00:08:29","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":21618,"visible":true,"origin":"","legend":"","description":"","filename":"TableS1.docx","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/a103860934199d42cb59f072.docx"},{"id":49150509,"identity":"395ddd9f-ab3e-411c-a5cb-80aa8fa699ce","added_by":"auto","created_at":"2024-01-04 00:08:29","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":22857,"visible":true,"origin":"","legend":"","description":"","filename":"TableS2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/6b65a24076673e838318a2ae.xlsx"},{"id":49150563,"identity":"b3448166-b53c-4ddb-9a53-5f2f4db00db1","added_by":"auto","created_at":"2024-01-04 00:16:29","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":9525,"visible":true,"origin":"","legend":"","description":"","filename":"TableS3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-3791877/v1/ff3d23ea524739bd5920889e.xlsx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Construction of a molecular diagnostic system for neurogenic rosacea by combining transcriptome sequencing and machine learning","fulltext":[{"header":"Introduction","content":"\u003cp\u003eRosacea is a prevalent chronic inflammatory cutaneous disorder of the face. Epidemiological surveys report its prevalence to range between 0% and 22%. (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) Often manifested as erythema likened to a \u0026ldquo;drunken\u0026rdquo; face, accompanied by sensations of burning and tingling, rosacea markedly impacts the patient's appearance, quality of life, and psychological well-being.(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) A subset of rosacea patients exhibits distinct clinical features, characterized by pronounced facial redness, burning, tingling, and sensory disturbances extending beyond mere inflammation. In 2011, Scharschmidt designated this subgroup as neurogenic rosacea (NR),(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e) based on sensations rather than cutaneous manifestations. NR patients commonly present with neurological or neuropsychiatric conditions such as complex regional pain syndrome, idiopathic tremor, depression, and obsessive-compulsive disorder.(\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e) Conventional rosacea treatments like metronidazole,(\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e) isotretinoin,(\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e) and oral tetracycline,(\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e) are largely ineffective for NR, while neurological treatments, including oral gabapentin, pregabalin, and tricyclic antidepressants,(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e) may yield superior therapeutic outcomes. Despite this, NR is not internationally recognized, and diagnosis, largely reliant on subjective clinical judgment, lacks an objective molecular diagnostic model. This oversight, particularly among primary care dermatologists, often leads to inappropriate treatments and exacerbation of NR symptoms, underscoring the pressing need for precise NR diagnosis.\u003c/p\u003e \u003cp\u003eAlthough the pathogenesis of rosacea remains elusive, factors including innate immune system imbalance,(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e) gut microbiota,(\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e) physical triggers like ultraviolet radiation,(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e) skin barrier dysfunction,(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e) and aberrant neurovascular signaling (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e) are implicated in rosacea's onset and progression. The role of neurovascular homeostasis in rosacea has gained recognition, particularly regarding the hypersensitivity of neural responses in NR patients and prevalent neurological complications(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). This highlights the significance of neurovascular dysfunction in NR, warranting exploration of the molecular mediators involved. Two major molecular categories are involved in mediating neural functions in rosacea: ion channel-associated proteins (including the Mas-related G protein-coupled receptors (Mrgpr) family and TRP channels) and neuropeptides (such as substance P and calcitonin gene-related peptide (CGRP)).(\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e) The activation of neurogenic TRP channels may stimulate skin vascular systems, leading to flushing, a hallmark rosacea feature. Specific G-protein\u0026ndash;coupled receptors in the Mrgpr family are principally engaged in cutaneous neurogenic inflammation, facilitating mast cell communication with sensory nerves and cutaneous cells. (\u003cspan additionalcitationids=\"CR19\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e) These molecules might be instrumental in NR development.\u003c/p\u003e \u003cp\u003eBiomarker-based diagnostics, owing to its objectivity and accuracy, is employed across a range of diseases. (\u003cspan additionalcitationids=\"CR22\" citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e) Construction of disease-specific biomarkers and auxiliary diagnostic models have become the prevailing diagnostic methodology. In this context, our study assembled 43 biomarkers, encompassing mRNAs encoding neuropeptides like substance P, PACAP, VIP, and CRGP, as well as mRNAs for ion channel-related proteins like TRP channels and the Mrgpr family. Through clustering, we differentiated two patient groups and constructed an auxiliary diagnostic model to facilitate NR diagnosis.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eClinical diagnosis\u003c/h2\u003e \u003cp\u003eThree seasoned dermatologists from the Xiangya Hospital's dermatology department, well-versed in diagnosing and treating rosacea, collaborated with three highly experienced psychiatric clinicians from the same hospital to categorize 46 rosacea patients in our cohort. The diagnostic criteria for NR encompassed:: First, rosacea could be diagnosed according to the 2017 edition of the International Diagnostic guidelines (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e) for rosacea and the patient\u0026rsquo;s facial symptoms and other clinical data. Second, there were no obvious papules and pustules in the patient's facial photos. Third, the patient had obvious facial redness, burning, tingling, and sensory disorders beyond blushing or inflammation. Fourth, patients might have had neurological or neuropsychiatric disorders, including complex regional pain syndrome, idiopathic tremor, depression, and obsessive-compulsive disorder.(\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e) Fifth, topical or oral metronidazole, isotretinoin, and other first-line treatment drugs were ineffective, but the symptoms significantly improved after using gabapentin or tricyclic antidepressants.(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e) Finally, demodex acne, contact dermatitis, and other differential diagnoses were excluded. Participants had to meet the above six diagnostic criteria at the same time to be diagnosed as NR. We used a questionnaire survey and physician evaluation to evaluate the clinical phenotype of patients with rosacea. Specifically, we used the Global flush severity score (GFSS) scale to evaluate paroxysmal flashes.(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e) The researcher erythema rating scale (CEA)(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e) and the patient erythema self-rating scale (PSA)(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e) were used to evaluate persistent erythema. The evaluation of papules and pustules was carried out by the combination of \u0026ldquo;Statistics of the number of papules and pustules\u0026rdquo; and \u0026ldquo;researcher Global Evaluation (IGA) scale.\u0026rdquo; The psychological burden of patients with rosacea was mainly assessed by the Chinese version of the Rosacea Quality of Life (RosaQoL) scale, the Dermatology Life Quality Index (DLQI),(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e) and the Penn State Worry Questionnaire (PSWQ).(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e) All evaluation forms are available in \u003cb\u003eTable \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e.\u003c/p\u003e \u003cp\u003eThe clinical data (\u003cb\u003eTable \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e)\u003c/b\u003e of 46 patients with rosacea are provided as \u003cb\u003esupplementary data\u003c/b\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eData acquisition and integration\u003c/h2\u003e \u003cp\u003eOur team's previous research collected skin biopsy samples from the central facial area of patients diagnosed with rosacea, aged between 20 and 60, as well as from healthy females undergoing cosmetic procedures (referred to as the HS group), at the Dermatology Department of Xiangya Hospital, Central South University. The collection period spanned from June 1, 2014, to April 4, 2020. This study involved 46 clinically and pathologically diagnosed rosacea patients and 19 age-similar HS participants. Post-collection, all samples were immediately stored at \u0026minus;\u0026thinsp;80\u0026deg;C pending analysis. Every sample and its associated clinical information were gathered with informed consent. The study's protocol underwent rigorous review and received approval from the Xiangya Hospital Central South University's ethics review board. all methods were performed in accordance with the relevant guidelines and regulations. Our sequencing data from rosacea patients were deposited in the genome sequence archive under accession number HRA000379 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://bigd.big.ac.cn/gsa-human/\u003c/span\u003e\u003cspan address=\"http://bigd.big.ac.cn/gsa-human/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The GSE65914 datasets (microarray chip ), which contain 38 rosacea and 20 normal samples, were downloaded from the GEO (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/gds/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/gds/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) database with a log2 transform and quantile normalization matrix. We used the combat function of R package \u0026ldquo;sva\u0026rdquo; to combine the GSE65914 dataset and our local dataset and remove the batch effect.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eUnsupervised clustering\u003c/h2\u003e \u003cp\u003eUtilizing the expression patterns of the 43 identified biomarkers (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), we embarked on an unsupervised cluster analysis for the 84 rosacea patients via the ConsensusClusterPlus software package (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Detailed information about these 43 biomarkers is shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Using agglomerative km clustering with a Pearson correlation distance of 1 and resampling 80% of the samples for 10 repetitions, the number and stability of clusters were determined according to the consensus clustering algorithm.(\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e) To enhance the stability of classification, this process was repeated 1,000 times.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDetailed information of 43 neurogenic rosacea related genes\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSYMBOL\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSEMBL\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eENTREZID\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGENENAME\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADCYAP1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000141433\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e116\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadenylate cyclase activating polypeptide 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADCYAP1R1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000078549\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e117\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eADCYAP receptor type I\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000148926\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e133\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenomedullin\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADM2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000128165\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e79924\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenomedullin 2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRA1A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000120907\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e148\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor alpha 1A\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRA1B\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000170214\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e147\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor alpha 1B\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRA1D\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000171873\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e146\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor alpha 1D\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRA2A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000150594\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e150\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor alpha 2A\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRA2B\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000274286\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e151\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor alpha 2B\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRA2C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000184160\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e152\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor alpha 2C\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRB1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000043591\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e153\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor beta 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRB2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000169252\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e154\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor beta 2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eADRB3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000188778\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e155\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eadrenoceptor beta 3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCALCA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000110680\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e796\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ecalcitonin related polypeptide alpha\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCALCR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000004948\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e799\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ecalcitonin receptor\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCALCRL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000064989\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e10203\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ecalcitonin receptor like receptor\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDRD1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000184845\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1812\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003edopamine receptor D1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eF2RL1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000164251\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2150\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eF2R like trypsin receptor 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eF2RL3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000127533\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e9002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eF2R like thrombin or trypsin receptor 3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHRH1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000196639\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3269\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ehistamine receptor H1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHRH2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000113749\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3274\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ehistamine receptor H2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHRH3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000101180\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e11255\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ehistamine receptor H3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHRH4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000134489\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e59340\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ehistamine receptor H4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHTR2A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000102468\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3356\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5-hydroxytryptamine receptor 2A\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHTR3A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000166736\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3359\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5-hydroxytryptamine receptor 3A\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMMEL1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000277131\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e79258\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003emembrane metalloendopeptidase like 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMRGPRX1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000170255\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e259249\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMAS related GPR family member X1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMRGPRX2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000183695\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e117194\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMAS related GPR family member X2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMRGPRX4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000179817\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e117196\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMAS related GPR family member X4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNGFR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000064300\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4804\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003enerve growth factor receptor\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNPFFR2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000056291\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e10886\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eneuropeptide FF receptor 2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNPY\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000122585\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4852\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eneuropeptide Y\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTAC1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000006128\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e6863\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003etachykinin precursor 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTACR1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000115353\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e6869\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003etachykinin receptor 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTRPA1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000104321\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e8989\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003etransient receptor potential cation channel subfamily A member 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTRPV2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000187688\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e51393\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003etransient receptor potential cation channel subfamily V member 2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTRPV3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000167723\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e162514\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003etransient receptor potential cation channel subfamily V member 3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTRPV4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000111199\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e59341\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003etransient receptor potential cation channel subfamily V member 4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTRPV5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000127412\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e56302\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003etransient receptor potential cation channel subfamily V member 5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTRPV6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000276971\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e55503\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003etransient receptor potential cation channel subfamily V member 6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVIP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000146469\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7432\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003evasoactive intestinal peptide\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVIPR1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000114812\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7433\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003evasoactive intestinal peptide receptor 1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVIPR2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eENSG00000106018\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7434\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003evasoactive intestinal peptide receptor 2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eGene differential expression analysis\u003c/h2\u003e \u003cp\u003eThe linear model for microarray analysis (Limma) (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e) is a differential expression screening method based on the generalised linear model. Here, we used the R software package limma (version 3.40.6) for differential analysis to obtain differential genes between different comparison and control groups. We used the criterion that the absolute value of the fold change (FC) is greater than 1.5 and the adjusted-P value is less than 0.05 to screen for significant differentially expressed genes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003ePrincipal component analysis (PCA)\u003c/h2\u003e \u003cp\u003eWe executed PCA through the \"FactoMineR\" package of R software. Briefly, we first standardized the data, then calculated the eigenvalues and variances, and simultaneously calculated the contribution of each variable to PC. Finally, the PCA results were visualised through the \"factoextra\" package.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eAnalysis of immune cell infiltration\u003c/h2\u003e \u003cp\u003eBased on the merged data matrix, we used the ssGSEA method of the GSVA package (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e), MCPcounter (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e), and Xcell (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://xcell.ucsf.edu/\u003c/span\u003e\u003cspan address=\"https://xcell.ucsf.edu/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e) to calculate the abundance of immune cells and excluded the samples with P\u0026thinsp;\u0026gt;\u0026thinsp;0.05. Finally, the Mann\u0026ndash;Whitney U test was used to analyse the differences in immune cell subtypes among different groups.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eGene enrichment analysis and gene set enrichment analysis\u003c/h2\u003e \u003cp\u003eWe used the clusterProfiler (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e) package for GO-BP and KEGG enrichment analysis, and the cutoff values of the P value and adjusted-P value were 0.05. There were fewer than three mRNAs covered in the enrichment path. The clusterProfiler (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e) package was used for gene set enrichment analysis (GSEA). The input files were gene names, and their values of log\u003csub\u003e2\u003c/sub\u003e (fold change) were between the high- and low-risk groups.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e\u003cb\u003eWeighted gene co-expression network analysis (WGCNA)\u003c/b\u003e\u003c/h2\u003e \u003cp\u003eFirst, based on the gene expression profile, we calculated the Median Absolute Deviation (MAD) of each gene, eliminated the top 50% of the genes with the smallest MAD, removed the outlier genes and samples by using the good Sample Genes method of the R software package WGCNA (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e), and further used WGCNA to construct a scale-free co-expression network. Specifically, first, Pearson's correlation matrices and average linkage method were performed for all pair-wise genes. Then, a weighted adjacency matrix was constructed using the power function A_mn=|C_mn|^β (C_mn\u0026thinsp;=\u0026thinsp;Pearson's correlation between Gene_m and Gene_n; A_mn\u0026thinsp;=\u0026thinsp;adjacency between Gene m and Gene n). Here, β is a soft-thresholding parameter that emphasises strong correlations between genes and penalises weak correlations. After choosing the power of 4, we transformed the adjacency into a topological overlap matrix (TOM), which could measure the network connectivity of a gene defined as the sum of its adjacency with all other genes for the network gene ratio, and the corresponding dissimilarity (1-TOM) was calculated. To classify genes with similar expression profiles into gene modules, average linkage hierarchical clustering was conducted according to the TOM-based dissimilarity measure, with a minimum size (Gene group) of 30 for the genes\u0026rsquo; dendrogram. We set the sensitivity to 3. To further analyse the module, we calculated the dissimilarity of module eigen genes, chose a cut line for the module dendrogram and merged some modules. In addition, we merged the modules with a distance of less than 0.25 and finally obtained three co-expression modules. Note that the grey module is considered a collection of genes that cannot be assigned to any module.\u003c/p\u003e \u003cp\u003eWe calculated the correlation between the module and the sample characters and drew the correlation heatmap. The modules related to traits were mined according to the correlation between traits, module feature vector genes, and the p-value. According to the significance of the gene network, that is, the mean value of the correlation between the trait and each gene expression in each module as the significance of the trait in this module, the module with the greatest significance was most related to the trait. Looking for the hub gene (hub genes) related to this trait, we first calculated the internal connectivity and module identity of the gene. The internal connectivity measures the status of the gene within the module, while the module identity indicates to which module the gene belongs.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eProtein-protein interaction analysis\u003c/h2\u003e \u003cp\u003eWe used the Metascape (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://metascape.org/gp/index.html#/main/step1\u003c/span\u003e\u003cspan address=\"https://metascape.org/gp/index.html#/main/step1\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) website (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e) for protein-protein (PPI) analysis of the modular hub gene, and the PPI analysis of the website was based on the Molecular Complex Detection (MCODE (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e)) tool. Finally, we used Cytoscape (version 3.8.2) to visualise the PPI network.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eDrug prediction\u003c/h2\u003e \u003cp\u003eThe Connectivity Map is a growing resource of over 3\u0026nbsp;million perturbational profiles available to the wider scientific community. The clue.io platform (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://clue.io\u003c/span\u003e\u003cspan address=\"https://clue.io\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.) offers a range of apps to facilitate the analysis of these data. Based on the QUERY CMap module on the CLUE website, we uploaded the hub gene in the WGCNA module and set the query parameters to \"Gene expression (L1000).\"\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eRandom forest analysis\u003c/h2\u003e \u003cp\u003eWe used the expression matrix of 696 differential genes and 84 rosacea samples as input files and standardized the data. After setting the random number seed, we analyzed the data with the \u0026ldquo;randomForest\u0026rdquo; R software package. When building 500 trees, the error rate estimated by OOB was 5.3%. The experimental and control group's classification error rates were 0.02 and 0.06, respectively. Then, we used the \u0026ldquo;caret\u0026rdquo; package to split the training and test sets and the \u0026ldquo;Boruta\u0026rdquo; package to select and identify the key classification variables. A total of 51 important variables, 40 potentially important variables (tentative variable, no statistical difference between the importance score and the best shadow variable score), and 605 unimportant variables were identified. Next, we used the \u0026ldquo;dplyr\u0026rdquo; package to define a function to extract the important values corresponding to each variable and used the \u0026ldquo;ImageGP\u0026rdquo; package to draw important variable results. Finally, we used the \u0026ldquo;Caret\u0026rdquo; package cross-validation to select parameters and fit the model; specifically, we defined a function to generate some columns of mtry for testing (a series of values not greater than the total number of variables), select data related to key feature variables, and select data related to key feature variables.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eConstruction of the diagnostic model\u003c/h2\u003e \u003cp\u003eThe least absolute shrinkage and selection operator (LASSO) method, which is suitable for a reduction in high-dimensional data,(\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e) was used to select the best predictive characteristics. Features with nonzero coefficients in the LASSO regression model were selected.(\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e) All of the potential predictors selected by lasso analysis were applied to develop a predicting model for the diagnosis of NR (R packages \u0026ldquo;glmnet\u0026rdquo; and \u0026ldquo;rms\u0026rdquo;).(\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e)\u003c/p\u003e \u003cp\u003eCalibration curves were plotted to assess the calibration of the diagnosis nomogram (R package \u0026ldquo;rms\u0026rdquo;). We employed the bootstrap method to assess both the stability and performance of a given model. To quantify the discrimination performance of the diagnosis nomogram, Harrell\u0026rsquo;s C-index was measured (R package \u0026ldquo;rms\u0026rdquo;).(\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e) Decision curve analysis was conducted to determine the clinical usefulness of the diagnosis nomogram by quantifying the net benefits at different threshold probabilities in the cohort.(\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e) The net benefit was calculated by subtracting the proportion of all of the patients with false-positive results from the proportion of patients with true-positive results and by weighing the relative harm of forgoing interventions compared with the negative consequences of an unnecessary intervention (R packages \u0026ldquo;rms\u0026rdquo; and \u0026ldquo;rmda\u0026rdquo;). Receiver operating characteristic (ROC) curve analysis was employed to compare predictions concerning the sensitivity and specificity (R packages \u0026ldquo;pROC\u0026rdquo; and \u0026ldquo;rms\u0026rdquo;).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eData processing\u003c/h2\u003e \u003cp\u003eWe integrated GSE65914 with our own sequencing cohort after removing the batch effect through the sva package as a training cohort. As delineated in \u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA and S1B\u003c/b\u003e's bar chart, post-elimination of the batch effect, the data distribution between the two datasets demonstrated a trend towards uniformity with the median lying on the same axis Analysis through UPSET(\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e) revealed the intersection of the two cohorts yielding a total of 18,384 common protein-coding genes (\u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eC\u003c/b\u003e). The density maps (\u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eD and S1E\u003c/b\u003e) reveal significant discrepancies in the sample distribution across each dataset prior to batch effect removal, indicative of a batch effect. Subsequent to this removal, data distribution within each dataset appeared congruent, with analogous mean and variance. Furthermore, the PCA diagrams (\u003cb\u003eFigure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eF and S1G\u003c/b\u003e) illustrated that prior to batch effect removal, the dataset samples were homogeneously clustered, suggestive of a batch effect. However, post-removal, samples from each dataset were both clustered and interspersed, a strong indication of effective batch effect elimination. The final product was a data matrix comprised of 123 samples and 18,384 protein-coding genes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of Two Distinctive Clusters through Unsupervised Clustering\u003c/h2\u003e \u003cp\u003eCluster analysis was carried out utilizing using ConsensusClusterPlus (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). The optimal number of clusters was discerned via the empirical cumulative distribution function (CDF) plot (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC) and the evaluation of average intra-group consistency (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eE). The analysis revealed that at K\u0026thinsp;=\u0026thinsp;2, intra-group consistency was at its zenith, thereby resulting in the most efficacious clustering. Consequently, 84 rosacea patients were categorized into two clusters: Cluster 1 (n\u0026thinsp;=\u0026thinsp;46) and Cluster 2 (n\u0026thinsp;=\u0026thinsp;38). The expression differences of 43 established neurovascular loop biomarkers in rosacea between clusters 1 and 2 were then assessed. A total of 17 biomarkers were discovered to be highly expressed in Cluster 1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eF). These included nerve growth factor receptor (NGFR), tachykinin precursor 1 (TAC1), adrenomedullin (ADM and ADM2), adrenoceptor (ADRA1A, ADRB1, and ADRA2A), calcitonin (CALCA and CALCRL), histamine (HRH1, HRH2, and HTR3A), and TRPV family genes (TRPA1, TRPV2, TRPV3, and TRPV6). TAC1 encodes four products of the tachykinin family: substance P and neurokinin A, and neuropeptide K and neuropeptide γ. These hormones are thought to be neurotransmitters that interact with neuroreceptors and smooth muscle cells. It is well known that they induce behavioral responses and act as vasodilators and secretagogues.(\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e) CGRP induces vasodilation. It dilates various blood vessels, including coronary arteries, brain arteries, and systemic vascular system. The anchor-like receptor family (TRPA1) is a member of the ion channel transient receptor potential (TRP) superfamily. Receptor-activated non-selective cationic channels are involved in pain detection and may also be involved in cold perception (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e) and itching.(\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e)\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe abundance of inflammatory cell infiltration in cluster 2 was significantly higher than that in cluster 1\u003c/b\u003e \u003c/p\u003e \u003cp\u003eInitially, PCA was performed on 84 rosacea patients using 43 biomarkers (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA), followed by analysis on a complete cohort (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB) that included the aforementioned patients and 39 normal samples. The results showed that these 43 biomarkers were able to distinguish between cluster 1, cluster 2, and normal people in the two cohorts. According to the predicted results of Xcell (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://xcell.ucsf.edu/\u003c/span\u003e\u003cspan address=\"https://xcell.ucsf.edu/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e), the abundance of nerve-related biomarkers in cluster 1 was significantly higher than in cluster 2 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). Subsequently, ssGSEA(\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e) was employed to assess differences in immune cell abundance between the clusters. The results indicated that the immuno-inflammatory cell abundance in Cluster 1 was significantly reduced compared to Cluster 2 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE). Specifically, most immune cells, including B cells, CD4-positive T cells, CD8-positive T cells, macrophages, mast cells, natural killer cells, and neutrophils were considerably less abundant in cluster1 than in cluster2 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eF). Consistency in immunocyte infiltration abundance was observed in the MCP (\u003cb\u003eFigure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/b\u003e) and Xcell (\u003cb\u003eFigure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e\u003c/b\u003e) analyses, with Cluster 1 significantly lower than Cluster 2.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe pathways in cluster 1 are mostly concentrated in the synthesis, transmission, and function of neurotransmitters\u003c/b\u003e \u003c/p\u003e \u003cp\u003eIn an endeavor to delineate the disparities in functionality and pathway engagement between clusters 1 and 2, Gene Set Enrichment Analysis (GSEA) was executed for both groups. Utilizing the KEGG database, the GSEA findings revealed that the functional pathways within cluster 1 predominantly encompass the synaptic vesicle cycle, cholesterol metabolism, valine, leucine and isoleucine degradation, in addition to the biosynthesis of the amino acid pathway (\u003cb\u003eFigure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eA\u003c/b\u003e). Moreover, the GSEA analysis grounded in the GO database ascertained that the pathways of synaptic signaling, cellular amino acid metabolic process, and cholesterol biosynthesis are predominantly augmented in cluster 1 (\u003cb\u003eFigure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eB\u003c/b\u003e). Conversely, the GSEA-KEGG and GSEA-GO results of cluster 2 unveiled that the functional pathways within this cluster are primarily associated with immune- and inflammation-related processes, including but not limited to the IL\u0026ndash;17 signaling pathway, NF\u0026ndash;kappa B signaling pathway, TNF signaling pathway, positive regulation of inflammatory response to an antigenic stimulus, regulation of CD8-positive, alpha\u0026ndash;beta T cell activation, and Th1 and Th2 cell differentiation (\u003cb\u003eFigure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eC and S4D\u003c/b\u003e). The foregoing results underscore that the pathways in cluster 1 are overwhelmingly concentrated on the synthesis, transmission, and functionalization of neurotransmitters, whereas the functionality of cluster 2 is largely oriented toward inflammation and immune responsiveness.\u003c/p\u003e \u003cp\u003eSubsequently, we categorized cluster 1 as the rosacea group demonstrating a conspicuous neuromolecular phenotype (RHNAG) and cluster 2 as the rosacea group lacking a neuromolecular phenotype (RLNAG). As delineated in \u003cb\u003eTable \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/b\u003e, the onset symptoms of RHNAG patients predominantly manifested as paroxysmal flushing (n\u0026thinsp;=\u0026thinsp;16; 66.7%), while a minority exhibited persistent erythema (n\u0026thinsp;=\u0026thinsp;1; 4.2%) or papular pustules (n\u0026thinsp;=\u0026thinsp;6; 25.0%). Conversely, the RLNAG patients chiefly presented with papular pustules (n\u0026thinsp;=\u0026thinsp;15; 68.2%), with a smaller proportion experiencing paroxysmal flushing (n\u0026thinsp;=\u0026thinsp;7; 31.8%). Furthermore, an evaluation of the Global Flush Severity Score (GFSS) scale revealed significantly elevated scores in RHNAG patients relative to their RLNAG counterparts. Assessments employing the CEA, PSA, and IGA scales intimated that erythema was more pronounced in RHNAG patients, whereas papular pustules were markedly less prevalent compared to RLNAG patients. Noteworthy is the fact that according to the Dermatology Life Quality Index (DLQI), RHNAG patients suffered from a diminished quality of life compared to RLNAG patients, and the Penn State Worry Questionnaire (PSWQ) indicated exacerbated levels of anxiety and depression within the RHNAG group. Collectively, these findings elucidate that the phenotype of RHNAG, characterized by elevated expression of neurotransmitters and receptor-associated markers, aggravated mental health conditions, and comparatively subdued inflammation, bears significant resemblance to that of NR in a clinical context.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eScreening the gene modules most related to RHNAG or RLNAG by WGCNA\u003c/h2\u003e \u003cp\u003eIn the quest to isolate the gene modules most pertinent to RHNAG or RLNAG, we scrutinized the genetic variances between these patient cohorts. Our examination unveiled a total of 1071 differentially expressed genes, comprising 411 genes that were upregulated in the RHNAG group and 660 genes that were upregulated in the RLNAG group (\u003cb\u003eFigure \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003eA and S5B\u003c/b\u003e). Pursuant to this, we employed Weighted Gene Co-expression Network Analysis (WGCNA) on 3379 genes, all of which exhibited a variance exceeding 0.2 across the samples, and subsequently segmented these genes into four distinct modules (\u003cb\u003eFigure S6A-S6D\u003c/b\u003e). Correlational analysis between the modules and clinical attributes revealed that the blue module exhibited positive correlations with the RHNAG group and negative correlations with immune cell abundance, including T cells, cytotoxic lymphocytes, and natural killer cells (\u003cb\u003eFigure S6E\u003c/b\u003e). In contrast, the turquoise module manifested a pronounced positive correlation with the RLNAG group and an abundance of immune cell infiltration. The eigengene distance heatmap showcased the maximum distance between the blue and turquoise modules, indicative of a substantial divergence in gene expression patterns (\u003cb\u003eFigure S6F\u003c/b\u003e). Furthermore, the scatter plot correlating Gene Significance (GS) with Module Membership (MM) confirmed that the genes most intimately associated with the RHNAG group were also pivotal within the blue module (\u003cb\u003eFigure S6G\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003cb\u003eThe function of the hub gene in the blue module is mainly concentrated in the synthesis and function of neurotransmitters\u003c/b\u003e \u003c/p\u003e \u003cp\u003eTo further explore the function of hub genes in the blue and turquoise modules, we carried out PPI analysis and functional enrichment analysis of hub genes in the two modules. The analysis results of MCODE software show that the hub genes of the blue module were mainly focused on the synthesis and function of neurotransmitter materials, such as cholesterol synthesis, amino acid synthesis, and steroid metabolism (\u003cb\u003eFigure S7A)\u003c/b\u003e. The results of KEGG and GO-BP enrichment analysis of the hub genes of the blue module also indicate that the functions of these genes are mainly focused on the pathways related to the synthesis and function of neurotransmitters such as amino acid metabolism, cholesterol metabolism, and steroid metabolism (\u003cb\u003eFigures S7B and S7C)\u003c/b\u003e. On the contrary, the functions of the core genes of the turquoise module are mainly concentrated in immune and inflammatory pathways, such as activation of the immune response, T cell receptor signaling pathway, JAK\u0026ndash;STAT signaling pathway, NF\u0026ndash;kappa B signaling pathway, and positive regulation of T cell activation (\u003cb\u003eFigures S8A\u0026ndash;S8C\u003c/b\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eConstruction of the diagnostic model of RHNAG and RLNAG\u003c/h2\u003e \u003cp\u003eInitially, a random forest analysis was executed on a set of 696 differentially expressed genes contrasting RHNAG and RLNAG, from which 51 pivotal genes were identified as key differentiators between the phenotypes of RHNAG and RLNAG (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). Subsequently, a scatterplot delineating the top 20 genes based on importance scores was generated (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). From this pool, 16 molecular entities boasting importance scores surpassing 50 were earmarked for ROC analysis. Out of these, 10 molecules demonstrating AUC values exceeding 0.9 were further shortlisted for lasso regression scrutiny (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC) The subsequent lasso regression pinpointed four genes\u0026mdash;HMGCR, CUX2, PLIN2, and PNPLA3\u0026mdash;from the aforementioned 10 for model construction, with a lambda value set at 4 \u003cb\u003e(\u003c/b\u003eFigs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD \u0026amp; \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE\u003cb\u003e)\u003c/b\u003e Utilizing the nomogram approach, a linear diagnostic prediction model for RHNAG was formulated based on these four genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eG). According to the algorithm in the package, the scores corresponding to the expression of the four molecules are obtained from the following linear formulas, respectively.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e(\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\text{s}\\text{c}\\text{o}\\text{r}\\text{e} \\text{o}\\text{f} \\text{P}\\text{N}\\text{P}\\text{L}\\text{A}3=-20.154(\\text{exp }of PNPLA3)+147.73\\)\u003c/span\u003e\u003c/span\u003e;\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\text{s}\\text{c}\\text{o}\\text{r}\\text{e} \\text{o}\\text{f} \\text{C}\\text{U}\\text{X}2=19.133(\\text{exp }of CUX2)-19.4$$\u003c/div\u003e\u003c/div\u003e;\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\text{s}\\text{c}\\text{o}\\text{r}\\text{e} \\text{o}\\text{f} \\text{P}\\text{L}\\text{I}\\text{N}2=-21.473(\\text{exp }of PLIN2)+181.47$$\u003c/div\u003e\u003c/div\u003e;\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\text{s}\\text{c}\\text{o}\\text{r}\\text{e} \\text{o}\\text{f} \\text{H}\\text{M}\\text{G}\\text{C}\\text{R}=-10.7(\\text{exp }of HMGCR)+120.85\\)\u003c/span\u003e \u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe formula for calculating the RHNAG diagnostic probability corresponding to the total score of the four molecules is as follows:\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$probability=1.06777*ln\\left(total score\\right)-19.4$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eFor example, the expression of HMGCR, CUX2, PLIN2, and PNPLA3 in patient numbered GSM1611085 was 8.86, 4.4, 7.29, and 6.78, corresponding respectively to 73, 36, 74, and 91 points according to the formula. Therefore, the patient's overall score was 274, corresponding to the probability of 0.984 according to the formula that they had RHNAG. The final result shows that the patient was indeed an RHNAG patient. Harrell\u0026rsquo;s C-index of this model was 0.9683. The results of BOOTSTRAT indicated a robust C-index of 0.908 (95% CI [0.901, 0.915]), suggesting high predictive accuracy and reliability of our model. The calibration curve analysis results show that the model's predicted value was close to the ideal (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eH). The above results indicate that the diagnostic model can predict whether a patient has the RHNAG type of rosacea. Finally, the results of DCA show that the benefits order of each of the four genes as a model to predict the clinical rate of return of RHNAG was HMGCR\u0026thinsp;\u0026gt;\u0026thinsp;CUX2\u0026thinsp;\u0026gt;\u0026thinsp;PLIN2\u0026thinsp;\u0026gt;\u0026thinsp;PNPLA3. The clinical benefit rate of the predictive model of the four genes was higher than that of each gene alone (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eThe model has high sensitivity and specificity in the diagnosis of NR\u003c/h2\u003e \u003cp\u003eIn line with the diagnostic categorization by three experienced dermatologists from Xiangya Hospital, 18 of the 46 rosacea patients were diagnosed with Neurogenic Rosacea (NR) while the remaining 28 were classified as Non-Neurogenic Rosacea (NNR) patients (\u003cb\u003eTable \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e\u003c/b\u003e). Visual documentation, comprising images of five NR and five NNR patients, revealed a notable absence of papules and pustules on the facial region of NR patients, replaced instead by pervasive flushing (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Conversely, NNR patients exhibited pronounced papules and pustules either locally or encompassing the entire visible region (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). Finally, the total scores of 84 patients with rosacea were calculated according to the diagnostic model, and the ROC analysis was carried out according to the total scores. The results show that the AUC for the nomogram was 0.9376 (95% confidence interval (CI): 0.8898\u0026ndash;0.9855; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), the best cutoff value was 156, and the sensitivity and specificity were 0.974 and 0.861, respectively, in the total cohort of 84 patients with rosacea (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). Remarkably, an isolated ROC analysis on the cohort of 46 rosacea patients (NR: n\u0026thinsp;=\u0026thinsp;18; NNR: n\u0026thinsp;=\u0026thinsp;28), as assessed by the trio of dermatologists, showcased an AUC of 0.9023 for NR diagnosis (95% CI: 0.826\u0026ndash;0.9925; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The prime cutoff value was determined to be 169, with a corresponding sensitivity and specificity of 0.926 and 0.870, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD). Cumulatively, these findings underscore the robustness and elevated precision of our diagnostic model in clinical applications.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003ePrediction of Therapeutic Agents\u003c/b\u003eLeveraging the cardinal genes identified within the blue and turquoise modules, potential therapeutic agents were prognosticated utilizing the Connectivity Map function on the clue.io platform. Agents with predictive scores beneath \u0026minus;\u0026thinsp;0.6 were earmarked as potential treatment options. Our analysis indicated that the prospective therapeutic agents for RHNAG, as gleaned from the blue module, predominantly belonged to classes such as dopamine receptor antagonists, norepinephrine inhibitors, tricyclic antidepressants, opioid receptor antagonists, and sigma receptor antagonists (Refer to \u003cb\u003eFigure S9A\u003c/b\u003e). Contrarily, the turquoise module's hub gene projections for RLNAG treatments were primarily aligned with mTOR inhibitors, JAK inhibitors, protein kinase inhibitors, MEK inhibitors, and STAT inhibitors (\u003cb\u003eFigure S9B\u003c/b\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, leveraging neurovascular biomarkers, we stratified rosacea patients into two distinct groups: RHNAG and RLNAG. Intriguingly, the RLNAG group demonstrated a markedly heightened abundance of immune cells compared to the RHNAG cohort. Comprehensive GSEA, KEGG, and GO-BP enrichment analyses revealed that genes specific to RHNAG are predominantly enriched in pathways involving synaptic neurotransmitters, phospholipids, unsaturated fatty acids, amino acids, and cholesterol metabolism. Conversely, RLNAG-specific genes predominantly aligned with immune and inflammatory pathways. This molecular distinction was mirrored clinically; RHNAG patients typically presented absent of facial papules and pustules, but with pronounced flushing, burning, and tingling sensations. Additionally, a significant portion of the RHNAG group manifested neurotic symptoms including anxiety, depression, and neuroticism, and standard treatments often proved ineffectual. In stark contrast, the RLNAG patients, as observed in the clinical validation cohort, predominantly exhibited facial papules and pustules with a scant occurrence of neurological symptoms such as sensory aberrations and depression. Notably, their symptoms were ameliorated by conventional treatments. A salient observation from the clinical validation cohort, as assessed by a trio of seasoned dermatologists, was the unanimous categorization of all NR patients into the RHNAG group, while the majority of NNR patients predominantly clustered within the RLNAG cohort. Thus, we postulate that RHNAG patients share striking clinical parallels with NR patients, primarily presenting with neurological symptoms. In juxtaposition, RLNAG patients align closely with NNR patients, typified by pronounced inflammation and the presence of facial papules and pustules. This concordance underscores why the RHNAG diagnostic model boasts such elevated sensitivity and specificity in NR diagnosis, reinforcing its potential as a diagnostic surrogate for NR.\u003c/p\u003e \u003cp\u003eCholesterol stands as the primary lipid collaborator with sphingolipids within membrane microdomains. Within the nervous system, cholesterol predominantly constitutes the primary lipid component of myelin (28%).(\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e) Importantly, pivotal synaptic transmission processes, such as endocytosis, exocytosis, and the lateral diffusion of neurotransmitter receptors within the membrane, are profoundly modulated by cholesterol concentrations.(\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e) The functionality of neurotransmitter receptors is modulated by lipid domains,(\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e) cholesterol,(\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e) and sphingolipids.(\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e) Cholesterol's modulatory effects on membrane receptor function are largely attributed to either its direct receptor interaction or its overarching influence on the biophysical attributes of the lipid bilayer of membranes.(\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e, \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e) In the nervous milieu, lipids emerge as the predominant organic compounds. Glycerol phospholipids, sphingolipids, and cholesterol primarily constitute the neural lipid repertoire within both central and peripheral domains.(\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e) Sphingolipids, glycerolipids, and their derivates (namely glycosphingolipids, GSLs) are intricately associated with neurogenesis, synaptic transmission, and the synthesis, functionality, and transport of neurotransmitter receptors (\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e). Within the architecture of our NR diagnostic model, 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) occupies a crucial position. HMGCR facilitates the conversion of (3S)-hydroxy-3-methylglutaryl-CoA (HMG-CoA) to mevalonic acid, representing the pivotal regulatory step in cholesterol synthesis and other isoprenoid formations, thereby orchestrating cellular cholesterol equilibrium.(\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e, \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e) Consequently, we postulate HMGCR's involvement in the synthesis of neurotransmitters and their receptor engagement, potentially mediated through cholesterol synthesis regulation, which may underscore its role in NR's pathogenesis. Further, the protein encoded by perilipin 2 (PLIN2) is aligned with the perilipin family, known for coating intracellular lipid storage vesicles, suggesting its potential as a lipid accumulation marker across diverse cellular environments and pathologies.(\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e) Another protein of interest, encoded by the patatin-like phospholipase domain containing 3 (PNPLA3), functions as a triacylglycerol lipase, mediating triacylglycerol hydrolysis in adipocytes and intricately involved in glycerol phospholipid biosynthesis. (\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e) In our investigations, these phospholipid biosynthesis-centric genes exhibited a positive correlation with the composite score of the NR diagnostic model, suggesting their potential to amplify the synthesis, transport, and functional attributes of neurotransmitters and their receptors, possibly through phospholipid and sphingomyelin synthesis, thereby elucidating their role in NR's etiology.The Cut-like homeobox 2 (CUX2) molecule, uniquely characterized by its negative correlation with the model score, encodes a protein housing three CUT domains and a homeodomain, both serving as DNA-binding moieties. This transcription factor critically governs neuronal proliferation and differentiation within the cerebral landscape, specifically modulating dendrite evolution, branching, dendritic spine genesis, and synaptogenesis within cortical strata II-III while demonstrating sequence-specific DNA binding. Markedly, CUX2 epitomizes a differentiation hallmark of glutamatergic pathways.(\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e) Suzuki et al.(\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e) established that a deficit in CUX2 significantly augments glutamatergic synaptic transmission within the hippocampal domain. Our findings resonate with this observation, underscoring that reduced CUX2 expression correlates with an elevated predisposition towards RHNAG diagnosis in rosacea-afflicted patients. This suggests that rosacea patients, due to the attenuated expression of CUX2, might experience amplified glutamate-mediated synaptic transmission, potentially triggering neurological manifestations that evolve into NR.\u003c/p\u003e \u003cp\u003eOur investigation represents a pioneering endeavor in the formulation of an auxiliary diagnostic model for NR predicated on biomarker profiling. Within a clinical milieu, it becomes feasible to ascertain the expression levels of these quartet molecules in peripheral blood. This allows for computation of the susceptibility to NR using the provided model equation. Subsequently, the derived prognostication can inform the prescription of appropriate therapeutic agents, an advancement that holds profound implications for the clinical diagnostic and therapeutic paradigms of NR. To elucidate, RHNAG-afflicted individuals might benefit from the administration of tricyclic antidepressants such as nortriptyline and protriptyline (traditionally prescribed for depressive disorders), sigma receptor antagonists like rimcazole (prevalently utilized in schizophrenia management), and dopamine receptor antagonists, including trifluoperazine and fluphenazine (typically prescribed for schizophrenia). Conversely, for those diagnosed with RLNAG, therapeutic options might encompass JAK inhibitors like Ruxolitinib (predominantly used for primary myelodysplasia and post-thrombocythemic myelodysplasia, with recent trials for autoimmune pathologies such as psoriasis) and MTOR inhibitors, e.g., everolimus (commonly incorporated in immunosuppressive regimens post-transplantation).Abnormal neurovascular regulation and imbalance of the inflammatory immune system are two intertwined pathogenetic factors of rosacea. Depending on the different clinical phenotypes of rosacea, the emphasis on these two mechanisms may differ. An intriguing query emerges: Why does NR predominantly exhibit neurological symptoms while manifesting attenuated inflammation? Studies by Kronfol et al. and Rothermundt et al. revealed that individuals without melancholic tendencies or depression often display proinflammatory states. Conversely, those with melancholic characteristics or diagnosed depression typically demonstrate diminished proinflammatory cytokine production.(\u003cspan additionalcitationids=\"CR61\" citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e) Significantly, anti-inflammatory cytokines such as transforming growth factor (TGF)-β and IL-10 often present at elevated levels in major depression (MD).(\u003cspan additionalcitationids=\"CR64\" citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e) Melancholic and non-melancholic patients show different immune patterns. It's plausible that neurosystemic aberrations in RHNAG patients could be inhibiting systemic inflammation. On the other hand, RLNAG rosacea patients exhibit inflammatory levels that align with their clinical manifestations.\u003c/p\u003e \u003cp\u003eNonetheless, our study is not devoid of limitations. A primary limitation is the paucity of valid samples, amounting to only 84, used in constructing the model. While we integrated all publicly available rosacea transcriptome data with our sequencing cohort, current constraints preclude further expansion of the sequencing sample size. Should additional rosacea data emerge in public repositories, we are poised to reassess our model. Another shortcoming stems from our model's reliance solely on mRNA expression levels within lesions. In a clinical setting, biomarkers sourced from blood or urine might be more feasible and palatable to patients. It's also noteworthy that our study observed five RHNAG patients manifesting pronounced papules and pustules, which are uncharacteristic for NR. This highlights the inherent challenge in perfectly aligning the NR phenotype through molecular modeling. Nevertheless, our RHNAG diagnostic model retains commendable accuracy and sensitivity in diagnosing NR. Future endeavors encompassing more extensive sequencing of both NR and NNR patients might pave the way for a refined and more accurate NR diagnostic model.\u003c/p\u003e \u003cp\u003eIn general, our pioneering approach combined machine learning with linear regression to devise a molecular diagnostic model for NR. This model, characterized by its high sensitivity and specificity, heralds a promising advancement in the diagnosis and treatment of NR. This could be particularly transformative in regions or among clinicians who might have previously overlooked NR.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eGEO, Gene Expression Omnibus; PCA, Principal Component Analysis; PLIN2, Perilipin 2; PNPLA3, Patatin Like Phospholipase Domain Containing 3; CUX2, Cut Like Homeobox 2; HMGCR, 3-Hydroxy-3-Methylglutaryl-CoA Reductase.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll the authors in this study agreed to the publication of the manuscript.\u0026nbsp;Informed consent for the publication of identifying images has been obtained from all subjects. It is important to note that no minors were included in the study cohort.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe experiments conducted in this study were approved by the Clinical Medical Ethics Committee of Xiangya Hospital, Central South University (approval number: 201404361). Written informed consent was obtained from all participants prior to their participation in the study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSequencing data from rosacea patients have been deposited in the genome sequence archive under accession number HRA000379 (http://bigd.big.ac.cn/gsa-human/). Clinical mugshots of 46 rosacea patients can be obtained by contacting the corresponding author.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRui Mao\u003c/strong\u003e: Conceptualization, Methodology, Software, Investigation, Visualization, Writing an original draft, Investigation. \u003cstrong\u003eJi Li\u003c/strong\u003e: Conceptualization, Writing - original draft, Funding acquisition, Supervision.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the National Natural Science Funds for Distinguished Young Scholars (No. 82225039)，the National Natural Science Foundation of China (No.82273557 and 8217344)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to thank the National Centre for Biotechnology Information staff. We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eRainer BM, Fischer AH, Luz Felipe da Silva D, Kang S, Chien AL. Rosacea is associated with chronic systemic diseases in a skin severity-dependent manner: results of a case-control study. J Am Acad Dermatol. 2015;73:604\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThiboutot D, Anderson R, Cook-Bolden F, Draelos Z, Gallo RL, Granstein RD, et al. Standard management options for rosacea: The 2019 update by the National Rosacea Society Expert Committee. J Am Acad Dermatol. 2020;82:1501\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEgeberg A, Fowler JF Jr, Gislason GH, Thyssen JP. Nationwide Assessment of Cause-Specific Mortality in Patients with Rosacea: A Cohort Study in Denmark. Am J Clin Dermatol. 2016;17:673\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVera N, Patel NU, Seminario-Vidal L. Rosacea Comorbidities. Dermatol Clin. 2018;36:115\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilkin J, Dahl M, Detmar M, Drake L, Feinstein A, Odom R, et al. Standard classification of rosacea: Report of the National Rosacea Society Expert Committee on the Classification and Staging of Rosacea. J Am Acad Dermatol. 2002;46:584\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScharschmidt TC, Yost JM, Truong SV, Steinhoff M, Wang KC, Berger TG. Neurogenic rosacea: a distinct clinical subtype requiring a modified approach to treatment. Arch Dermatol. 2011;147:123\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParkins GJ, Maan A, Dawn G. Neurogenic rosacea: an uncommon and poorly recognized entity. Clin Exp Dermatol. 2015;40:930\u0026ndash;1.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim HO, Kang SY, Kim KE, Cho SY, Kim KH, Kim IH. Neurogenic rosacea in Korea. J Dermatol. 2021;48:49\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiyachi Y. Potential antioxidant mechanism of action for metronidazole: implications for rosacea management. Adv Ther. 2001;18:237\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Zuuren EJ, Fedorowicz Z. Low-Dose Isotretinoin: An Option for Difficult-to-Treat Papulopustular Rosacea. J Invest Dermatol. 2016;136:1081\u0026ndash;3.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDel Rosso JQ, Webster GF, Jackson M, Rendon M, Rich P, Torok H, et al. Two randomized phase III clinical trials evaluating anti-inflammatory dose doxycycline (40-mg doxycycline, USP capsules) administered once daily for treatment of rosacea. J Am Acad Dermatol. 2007;56:791\u0026ndash;802.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYamasaki K, Di Nardo A, Bardan A, Murakami M, Ohtake T, Coda A, et al. Increased serine protease activity and cathelicidin promotes skin inflammation in rosacea. Nat Med. 2007;13:975\u0026ndash;80.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLacey N, Delaney S, Kavanagh K, Powell FC. Mite-related bacterial antigens stimulate inflammatory cells in rosacea. Br J Dermatol. 2007;157:474\u0026ndash;81.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYamasaki K, Kanada K, Macleod DT, Borkowski AW, Morizane S, Nakatsuji T, et al. TLR2 expression is increased in rosacea and stimulates enhanced serine protease production by keratinocytes. J Invest Dermatol. 2011;131:688\u0026ndash;97.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDirschka T, Tronnier H, F\u0026ouml;lster-Holst R. Epithelial barrier function and atopic diathesis in rosacea and perioral dermatitis. Br J Dermatol. 2004;150:1136\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChoi JE, Di Nardo A. Skin neurogenic inflammation. Semin Immunopathol. 2018;40:249\u0026ndash;59.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBuddenkotte J, Steinhoff M. Recent advances in understanding and managing rosacea. F1000Res 2018; 7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDel Rosso JQ. Management of facial erythema of rosacea: what is the role of topical α-adrenergic receptor agonist therapy. J Am Acad Dermatol. 2013;69:44\u0026ndash;56.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMetzler-Wilson K, Toma K, Sammons DL, Mann S, Jurovcik AJ, Demidova O, et al. Augmented supraorbital skin sympathetic nerve activity responses to symptom trigger events in rosacea patients. J Neurophysiol. 2015;114:1530\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteinhoff M, Schmelz M, Schauber J. Facial Erythema of Rosacea - Aetiology, Different Pathophysiologies and Treatment Options. Acta Derm Venereol. 2016;96:579\u0026ndash;86.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGillessen S, Armstrong A, Attard G, Beer TM, Beltran H, Bjartell A et al. Management of Patients with Advanced Prostate Cancer: Report from the Advanced Prostate Cancer Consensus Conference 2021. Eur Urol 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTayob N, Kanwal F, Alsarraj A, Hernaez R, El-Serag HB. The Performance of AFP, AFP-3, DCP as Biomarkers for Detection of Hepatocellular Carcinoma (HCC): A Phase 3 Biomarker Study in the United States. Clin Gastroenterol Hepatol; 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXie Q, Xue W. IgE-Mediated food allergy: Current diagnostic modalities and novel biomarkers with robust potential. Crit Rev Food Sci Nutr 2022: 1\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGallo RL, Granstein RD, Kang S, Mannis M, Steinhoff M, Tan J, et al. Standard classification and pathophysiology of rosacea: The 2017 update by the National Rosacea Society Expert Committee. J Am Acad Dermatol. 2018;78:148\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchram AM, James WD. Neurogenic rosacea treated with endoscopic thoracic sympathectomy. Arch Dermatol. 2012;148:270\u0026ndash;1.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNorquist JM, Watson DJ, Yu Q, Paolini JF, McQuarrie K, Santanello NC. Validation of a questionnaire to assess niacin-induced cutaneous flushing. Curr Med Res Opin. 2007;23:1549\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics. 2010;26:1572\u0026ndash;3.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcLachlan GJ, Bean RW, Ng SK. Clustering Methods Mol Biol. 2017;1526:345\u0026ndash;62.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRitchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eH\u0026auml;nzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBecht E, Giraldo NA, Lacroix L, Buttard B, Elarouci N, Petitprez F, et al. Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol. 2016;17:218.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAran D, Hu Z, Butte AJ. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol. 2017;18:220.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16:284\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLangfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10:1523.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBader GD, Hogue CW. An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinformatics. 2003;4:2.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSauerbrei W, Royston P, Binder H. Selection of important variables and determination of functional form for continuous predictors in multivariable model building. Stat Med. 2007;26:5512\u0026ndash;28.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFriedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw. 2010;33:1\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKidd AC, McGettrick M, Tsim S, Halligan DL, Bylesjo M, Blyth KG. Survival prediction in mesothelioma using a scalable Lasso regression model: instructions for use and initial performance using clinical predictors. BMJ Open Respir Res. 2018;5:e000240.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang YQ, Liang CH, He L, Tian J, Liang CS, Chen X, et al. Development and Validation of a Radiomics Nomogram for Preoperative Prediction of Lymph Node Metastasis in Colorectal Cancer. J Clin Oncol. 2016;34:2157\u0026ndash;64.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKramer AA, Zimmerman JE. Assessing the calibration of mortality benchmarks in critical care: The Hosmer-Lemeshow test revisited. Crit Care Med. 2007;35:2052\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVickers AJ, Cronin AM, Elkin EB, Gonen M. Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers. BMC Med Inform Decis Mak. 2008;8:53.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLex A, Gehlenborg N, Strobelt H, Vuillemot R, Pfister H. UpSet: Visualization of Intersecting Sets. IEEE Trans Vis Comput Graph. 2014;20:1983\u0026ndash;92.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGonz\u0026aacute;lez-Santana A, Marrero-Hern\u0026aacute;ndez S, Dorta I, Hern\u0026aacute;ndez M, Pinto FM, B\u0026aacute;ez D, et al. Altered expression of the tachykinins substance P/neurokinin A/hemokinin-1 and their preferred neurokinin 1/neurokinin 2 receptors in uterine leiomyomata. Fertil Steril. 2016;106:1521\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoparthi L, Survery S, Kreir M, Simonsen C, Kjellbom P, H\u0026ouml;gest\u0026auml;tt ED, et al. Human TRPA1 is intrinsically cold- and chemosensitive with and without its N-terminal ankyrin repeat domain. Proc Natl Acad Sci U S A. 2014;111:16901\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaulsen CE, Armache JP, Gao Y, Cheng Y, Julius D. Structure of the TRPA1 ion channel suggests regulatory mechanisms. Nature. 2015;520:511\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCuratolo W. Glycolipid function. Biochim Biophys Acta. 1987;906:137\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaulazo MA, Sodero AO. SIRT-1 Activity Sustains Cholesterol Synthesis in the Brain. Neuroscience. 2021;476:116\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCriado M, Eibl H, Barrantes FJ. Effects of lipids on acetylcholine receptor. Essential need of cholesterol for maintenance of agonist-induced state transitions in lipid vesicles. Biochemistry. 1982;21:3622\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurger K, Gimpl G, Fahrenholz F. Regulation of receptor function by cholesterol. Cell Mol Life Sci. 2000;57:1577\u0026ndash;92.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaier CJ, Barrantes FJ. Sphingolipids are necessary for nicotinic acetylcholine receptor export in the early secretory pathway. J Neurochem. 2007;101:1072\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFantini J, Garmy N, Mahfoud R, Yahi N. Lipid rafts: structure, function and role in HIV, Alzheimer's and prion diseases. Expert Rev Mol Med. 2002;4:1\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSonnino S, Chigorno V. Ganglioside molecular species containing C18- and C20-sphingosine in mammalian nervous tissues and neuronal cell cultures. Biochim Biophys Acta. 2000;1469:63\u0026ndash;77.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuskey KL, Stevens B. Human 3-hydroxy-3-methylglutaryl coenzyme A reductase. Conserved domains responsible for catalytic activity and sterol-regulated degradation. J Biol Chem. 1985;260:10271\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCuccioloni M, Mozzicafreddo M, Spina M, Tran CN, Falconi M, Eleuteri AM, et al. Epigallocatechin-3-gallate potently inhibits the in vitro activity of hydroxy-3-methyl-glutaryl-CoA reductase. J Lipid Res. 2011;52:897\u0026ndash;907.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaushik S, Cuervo AM. Degradation of lipid droplet-associated proteins by chaperone-mediated autophagy facilitates lipolysis. Nat Cell Biol. 2015;17:759\u0026ndash;70.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRomeo S, Kozlitina J, Xing C, Pertsemlidis A, Cox D, Pennacchio LA, et al. Genetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease. Nat Genet. 2008;40:1461\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeissier A, Griveau A, Vigier L, Piolot T, Borello U, Pierani A. A novel transient glutamatergic population migrating from the pallial-subpallial boundary contributes to neocortical development. J Neurosci. 2010;30:10563\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSuzuki T, Tatsukawa T, Sudo G, Delandre C, Pai YJ, Miyamoto H, et al. CUX2 deficiency causes facilitation of excitatory synaptic transmission onto hippocampus and increased seizure susceptibility to kainate. Sci Rep. 2022;12:6505.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRothermundt M, Arolt V, Fenker J, Gutbrodt H, Peters M, Kirchner H. Different immune patterns in melancholic and non-melancholic major depression. Eur Arch Psychiatry Clin Neurosci. 2001;251:90\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRothermundt M, Arolt V, Peters M, Gutbrodt H, Fenker J, Kersting A, et al. Inflammatory markers in major depression and melancholia. J Affect Disord. 2001;63:93\u0026ndash;102.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKronfol Z. Immune dysregulation in major depression: a critical review of existing evidence. Int J Neuropsychopharmacol. 2002;5:333\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHowren MB, Lamkin DM, Suls J. Associations of depression with C-reactive protein, IL-1, and IL-6: a meta-analysis. Psychosom Med. 2009;71:171\u0026ndash;86.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim EK, et al. A meta-analysis of cytokines in major depression. Biol Psychiatry. 2010;67:446\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK\u0026ouml;hler CA, Freitas TH, Maes M, de Andrade NQ, Liu CS, Fernandes BS, et al. Peripheral cytokine and chemokine alterations in depression: a meta-analysis of 82 studies. Acta Psychiatr Scand. 2017;135:373\u0026ndash;87.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-genomics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mgnm","sideBox":"Learn more about [BMC Medical Genomics](http://bmcmedgenomics.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/mgnm/default.aspx","title":"BMC Medical Genomics","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"neurogenic rosacea, diagnostic system, transcriptome","lastPublishedDoi":"10.21203/rs.3.rs-3791877/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3791877/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePatients with neurogenic rosacea (NR) frequently demonstrate pronounced neurological manifestations, often unresponsive to conventional therapeutic approaches. A molecular-level understanding and diagnosis of this patient cohort could significantly guide clinical interventions. In this study, we amalgamated our sequencing data (n\u0026thinsp;=\u0026thinsp;46) with a publicly accessible database (n\u0026thinsp;=\u0026thinsp;38) to perform an unsupervised cluster analysis of the integrated dataset. The eighty-four rosacea patients were partitioned into two distinct clusters. Neurovascular biomarkers were found to be elevated in cluster 1 compared to cluster 2. Pathways in cluster 1 were predominantly involved in neurotransmitter synthesis, transmission, and functionality, whereas cluster 2 pathways were centered on inflammation-related processes. Differential gene expression analysis and WGCNA were employed to delineate the characteristic gene sets of the two clusters. Subsequently, a diagnostic model was constructed from the identified gene sets using linear regression methodologies. The model's C index, comprising genes PNPLA3, CUX2, PLIN2, and HMGCR, achieved a remarkable value of 0.9683, with an area under the curve (AUC) for the training cohort's nomogram of 0.9376. Clinical characteristics from our dataset (n\u0026thinsp;=\u0026thinsp;46) were assessed by three seasoned dermatologists, forming the NR validation cohort (NR, n\u0026thinsp;=\u0026thinsp;18; non-neurogenic rosacea, n\u0026thinsp;=\u0026thinsp;28). Upon application of our model to NR diagnosis, the model's AUC value reached 0.9023. Finally, potential therapeutic candidates for both patient groups were predicted via the Connectivity Map. In summation, this study unveiled two clusters with unique molecular phenotypes within rosacea, leading to the development of a precise diagnostic model instrumental in NR diagnosis.\u003c/p\u003e","manuscriptTitle":"Construction of a molecular diagnostic system for neurogenic rosacea by combining transcriptome sequencing and machine learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-01-04 00:08:24","doi":"10.21203/rs.3.rs-3791877/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-08-29T09:36:39+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-08-15T00:59:25+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"21864480146484951008658172484849122372","date":"2024-06-28T20:27:21+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"255936688076443281965114076005493976656","date":"2024-06-28T14:53:36+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-06-27T03:26:56+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"118744573402577382280459604075896616251","date":"2024-06-24T08:34:44+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"6f9cd7dc-3742-4f23-b363-e8450df8a280","date":"2024-01-29T14:25:22+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2023-12-29T08:40:11+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2023-12-29T08:29:40+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2023-12-26T07:36:41+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2023-12-26T07:34:02+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Genomics","date":"2023-12-22T12:12:50+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-genomics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mgnm","sideBox":"Learn more about [BMC Medical Genomics](http://bmcmedgenomics.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/mgnm/default.aspx","title":"BMC Medical Genomics","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"75f8112a-7b86-4d68-aaaf-7694530628c9","owner":[],"postedDate":"January 4th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-09-16T16:10:22+00:00","versionOfRecord":{"articleIdentity":"rs-3791877","link":"https://doi.org/10.1186/s12920-024-02008-0","journal":{"identity":"bmc-medical-genomics","isVorOnly":false,"title":"BMC Medical Genomics"},"publishedOn":"2024-09-13 15:58:31","publishedOnDateReadable":"September 13th, 2024"},"versionCreatedAt":"2024-01-04 00:08:24","video":"","vorDoi":"10.1186/s12920-024-02008-0","vorDoiUrl":"https://doi.org/10.1186/s12920-024-02008-0","workflowStages":[]},"version":"v1","identity":"rs-3791877","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3791877","identity":"rs-3791877","version":["v1"]},"buildId":"ehx78VzkSd0WSzXnipQa-","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}