Dataset Information

Machine Learning-Based Genome-Wide Salivary DNA Methylation Analysis for Identification of Noninvasive Biomarkers in Oral Cancer Diagnosis.

ABSTRACT: This study aims to examine the feasibility of ML-assisted salivary-liquid-biopsy platforms using genome-wide methylation analysis at the base-pair and regional resolution for delineating oral squamous cell carcinoma (OSCC) and oral potentially malignant disorders (OPMDs). A nested cohort of patients with OSCC and OPMDs was randomly selected from among patients with oral mucosal diseases. Saliva samples were collected, and DNA extracted from cell pellets was processed for reduced-representation bisulfite sequencing. Reads with a minimum of 10× coverage were used to identify differentially methylated CpG sites (DMCs) and 100 bp regions (DMRs). The performance of eight ML models and three feature-selection methods (ANOVA, MRMR, and LASSO) were then compared to determine the optimal biomarker models based on DMCs and DMRs. A total of 1745 DMCs and 105 DMRs were identified for detecting OSCC. The proportion of hypomethylated and hypermethylated DMCs was similar (51% vs. 49%), while most DMRs were hypermethylated (62.9%). Furthermore, more DMRs than DMCs were annotated to promoter regions (36% vs. 16%) and more DMCs than DMRs were annotated to intergenic regions (50% vs. 36%). Of all the ML models compared, the linear SVM model based on 11 optimal DMRs selected by LASSO had a perfect AUC, recall, specificity, and calibration (1.00) for OSCC detection. Overall, genome-wide DNA methylation techniques can be applied directly to saliva samples for biomarker discovery and ML-based platforms may be useful in stratifying OSCC during disease screening and monitoring.

SUBMITTER: Adeoye J

PROVIDER: S-EPMC9563273 | biostudies-literature | 2022 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine Learning-Based Genome-Wide Salivary DNA Methylation Analysis for Identification of Noninvasive Biomarkers in Oral Cancer Diagnosis.

Adeoye John J Wan Chi Ching Joan CCJ Zheng Li-Wu LW Thomson Peter P Choi Siu-Wai SW Su Yu-Xiong YX

Cancers 20221008 19

This study aims to examine the feasibility of ML-assisted salivary-liquid-biopsy platforms using genome-wide methylation analysis at the base-pair and regional resolution for delineating oral squamous cell carcinoma (OSCC) and oral potentially malignant disorders (OPMDs). A nested cohort of patients with OSCC and OPMDs was randomly selected from among patients with oral mucosal diseases. Saliva samples were collected, and DNA extracted from cell pellets was processed for reduced-representation b ...[more]

PMID: 36230858

Similar Datasets

Project description:Osteoarthritis (OA) is thought to be the most prevalent chronic joint disease. The incidence of OA is rising because of the ageing population and the epidemic of obesity. This research was designed for the identification of novel diagnostic biomarkers for OA and analyzing the possible association between critical genes and infiltrated immune cells. 10 OA samples from patients with spinal OA and 10 normal samples were collected. GSE55235 and GSE55457 datasets including human OA and normal samples were downloaded from the GEO datasets. Differentially expressed genes (DEGs) were identified between 20 OA and 20 controls. SVM-RFE analysis and LASSO regression model were carried out to screen possible markers. The compositional patterns of the 22 types of immune cell fraction in OA were determined by the use of CIBERSORT. The expression level of the biomarkers in OA was examined by the use of RT-PCR. In this study, an overall 44 DEGs were identified: 18 genes were remarkably upregulated and 26 genes were distinctly downregulated. KEGG pathway analyses revealed that pathways were significantly enriched including IL-17 signal path, rheumatoid arthritis, TNF signal path, and lipid and atherosclerosis. Based on the results of machine learning, we identified APOLD1 and EPYC as critical diagnostic genes in OA, which were further confirmed using ROC assays. Immune cell infiltration analysis revealed that APOLD1 was correlated with mastocytes stimulated, NK cells resting, T cells CD4 memory resting, DCs stimulated, T cells gamma delta, macrophages M0, NK cells stimulated, and mastocytes resting. Moreover, we found that EPYC was correlated with mastocytes stimulated, NK cells resting, T cells CD4 memory resting, DCs stimulated, T cells gamma delta, macrophages M0, NK cells stimulated, and mastocytes resting. Overall, our findings might provide some novel clue for the exploration of novel markers for OA diagnosis. The critical genes and their associations with immune infiltration may offer new insight into understanding OA developments.

Project description:Sepsis is a systemic inflammatory response syndrome caused by bacteria and other pathogenic microorganisms. Every year, approximately 31.5 million patients are diagnosed with sepsis, and approximately 5.3 million patients succumb to the disease. In this study, we identified biomarkers for diagnosing sepsis analyzed the relationships between genes and Immune cells that were differentially expressed in specimens from patients with sepsis compared to normal controls. Finally, We verified its effectiveness through animal experiments. Specifically, we analyzed datasets from four microarrays(GSE11755、GSE12624、GSE28750、GSE48080) that included 106 blood specimens from patients with sepsis and 69 normal human blood samples. SVM-RFE analysis and LASSO regression model were carried out to screen possible markers. The composition of 22 immune cell components in patients with sepsis were also determined using CIBERSORT. The expression level of the biomarkers in Sepsis was examined by the use of qRT-PCR and Western Blot (WB). We identified 50 differentially expressed genes between the cohorts, including 2 significantly upregulated and 48 significantly downregulated genes, and KEGG pathway analysis identified Salmonella infection, human T cell leukemia virus 1 infection, Epstein-Barr virus infection, hepatitis B, lysosome and other pathways that were significantly enriched in blood from patients with sepsis. Ultimately, we identified COMMD9, CSF3R, and NUB1 as genes that could potentially be used as biomarkers to predict sepsis, which we confirmed by ROC analysis. Further, we identified a correlation between the expression of these three genes and immune infiltrate composition. Immune cell infiltration analysis revealed that COMMD9 was correlated with T cells regulatory (Tregs), T cells follicular helper, T cells CD8, et al. CSF3R was correlated with T cells regulatory (Tregs), T cells follicular helper, T cells CD8, et al. NUB1 was correlated with T cells regulatory (Tregs), T cells gamma delta, T cells follicular helper, et al. Taken together, our findings identify potential new diagnostic markers for sepsis that shed light on novel mechanisms of disease pathogenesis and, therefore, may offer opportunities for therapeutic intervention.

Project description:BackgroundIntracerebral hemorrhage (ICH) is a severe stroke subtype with high morbidity, disability, and mortality rates. Currently, no biomarkers for ICH are available for use in clinical practice. We aimed to explore the roles of RNAs in ICH pathogenesis and identify potential diagnostic biomarkers.MethodsWe collected 233 individual blood samples from two independent cohorts, including 64 patients with ICH, 59 patients with ischemic stroke (IS), 60 patients with hypertension (HTN) and 50 healthy controls (CTRL) for RNA sequencing. Differentially expressed genes (DEGs) analysis, gene set enrichment analysis (GSEA), and weighted correlation network analysis (WGCNA) were performed to identify ICH-specific modules. The immune cell composition was evaluated with ImmuneCellAI. Multiple machine learning algorithms to select potential biomarkers for ICH diagnosis, and further validated by quantitative real-time polymerase chain reaction (RT-PCR). Receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA) were performed to evaluate the diagnostic value of the signature for ICH. Finally, we generated M1 and M2 macrophages to investigate the expression of candidate genes.ResultsIn both cohorts, 519 mRNAs and 131 lncRNAs were consistently significantly differentially expressed between ICH patients and HTN controls. Gene function analysis suggested that immune system processes may be involved in ICH pathology. ImmuneCellAI analysis revealed that the abundances of 11 immune cell types were altered after ICH in both cohorts. WGCNA and GSEA identified 18 immune-related DEGs. Multiple algorithms identified an RNA panel (CKAP4, BCL6, TLR8) with high diagnostic value for discriminating ICH patients from HTN controls, CTRLs and IS patients (AUCs: 0.93, 0.95 and 0.82; sensitivities: 81.3%, 84.4% and 75%; specificities: 100%, 96% and 79.7%, respectively). Additionally, CKAP4 and TLR8 mRNA and protein levels decreased in RAW264.7 M1 macrophages and increased in RAW264.7 M2 macrophages, while BCL6 expression increased in M1 macrophages but not in M2 macrophages, which may provide potential therapeutic targets for ICH.ConclusionsThis study demonstrated that the expression levels of lncRNAs and mRNAs are associated with ICH, and an RNA panel (CKAP4, BCL6, TLR8) was developed as a potential diagnostic tool for distinguishing ICH from IS and controls, which could provide useful insight into ICH diagnosis and pathogenesis.

Dataset Information

Machine Learning-Based Genome-Wide Salivary DNA Methylation Analysis for Identification of Noninvasive Biomarkers in Oral Cancer Diagnosis.

Publications

Machine Learning-Based Genome-Wide Salivary DNA Methylation Analysis for Identification of Noninvasive Biomarkers in Oral Cancer Diagnosis.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets