همایش ملی بیوانفورماتیک ایران

صفحه اصلی / 4th international edition and 13th Iranian Conference on Bioinformatics

Innovative Biomarkers for Lung Cancer Classification and Prediction Using High-Dimensional Machine Learning: A Novel Approach to Targeted Therapies

نویسندگان :

Marzie Shadpirouz¹ Morteza Hadizadeh² Zahra Salehi³ Maziyar Veisi⁴ Reza Maddah⁵ Mohammad Shirinpour⁶ Kazem Abbaszadeh-Goudarzi⁷ Kaveh Kavousi⁸

1- University of Tehran 2- Kerman University of Medical Sciences 3- Tehran University of Medical Sciences 4- Shahrekord University 5- National Institute of Genetic Engineering and Biotechnology 6- Shahid Beheshti University 7- Sabzevar University of Medical Sciences 8- University of Tehran

کلمات کلیدی :

Lung adenocarcinoma،Machine learning،Precision Medicine،Drug repositioning

چکیده :

Lung adenocarcinoma (LUAD) represents the most prevalent pathological subtype of lung cancer. Unfortunately, a significant proportion of LUAD cases are detected at advanced stages, where the prognosis remains unfavorable. Therefore, our objective was to discover innovative biomarkers to enhance the diagnosis and treatment of early-stage LUAD and to develop targeted therapeutic strategies. In this study, two microarray datasets, GSE75037 and GSE32863, were sourced from the Gene Expression Omnibus (GEO) database for analysis. Data preprocessing and meta-analysis were conducted using the R statistical programming language. Feature selection was carried out through analysis of variance (ANOVA). To predict cancer stages, ordinal regression, ordinal tree, and ordinal forest models were developed, and their predictive performance was evaluated. Differentially expressed genes (DEGs) were identified and subjected to gene set enrichment analysis to uncover significant biological pathways. The validation and diagnostic accuracy of the DEGs were further examined using UALCAN and ROC curve analysis. Additionally, the selected genes were validated using an independent, comprehensive LUAD RNA-Seq dataset from The Cancer Genome Atlas (TCGA). To identify cell types associated with these genes, scRNA-Seq data from the GEO database (dataset GSE131907) including lung cancer and normal samples were analyzed. Differential gene expression was examined using the FindMarkers function from the Seurat package. Finally, potential therapeutic agents were identified through the DGIdb database, followed by molecular docking and molecular dynamics simulations to evaluate their potential efficacy as treatments. Analysis revealed that the top 40 differentially expressed genes (DEGs) were primarily associated with pathways involving drug metabolism via cytochrome P450, xenobiotic metabolism by cytochrome P450, and retinol metabolism. Among these, genes ADH1A, ADH1B, and F10 stood out due to their significant interactions within these pathways. Notably, a negative correlation was observed between the expression levels of these genes and patient survival rates. Additionally, their expression was significantly reduced in lung adenocarcinoma (LUAD) tissues compared to adjacent non-tumor tissues. The diagnostic potential of these genes was validated through ROC analysis, highlighting their ability to distinguish between cancerous and non-cancerous tissues. TCGA analysis showed that the selected genes were statistically significant across all stages, from I to IV .Based on the single-cell analysis, the ADH1B gene shows a significant differential expression in the Mesothelial cell type, an average log2 fold change of -6.1985, and an adjusted P-value of 0.0043. Moreover, drugs such as abacavir, rivaroxaban, and edoxaban tosylate demonstrated a strong binding affinity to these genes, suggesting their therapeutic relevance. These findings emphasize the value of leveraging bioinformatics and machine learning approaches to uncover new pathways for LUAD diagnosis and treatment, paving the way for more targeted and effective interventions.