Identification of Biomarker in Lung Adenocarcinoma Using Machine Learning and Neural Network

Abstract

Lung cancer is one of the leading causes globally. The survival rate is relatively low because symptoms of lung
adenocarcinoma frequently do not appear until the illness has spread to every part of the lung. Early disease detection can prevent
the spread of cancer and reduce cancer-related mortality. Biomarkers help in early disease detection. Machine learning and neural
network approaches, which use mathematical techniques to train a model to learn from data for a particular task, have been widely
used in biomarker discovery because identifying biomarkers is a time-consuming procedure. Based on the expression of
biomarkers in the various groups, the "Pathway analysis" service evaluates the enrichment of biological processes, gene sets, and
signaling pathways. The study aims to find the expressed gene sets, enriched signaling pathways, and biomarkers for lung
adenocarcinoma. The "TCGA-LUAD" project's TCGA data is used to identify biomarkers, and PCA analysis reveals that most lung
adenocarcinoma patients have no history of other malignancies. Our examination of the GO biological process overrepresentation
reveals that 499/6818 represents the BgRatio of cytokine response, and 129 GOs have P values less than 0.0005,
indicating that they are strongly affected by biological processes. The GO Molecular Function Over-Representation Analysis reveals
that 447 biomarkers with differential expression and 20 GOs with P values less than 0.001 are substantially affected by molecular
function, with a BgRatio of transporter activity of 472/6790. Additionally, GO Cellular Component Over-Representation Analysis
reveals that 339 biomarkers with differential expression and 21 GOs with P values less than 1e-07, where the BgRatio of Cell
Surface is 495/7043, are substantially affected cellular components.