Extra Tree Classifier Predicts an Interactome Hub Gene as HSPB1 in Oral Cancer: A Bioinformatics Analysis

Introduction Oral cancer is a significant global health issue that is mainly caused by factors, such as smoking, alcohol consumption, poor oral hygiene, age, and the human papillomavirus. Unfortunately, delayed diagnosis contributes to high rates of illness and mortality. However, saliva shows promise as a potential source for early detection, prognosis, and treatment. By analyzing the proteins and their interactions in saliva, we can gain insights that can assist in early detection and prediction. In this study, we aim to identify and predict the key genes, known as hub genes, in the salivary transcriptomics data of oral cancer patients and healthy individuals. Methods The data used for the analysis were obtained from salivaryproteome.org (https://salivaryproteome.org/) . The retrieved data consisted of individuals with oral cancer who were assigned unique identification numbers (IDs) 1025, 1030, 1027, and 1029, while the healthy individuals were assigned IDs 4256, 4257, 4255, and 4258, respectively. Differential gene expression analysis was used to identify genes that showed significant differences between the two groups. Uniformity and clustering were assessed through heatmaps and principal component analysis. Protein-protein interactions were investigated using the STRING database and Cytoscape. In addition, machine learning algorithms were employed to identify key genes involved in the interatomic interactions by analyzing transcriptomics data generated from the differential gene expression analysis. Results The accuracy and class accuracy of the extra tree classifier showed 98% and 97% in predicting interactomic hub genes, and HSPB1 was identified as a hub gene using Cytohubba from Cytoscape. Conclusion The predictive extra tree classifier, with its high accuracy in analysing interactomic hub genes in oral cancer, can potentially improve diagnosis and treatment strategies.