Towards the Identification of Multiclass Lung Cancer-Related Genes: An Evolutionary and Intelligent Procedure

The amount of available transcriptomic data from which relevant knowledge can be extracted has rapidly increased. Besides, with the advances in areas such as machine learning and high-performance computing, the time and computing efforts for analyzing those data are being reduced, leading to the design of Clinical-Decision Support Systems (CDSS) for the precision medicine paradigm. As a result of this increase, the use of intelligent and evolutionary methods to study and classify cancer diseases has been proposed in the literature showing promising results. This study is aimed at identifying a set of genes able to distinguish between the following types of lung cancer: Adenocarcinoma (ACC), Squamous Cell Carcinoma (SCC), and healthy lung. The Differentially Expressed Genes (DEGs) analysis was carried out through RNA-seq data coming from The Cancer Genome Atlas (TCGA). An optimized evolutionary procedure has been developed with the purpose of finding the optimal combinations among the hundreds of candidate DEGs. Our custom method can maximize multiclass lung cancer recognition while minimizing the number of selected DEGs. The results show an outstanding classification rate with a significantly reduced number of DEGs biologically related to lung cancer.