Identifying Clock Gene Signatures for Improved Cancer Chronotherapy Through Gene Expression Analysis

Cancer is a leading cause of death worldwide, and disruptions to the circadian system, which is regulated by a set of clock genes, can lead to tumorigenesis. Previous studies have used statistical analysis to identify differentially expressed clock genes (DECGs) between tumor and healthy cells. This research focuses on extracting single-gene circadian signatures from DECGs as potential biomarkers of specific cancer types using a novel component and system-level hybrid strategy. At the component level, each circadian gene was studied through four data analysis approaches to establish circadian signature candidates for breast, lung, kidney, and thyroid cancer. Genes with standout metrics from hypothesis testing and machine learning-based classification
$(\mathrm{p} < 0.01, 74-91\%$
accuracy) were examined further through machine learning regression and principal component analysis. At the system level, the circadian signature candidates were analyzed for involvement in cell cycle checkpoints, cell signaling, apoptosis regulation, and other biological processes associated with tumorigenesis. The findings demonstrate that individual data analysis approaches only detect specific patterns of circadian disruption. In contrast, the proposed hybrid approach defines for the first time an important set of single-gene circadian signatures for the four cancer types: CRY2 and PER1 were identified as circadian signatures for breast cancer, CRY2 and CSNK1D for lung cancer, CSNK1E and BHLHE41 for kidney cancer, and NRID1, ARNTL, and CSNK1E for thyroid cancer. The isolation of contributing factors to tumorigenesis in single-gene circadian signatures provides critical insights to advancing new time-based strategies for treating cancer.