Inferring circadian rhythm disruptions in cancer using phase differences between clock genes

Circadian rhythms exist across various levels of biological activities, from molecular processes to behaviors. Circadian clock genes are closely linked to the onset and progression of cancer; however, systematic analysis of their rhythmic expression in human tumors is still lacking due to difficulties in time-series sampling. In this study, we examined and improved the method for inferring phase differences through clock gene co-expression to investigate circadian rhythms in timestamp-free samples. We found that the co-expression levels of rhythmic genes are primarily determined by phase differences and are influenced by the strength and tissue specificity of rhythmic expression. Thus, we identified evolutionarily conserved rhythmic genes across multiple tissues to construct tissue-specific phase difference matrices. On this basis, we developed a method for inferring phase difference variations and extended it to the single-sample level as ssDistance to evaluate circadian rhythm disruptions in tumors. Results revealed that the significant alterations in phase differences between clock genes in tumors are cancer-type-specific and have complex effects on patient survival. This method provides an effective tool for studying circadian rhythms in large-scale public datasets lacking temporal information.