A comparative study of algorithms detecting differential rhythmicity in transcriptomic data

Rhythmic transcripts play pivotal roles in driving the daily oscillations of various biological processes. Genetic or environmental disruptions can lead to alterations in the rhythmicity of transcripts, ultimately impacting downstream circadian outputs, including metabolic processes and even behavior. To statistically compare the differences in transcript rhythms between two or more conditions, several algorithms have been developed to analyze circadian transcriptomic data, each with distinct features. In this study, we compared the performance of seven algorithms that were specifically designed to detect differential rhythmicity. We found that even when applying the same statistical threshold, these algorithms yielded varying numbers of differentially rhythmic transcripts. Nevertheless, the set of transcripts commonly identified as differentially rhythmic exhibited substantial overlap among algorithms. Furthermore, the phase and amplitude differences calculated by these algorithms displayed significant correlations. In summary, our study highlights a high degree of similarity in the results produced by these algorithms. Furthermore, when selecting an algorithm for analysis, it is crucial to ensure the compatibility of input data with the specific requirements of the chosen algorithm and to assess whether the algorithm's output fits the needs of the user.