Abstract
Motivation: Differential rhythmicity analysis identifies condition-specific changes in biological oscillations, but diverse algorithmic assumptions complicate method selection. Furthermore, integrating complementary evidence across different tools while accounting for statistical dependence remains an unresolved challenge.
Results: We benchmarked seven algorithms across diverse rhythmic scenarios, revealing that optimal selection depends heavily on specific biological variations and experimental designs. To overcome single-algorithm limitations, we introduced Kost's Method, a dependence-aware p-value integration approach. Building on this framework, we developed DiffRhyShiny, an interactive web platform that automates robust differential rhythmicity analysis, consensus evaluation, and visualization. Furthermore, we validated the platform's performance and statistical rigor using both extensive simulated scenarios and real-world transcriptomic datasets.
Availability and implementation: DiffRhyShiny is freely available at https://omicsai.shinyapps.io/diffrhyshiny, with the latest source code at https://github.com/liuchuan111/diffrhyshiny. The code and data to reproduce all benchmarking results are archived on Figshare at https://doi.org/10.6084/m9.figshare.33070100.