Detection of rhythmic signals in Oscillatory Systems using order Restricted Inference

Abstract

The determination of rhythmic signals in oscillatory systems, such as cell cycle or circadian clock, is essential for biologists to know which genes are associated to the system. While there are several procedures available for this task in the literature, none of them is satisfactory enough. One of the reasons for this is the absence of a good definition of rhythmic signal. We propose a new definition of rhythmic signal using order restrictions and taking into account the needs of the biologists, and an algorithm based on order restricted inference and conditional tests to detect and classify the signals in different groups. We test the algorithm in simulations and with real databases from circadian clock, and compare it with the most usual methods available showing its good performance.