B0802
Title: Taking advance of the circular manifold for the analysis of circadian gene expression data
Authors: Yolanda Larriba - University of Valladolid (Spain) [presenting]
Cristina Rueda - Universidad de Valladolid (Spain)
Abstract: Circadian rhythms are genetically encoded by a molecular clock that generates internal timing of approximately 24 hours. At a molecular level, circadian genes display daily rhythmic expression patterns which vary across tissues or species. The underlying circular structure of such molecular rhythms, which can be seen as oscillatory processes, makes the circular space a suitable manifold to formulate and solve key problems in circadian biology. Among others, within the circular space, the sampling temporal order estimation problem is efficiently addressed. The solution to this problem is based both on a circular approach of the well-known principal component analysis method and on the definition of circular order. In addition, it is proposed the use of a novel non-linear parametric regression model capable of adapting to non-sinusoidal shapes, like those observed in the molecular rhythms, and whose parameters are easily interpretable. The results provide accurate genes' peak phase estimates across different tissues and allow a better comprehension of the complex molecular clock network.