CFE 2020: Start Registration
View Submission - CMStatistics
Title: Circular order aggregation as a novel proposal to estimate sampling times in chronobiology Authors:  Yolanda Larriba - University of Valladolid (Spain) [presenting]
Cristina Rueda - University of Valladolid (Spain)
Abstract: Core clock circadian genes such as PERs, CRYs or ARNTL display clearly rhythmic expression patterns every 24 hours. The analysis of rhythmic genes, including pattern comparisons, amplitude or time point estimates, is crucial in chronobiology to understand biological functions correctly. However, daily expression data collection is expensive and may suppose a risk for health (e.g. human biopsies). Typically, in chronobiology, gene expression data are given from post-mortem specimens where expression data are measured at different times across several organs. This process involves a high uncertainty such as the exactly specimen time of death is usually unknown or not accurate; samples are taken from a large number of specimens; and because more than 30,000 expression data, including both rhythmic and non-rhythmic genes, are obtained from each organ simultaneously. Based on core clock genes, several directional approaches are evaluated, including circular principal component analysis, to solve temporal order estimation as a circular order aggregation problem. Specifically, statistical methods are applied to GTEx data collection, which contains more than 17,000 post-mortem RNA-Seq expression data across 54 human tissues. Preliminary results provide accurate sampling time estimates and enhance the knowledge of circadian biology.