Improved Statistical Methods Enable Greater Sensitivity in Rhythm Detection for Genome-Wide Data

Abstract

Robust methods for identifying patterns of expression in genome-wide data are important for generating hypotheses regarding gene function. To this end, several analytic methods have been developed for detecting periodic patterns. We improve one such method, JTK_CYCLE, by explicitly calculating the null distribution such that it accounts for multiple hypothesis testing and by including non-sinusoidal reference waveforms. We term this method empirical JTK_CYCLE with asymmetry search, and we compare its performance to JTK_CYCLE with Bonferroni and Benjamini-Hochberg multiple hypothesis testing correction, as well as to five other methods: cyclohedron test, address reduction, stable persistence, ANOVA, and F24. We find that ANOVA, F24, and JTK_CYCLE consistently outperform the other three methods when data are limited and noisy; empirical JTK_CYCLE with asymmetry search gives the greatest sensitivity while controlling for the false discovery rate. Our analysis also provides insight into experimental design and we find that, for a fixed number of samples, better sensitivity and specificity are achieved with higher numbers of replicates than with higher sampling density. Application of the methods to detecting circadian rhythms in a metadataset of microarrays that quantify time-dependent gene expression in whole heads of Drosophila melanogaster reveals annotations that are enriched among genes with highly asymmetric waveforms. These include a wide range of oxidation reduction and metabolic genes, as well as genes with transcripts that have multiple splice forms.

A Conserved Bicycle Model for Circadian Clock Control of Membrane Excitability

“…Using empirical JTK_CYCLE (Hutchison et al, 2015), an updated version of JTK_CYCLE (Hughes et al, 2010), to detect rhythmic transcripts at a false discovery rate of 5% (Benjamini-Hochberg corrected p<0.05), we observed robust 24 h rhythms in CG33988, the fly ortholog of the NCA Localization Factor-1 (NLF-1) but not in na itself nor its regulatory subunits unc79 , unc80 (Lear et al, 2013) and NALCN-activators, Src family kinases (Lu et al, 2009), Src42 and Src64b in flies (Fig. 5A).…”

“…Quantification was performed using eXpress (Roberts et al, 2011). Rhythmic transcripts were detected using empirical JTK_CYCLE (Hutchison et al, 2015), transcripts were considered robustly rhythmic when the Benjamini-Hochberg corrected p value or false discovery rate<0.05. Empirical JTK_CYCLE derives p values empirically considering asymmetric waveforms.…”

Analysis of the circadian rhythm of biological processes in mouse liver and kidney

“…Идентификация генов, демонстрирующих циркад-ную ритмику трансляции, была осуществлена методом, основанным на комбинации критерия, используемого в программе eJTK_CYCLE (empP < 0.05) (Hughes et al, 2010;Hutchison et al, 2015), с последующим анализом разброса измерений в каждой временной точке суточного ритма для выявления устойчивых паттернов экспрессии, воспроизводимых в каждом суточном цикле наблюдения для нескольких реплик (в нашем случае две реплики). Показатель empP в программе eJTK_CYCLE соответ-ствует эмпирической оценке уровня значимости (p-value) с учетом множественности выбора, полученной путем имитационного моделирования (Hutchison et al, 2015).…”

“…Показатель empP в программе eJTK_CYCLE соответ-ствует эмпирической оценке уровня значимости (p-value) с учетом множественности выбора, полученной путем имитационного моделирования (Hutchison et al, 2015). Ритмичный ген должен также удовлетворять допол-нительным условиям: 1) амплитуда колебаний уровня показателя профилирования рибосом гена должна быть не менее 20 % от его среднего значения за сутки; 2) хотя бы для одной временной точки показатель профилиро-вания рибосом в обеих репликах должен превышать его среднесуточное значение.…”

Direct nonparametric estimation of the period and the shape of a periodic component in short duration signals

“…Classically, Fourier Transform (FT) methods are used for this task, but when the duration of the signal is very short, for example only a few periods, these methods lack precision and resolution. Many parametric methods such as Fourier series decompositions can obtain better resolution [2], [3], [4], [5], but the precision of the period and the shape of the component is not often good enough when the shape of the periodic component is far from a sinusoidal form [6], [7], [8], [9] In this work, a new method is presented to determine the period very precisely and estimate the nonparametric shape of that periodic component. The method is based on the description of the problem directly as a linear inverse problem and the Bayesian inference.…”