LimoRhyde2: genomic analysis of biological rhythms based on effect sizes

Abstract

Genome-scale data have revealed daily rhythms in various species and tissues. However, current methods to assess rhythmicity largely restrict their focus to quantifying statistical significance, which may not reflect biological relevance. To address this limitation, we developed a method called LimoRhyde2 (the successor to our method LimoRhyde), which focuses instead on rhythm-related effect sizes and their uncertainty. For each genomic feature, LimoRhyde2 fits a curve using a series of linear models based on periodic splines, moderates the fits using an Empirical Bayes approach called multivariate adaptive shrinkage (Mash), then uses the moderated fits to calculate rhythm statistics such as peak-to-trough amplitude. The periodic splines capture non-sinusoidal rhythmicity, while Mash uses patterns in the data to account for different fits having different levels of noise. To demonstrate LimoRhyde2's utility, we applied it to multiple circadian transcriptome datasets. Overall, LimoRhyde2 prioritized genes having high-amplitude rhythms in expression, whereas a prior method (BooteJTK) prioritized "statistically significant" genes whose amplitudes could be relatively small. Thus, quantifying effect sizes using approaches such as LimoRhyde2 has the potential to transform interpretation of genomic data related to biological rhythms.

A comparative study of algorithms detecting differential rhythmicity in transcriptomic data

“…Positive integer count data that have similar characteristics to RNA-seq read counts (negative binomial residual distribution and variance that changes with the mean of the data) can be analyzed with LimoRhyde2, compareRhythms, and dryR. However, in this case, users should select an option designed for RNA-seq read count data (e.g., voom, DESeq2, or edgeR based method) (21, 25, 27). Alternatively, data can potentially be transformed before analysis so that the residuals have a Gaussian distribution with constant variance, and then analyzed using one of the Gaussian-assuming approaches listed above (23).…”

“…The p-values were manually adjusted using the Benjamini-Hochberg procedure. The LimoRhyde2 R package (version 0.1.0) (25) was run using the RPKM data with the option of “sinusoid = TRUE” for the function of getModelFit(), and the default options were used for other functions.…”

“…In addition, the requirements for input datasets and the output information are also often different. Here, we focus on seven of these algorithms published or posted on bioRxiv in or before 2022: DODR (22), LimoRhyde/LimoRhyde2 (24, 25), CircaCompare (26), compareRhythms (21), diffCircadian (23), dryR (27), and RepeatedCircadian (28), and perform a systematic comparison of their characteristics and performance. We highlight three features of these algorithms: 1) the requirements for input data, 2) the number of rhythmic and differentially rhythmic transcripts detected, and 3) the differences in their outputs.…”

Circatidal control of gene expression in the deep-sea hot vent shrimpRimicaris leurokolos

“…Transcripts with an FDR ≤ 0.05 were considered statistically rhythmic. The root mean square (RMS) amplitude and Rhythmicity Index (RI) of rhythmic transcripts were calculated by LimoRhyde2 101 and autocorrelation analysis, respectively. Significance of amplitude and detected P -values between tidal and daily oscillations was examined with the unpaired Student’s t -test, while the paired Student’s t -test was conducted on the tidal and daily components of rhythmic transcripts.…”

GLMMcosinor: Flexible cosinor modeling with a generalized linear mixed modeling framework to characterize rhythmic time series

“…Differences in such parameters between multiple rhythms can also be estimated for rhythmic time series datasets belonging to two or more experimental groups. Quantifying and estimating the characteristics of a rhythm is distinctly different from detecting whether a rhythm exists, for which there are several other methods available (Hughes, et al, 2010; Hutchison, et al, 2015; Obodo, et al, 2023). Our study focuses on advancing the methods to characterize rhythms, not detecting them.…”