discoODAs: Execute Oscillation Detection Using DiscoRhythm

In matthewcarlucci/DiscoRhythm: Interactive Workflow for Discovering Rhythmicity in Biological Data

Execute Oscillation Detection Using DiscoRhythm

Description

Runs specified oscillation detection algorithms (ODAs) sequentially to obtain oscillation characteristics for each row of the input data.

Usage

discoGetODAs(se, method = NULL, period, circular_t = FALSE)

discoODAs(
  se,
  period = 24,
  method = c("CS", "JTK", "LS", "ARS"),
  circular_t = FALSE,
  ncores = 1
)

Arguments

se	SummarizedExperiment, the main data object used by DiscoRhythm expected to contain se$ID, se$ReplicateID, se$Time sample metadata and non-null rownames. See the vignette for more details.
method	character, short names of ODAs to use. If length>1 all input method names will be evaluated.
period	numeric, the hypothesized period to test for.
circular_t	logical, is time circular on some base-cycle (ex. time of day). See the DiscoRhythm vignette for details.
ncores	numeric, number of cores to parallelize with (applicable to JTK, ARSER and LS only). If 1, will execute in serial.

Details

There are currently 4 available algorithms for rhythm detection:

• CS = Cosinor (Cornelissen,G. 2014): a.k.a “Harmonic Regression” fits a sinusoid with a free phase parameter.

• LS = Lomb-Scargle (Glynn, 2006): an approach using spectral power density.

• ARS = ARSER (Yang, 2010): removes linear trends and performs the Cosinor test.

• JTK = JTK Cycle (Hughes, 2010): non-parametric test of rhythmicity robust to outliers.

LS, ARS, and JTK results come directly from MetaCycle meta2d() output using the specified fixed period. ARSmle is set to "nomle" and no method integration is used (see meta2d documentation for details).

CS is implemented directly in DiscoRhythm's lmCSmat() as the single-component cosinor described in Cornelissen,G. (2014).

All q-values are calculated by performing p.adjust() on the resulting p-values with method="fdr".

Technical replicates are expected to be merged (likely by discoRepAnalysis) prior to usage of discoODAs.

The discoGetODAs function is called by discoODAs to determine if the selected methods may be used. If any methods are not valid, a warning will be thrown and only valid methods will be computed. discoGetODAs is not typically used directly, however, it may be called by the user to determine if the provided SummarizedExperiment is suitable for use with the specified methods.

Value

A named list of results where each element is a data.frame for the corresponding method with rownames corresponding to the feature identifiers and columns containing estimates for:

• acrophase

• amplitude

• p-value

• q-value

Additional columns relevant to each method will be present.

References

Yang R. and Su Z. (2010). Analyzing circadian expression data by harmonic regression based on autoregressive spectral estimation. Bioinformatics, 26(12), i168–i174.

Hughes M. E., Hogenesch J. B. and Kornacker K. (2010). JTK_CYCLE: an efficient nonparametric algorithm for detecting rhythmic components in genome-scale data sets. Journal of Biological Rhythms, 25(5), 372–380.

Glynn E. F., Chen J. and Mushegian A. R. (2006). Detecting periodic patterns in unevenly spaced gene expression time series using Lomb-Scargle periodograms. Bioinformatics, 22(3), 310–316.

Cornelissen,G. (2014) Cosinor-based rhythmometry. Theor. Biol. Med. Model., 11, 16.

See Also

lmCSmat meta2d

Examples

# Return valid ODAs for example dataset
discoGetODAs(discoGetSimu(as_se=TRUE),period=24)

# Import the simulated example dataset
se <- discoCheckInput(discoGetSimu(TRUE))

# Use discoRepAnalysis to average technical replicates
se_merged <- discoRepAnalysis(se,aov_pcut=1)$se

# Execute the Cosinor and JTK methods with a 24hr period
discoODAres <- discoODAs(se_merged,method=c("CS","JTK"))

# Get the index of rhythmic features detected by both methods at qvalue<0.05
idx <- which(discoODAres$CS$qvalue<0.05 & discoODAres$JTK$qvalue<0.05)

# Get the identifiers for common rhythmic features
rownames(se_merged)[idx]

matthewcarlucci/DiscoRhythm documentation built on March 24, 2022, 10:22 a.m.