Correcting for Dependent P-values in Rhythm Detection

Mapping Intimacies ◽

10.1101/118547 ◽

2017 ◽

Cited By ~ 5

Author(s):

Alan L. Hutchison ◽

Aaron R. Dinner

Keyword(s):

Time Series ◽

Statistical Significance ◽

P Values ◽

Circadian Genes ◽

Genome Wide ◽

Genome Wide Expression ◽

Expression Time

AbstractThere is much interest in using genome-wide expression time series to identify circadian genes. Several methods have been developed to test for rhythmicity in sparsely sampled time series typical of such measurements. Because these methods are statistical in nature, they rely on estimating the probabilities that patterns arise by chance (i.e., p-values). Here we show that leading methods implicitly make inappropriate assumptions of independence when estimating p-values. We show how to correct for the dependence to obtain accurate estimates for statistical significance during rhythm detection.

Download Full-text

Bootstrapping and Empirical Bayes Methods Improve Rhythm Detection in Sparsely Sampled Data

10.1101/118521 ◽

2017 ◽

Cited By ~ 2

Author(s):

Alan L. Hutchison ◽

Ravi Allada ◽

Aaron R. Dinner

Keyword(s):

Time Series ◽

Empirical Bayes ◽

Rank Order ◽

High Time Resolution ◽

Periodic Patterns ◽

Bayes Methods ◽

Genome Wide ◽

Empirical Bayes Methods ◽

Genome Wide Expression ◽

Expression Time

AbstractMotivationThere is much interest in using genome-wide expression time series to identify circadian genes. However, the cost and effort of such measurements often limits data collection. Consequently, it is difficult to assess the experimental uncertainty in the measurements and, in turn, to detect periodic patterns with statistical confidence.ResultsWe show that parametric bootstrapping and empirical Bayes methods for variance shrinkage can improve rhythm detection in genome-wide expression time series. We demonstrate these approaches by building on the empirical JTK_CYCLE method (eJTK) to formulate a method that we term BooteJTK. Our procedure rapidly and accurately detects cycling time series by combining information about measurement uncertainty with information about the rank order of the time series values. We exploit a publicly available genome-wide dataset with high time resolution to show that BooteJTK provides more consistent rhythm detection thanexisting methods at typical sampling frequencies. Then, we apply BooteJTK to genome-wide expression time series from multiple tissues and show that it reveals biologically sensible tissue relationships that eJTK misses.AvailabilityBootstrap eJTK (BooteJTK) is implemented in Python and is freely available on GitHub at https://github.com/alanlhutchison/BooteJTK.

Download Full-text

Bootstrapping and Empirical Bayes Methods Improve Rhythm Detection in Sparsely Sampled Data

Journal of Biological Rhythms ◽

10.1177/0748730418789536 ◽

2018 ◽

Vol 33 (4) ◽

pp. 339-349 ◽

Cited By ~ 13

Author(s):

Alan L. Hutchison ◽

Ravi Allada ◽

Aaron R. Dinner

Keyword(s):

Time Series ◽

Empirical Bayes ◽

High Time Resolution ◽

Periodic Patterns ◽

Data Set ◽

Bayes Methods ◽

Genome Wide ◽

Empirical Bayes Methods ◽

Genome Wide Expression ◽

Expression Time

There is much interest in using genome-wide expression time series to identify circadian genes. However, the cost and effort of such measurements often limit data collection. Consequently, it is difficult to assess the experimental uncertainty in the measurements and, in turn, to detect periodic patterns with statistical confidence. We show that parametric bootstrapping and empirical Bayes methods for variance shrinkage can improve rhythm detection in genome-wide expression time series. We demonstrate these approaches by building on the empirical JTK_CYCLE method (eJTK) to formulate a method that we term BooteJTK. Our procedure rapidly and accurately detects cycling time series by combining information about measurement uncertainty with information about the rank order of the time series values. We exploit a publicly available genome-wide data set with high time resolution to show that BooteJTK provides more consistent rhythm detection than existing methods at typical sampling frequencies. Then, we apply BooteJTK to genome-wide expression time series from multiple tissues and show that it reveals biologically sensible tissue relationships that eJTK misses. BooteJTK is implemented in Python and is freely available on GitHub at https://github.com/alanlhutchison/BooteJTK .

Download Full-text

Comparison of Six Statistical Methods for Interrupted time Series Studies: Empirical Evaluation of 190 Published Series

10.21203/rs.3.rs-118335/v1 ◽

2020 ◽

Author(s):

Simon Turner ◽

Amalia Karahalios ◽

Andrew Forbes ◽

Monica Taljaard ◽

Jeremy Grimshaw ◽

...

Keyword(s):

Time Series ◽

Statistical Method ◽

Statistical Methods ◽

Statistical Significance ◽

Empirical Evaluation ◽

Interrupted Time Series ◽

Series Data ◽

Standard Errors ◽

P Values ◽

The Impact

Abstract Background The Interrupted Time Series (ITS) is a quasi-experimental design commonly used in public health to evaluate the impact of interventions or exposures. Multiple statistical methods are available to analyse data from ITS studies, but no empirical investigation has examined how the different methods compare when applied to real-world datasets. MethodsA random sample of 200 ITS studies identified in a previous methods review were included. Time series data from each of these studies was sought. Each dataset was re-analysed using six statistical methods. Point and confidence interval estimates for level and slope changes, standard errors, p-values and estimates of autocorrelation were compared between methods. ResultsFrom the 200 ITS studies, including 230 time series, 190 datasets were obtained. We found that the choice of statistical method can importantly affect the level and slope change point estimates, their standard errors, width of confidence intervals and p-values. Statistical significance (categorised at the 5% level) often differed across the pairwise comparisons of methods, ranging from 4% to 25% disagreement. Estimates of autocorrelation differed depending on the method used and the length of the series. ConclusionsThe choice of statistical method in ITS studies can lead to substantially different conclusions about the impact of the interruption. Pre-specification of the statistical method is encouraged, and naive conclusions based on statistical significance should be avoided.

Download Full-text