Smoothing by Adaptive Shrinkage

Fast, wavelet-based Empirical Bayes shrinkage methods for signal denoising, including smoothing Poisson-distributed data and Gaussian-distributed data with possibly heteroskedastic error. The algorithms implement the methods described Z. Xing, P. Carbonetto & M. Stephens (2021) < https://jmlr.org/papers/v22/19-042.html>.

smashr: smoothing using Adaptive Shrinkage in R

This R package implements fast, wavelet-based Empirical Bayes shrinkage methods for signal denoising. This includes smoothing Poisson-distributed data and Gaussian-distributed data, with possibly heteroskedastic error. The algorithms implement the methods described in Xing, Carbonetto & Stephens (2021).

If you find a bug, please post an issue.

License

Copyright (c) 2016-2021, Zhengrong Xing, Peter Carbonetto and Matthew Stephens.

All source code and software in this repository is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version.

Citing this work

If you find that this R package useful for your work, please cite our paper:

Zhengrong Xing, Peter Carbonetto and Matthew Stephens (2021). Flexible signal denoising via flexible empirical Bayes shrinkage. Journal of Machine Learning Research 22(93), 1-28.

Quick Start

Follow these steps to quickly get started using smashr.

1. In R, install the latest version of smashr using devtools:

   install.packages("devtools")
   library(devtools)
   install_github("stephenslab/smashr")
   

   If you are interested in replicating results from the paper, we recommendg installing smashr 1.2-7:

   install_github("stephenslab/[email protected]")
   

   This will build the smashr package without the vignettes. To build with the vignettes, do this instead:

   install_github("stephenslab/smashr",build_vignettes = TRUE)
   

   We caution that some of the simulation examples may take a long time to run (20--30 minutes, or possibly longer). Also note that the install_github call should also install any missing packages that are required for smashr to work.

2. Load the smashr package, and run the smashr demo:

   library(smashr)
   demo("smashr")
   

3. To learn more, see the smashr package help and the smashr vignette (which you can also view here):

   help(package = "smashr")
   vignette("smashr")
   

Credits

This R package was developed by Zhengrong Xing and Matthew Stephens at the University of Chicago, with contributions from Peter Carbonetto.