Documentation and vignettes are available here.

Quick start

Install the latest version of flashier from GitHub:

install.packages("remotes")
remotes::install_github("willwerscheid/flashier")

Once you have installed the package, load the package in R:

Next, run an example analysis of the GTEx data set:

data(gtex)
fl <- flash(gtex,greedy_Kmax = 3,backfit = TRUE)

For a more detailed introduction to flashier, see the introductory vignette and, later, the advanced vignette.

To learn more, visit the package website, and view the “flash” help page in R:

help("flash")

Citing this work

If you find the flashier package or any of the source code in this repository useful for your work, please cite:

Jason Willwerscheid, Peter Carbonetto and Matthew Stephens. ebnm: an R package for solving the empirical Bayes normal means problem using a variety of prior families. arXiv:2110.00152.

License

Copyright (c) 2018-2024, Jason Willwerscheid, Peter Carbonetto and Matthew Stephens.

All source code and software in this repository are made available under the terms of the MIT license.

Notes

Please note that function names changed in flashier version 0.2.44. If you are using flashier code from previous versions, please make the following substitutions, and note that many parameter names have changed as well:

Old Name New Name
as.ebnm.fn flash_ebnm
conv.crit.elbo flash_conv_crit_elbo_diff
conv.crit.factors flash_conv_crit_max_chg_F
conv.crit.loadings flash_conv_crit_max_chg_L
display.elbo flash_verbose_elbo
display.elbo.diff flash_verbose_elbo_diff
display.F.max.chg flash_verbose_max_chg_F
display.L.max.chg flash_verbose_max_chg_L
display.max.chg flash_verbose_max_chg
ff.elbo flash_fit_get_elbo
ff.est.tau flash_fit_get_est_tau
ff.fixed.tau flash_fit_get_fixed_tau
ff.g flash_fit_get_g
ff.KL flash_fit_get_KL
ff.p2m flash_fit_get_p2m
ff.pm flash_fit_get_pm
ff.tau flash_fit_get_tau
flash.add.greedy flash_greedy
flash.backfit flash_backfit
flash.fit flash_fit
flash.fix.factors flash_factors_fix
flash.init flash_init
flash.init.factors flash_factors_init
flash.nullcheck flash_nullcheck
flash.remove.factors flash_factors_remove
flash.reorder.factors flash_factors_reorder
flash.set.factors.to.zero flash_factors_set_to_zero
flash.set.verbose flash_set_verbose
init.fn.default flash_greedy_init_default
init.fn.irlba flash_greedy_init_irlba
init.fn.softImpute flash_greedy_init_softImpute

Credits

The flashier R package was developed by Jason Willwerscheid, Peter Carbonetto and Matthew Stephens, with many other contributors.