Cargando…

Optimal multiple testing under a Gaussian prior on the effect sizes

We develop a new method for large-scale frequentist multiple testing with Bayesian prior information. We find optimal [Formula: see text]-value weights that maximize the average power of the weighted Bonferroni method. Due to the nonconvexity of the optimization problem, previous methods that accoun...

Descripción completa

Detalles Bibliográficos
Autores principales: Dobriban, Edgar, Fortney, Kristen, Kim, Stuart K., Owen, Art B.
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Oxford University Press 2015
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4813057/
https://www.ncbi.nlm.nih.gov/pubmed/27046938
http://dx.doi.org/10.1093/biomet/asv050
_version_ 1782424225779286016
author Dobriban, Edgar
Fortney, Kristen
Kim, Stuart K.
Owen, Art B.
author_facet Dobriban, Edgar
Fortney, Kristen
Kim, Stuart K.
Owen, Art B.
author_sort Dobriban, Edgar
collection PubMed
description We develop a new method for large-scale frequentist multiple testing with Bayesian prior information. We find optimal [Formula: see text]-value weights that maximize the average power of the weighted Bonferroni method. Due to the nonconvexity of the optimization problem, previous methods that account for uncertain prior information are suitable for only a small number of tests. For a Gaussian prior on the effect sizes, we give an efficient algorithm that is guaranteed to find the optimal weights nearly exactly. Our method can discover new loci in genome-wide association studies and compares favourably to competitors. An open-source implementation is available.
format Online
Article
Text
id pubmed-4813057
institution National Center for Biotechnology Information
language English
publishDate 2015
publisher Oxford University Press
record_format MEDLINE/PubMed
spelling pubmed-48130572016-12-01 Optimal multiple testing under a Gaussian prior on the effect sizes Dobriban, Edgar Fortney, Kristen Kim, Stuart K. Owen, Art B. Biometrika Articles We develop a new method for large-scale frequentist multiple testing with Bayesian prior information. We find optimal [Formula: see text]-value weights that maximize the average power of the weighted Bonferroni method. Due to the nonconvexity of the optimization problem, previous methods that account for uncertain prior information are suitable for only a small number of tests. For a Gaussian prior on the effect sizes, we give an efficient algorithm that is guaranteed to find the optimal weights nearly exactly. Our method can discover new loci in genome-wide association studies and compares favourably to competitors. An open-source implementation is available. Oxford University Press 2015-12 2015-11-04 /pmc/articles/PMC4813057/ /pubmed/27046938 http://dx.doi.org/10.1093/biomet/asv050 Text en © 2015 Biometrika Trust
spellingShingle Articles
Dobriban, Edgar
Fortney, Kristen
Kim, Stuart K.
Owen, Art B.
Optimal multiple testing under a Gaussian prior on the effect sizes
title Optimal multiple testing under a Gaussian prior on the effect sizes
title_full Optimal multiple testing under a Gaussian prior on the effect sizes
title_fullStr Optimal multiple testing under a Gaussian prior on the effect sizes
title_full_unstemmed Optimal multiple testing under a Gaussian prior on the effect sizes
title_short Optimal multiple testing under a Gaussian prior on the effect sizes
title_sort optimal multiple testing under a gaussian prior on the effect sizes
topic Articles
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4813057/
https://www.ncbi.nlm.nih.gov/pubmed/27046938
http://dx.doi.org/10.1093/biomet/asv050
work_keys_str_mv AT dobribanedgar optimalmultipletestingunderagaussianpriorontheeffectsizes
AT fortneykristen optimalmultipletestingunderagaussianpriorontheeffectsizes
AT kimstuartk optimalmultipletestingunderagaussianpriorontheeffectsizes
AT owenartb optimalmultipletestingunderagaussianpriorontheeffectsizes