Exploratory analysis is an important first step for discovering latent struc-
ture and generating hypotheses in large biological data sets. However, when
the number of variables is large compared to the number of samples, standard
methods such as principal components analysis give results that are unstable
and difficult to interpret.
Here, we present adaptive generalized principal components analysis
(adaptive gPCA), a new method that solves these problems by incorporat-
ing information about the relationships among the variables. Adaptive gPCA
gives a low-dimensional representation of the samples with axes that are inter-
pretable in terms of groups of closely related variables. We show that adaptive
gPCA does well at reconstructing true latent structure in simulated data and
demonstrate its use on a study of the effect of antibiotics on the human gut
microbiota.
%0 Journal Article
%1 fukuyama2019adaptive
%A Fukuyama, Julia
%D 2019
%I Institute of Mathematical Statistics
%J The Annals of Applied Statistics
%K PCA methods phylogenetics
%N 2
%R 10.1214/18-aoas1227
%T Adaptive gPCA: A method for structured dimensionality reduction with applications to microbiome data
%U https://doi.org/10.1214%2F18-aoas1227
%V 13
%X Exploratory analysis is an important first step for discovering latent struc-
ture and generating hypotheses in large biological data sets. However, when
the number of variables is large compared to the number of samples, standard
methods such as principal components analysis give results that are unstable
and difficult to interpret.
Here, we present adaptive generalized principal components analysis
(adaptive gPCA), a new method that solves these problems by incorporat-
ing information about the relationships among the variables. Adaptive gPCA
gives a low-dimensional representation of the samples with axes that are inter-
pretable in terms of groups of closely related variables. We show that adaptive
gPCA does well at reconstructing true latent structure in simulated data and
demonstrate its use on a study of the effect of antibiotics on the human gut
microbiota.
@article{fukuyama2019adaptive,
abstract = {Exploratory analysis is an important first step for discovering latent struc-
ture and generating hypotheses in large biological data sets. However, when
the number of variables is large compared to the number of samples, standard
methods such as principal components analysis give results that are unstable
and difficult to interpret.
Here, we present adaptive generalized principal components analysis
(adaptive gPCA), a new method that solves these problems by incorporat-
ing information about the relationships among the variables. Adaptive gPCA
gives a low-dimensional representation of the samples with axes that are inter-
pretable in terms of groups of closely related variables. We show that adaptive
gPCA does well at reconstructing true latent structure in simulated data and
demonstrate its use on a study of the effect of antibiotics on the human gut
microbiota.},
added-at = {2021-05-27T19:04:49.000+0200},
author = {Fukuyama, Julia},
biburl = {https://www.bibsonomy.org/bibtex/26bae66863ca15b930b79480349f8c888/peter.ralph},
doi = {10.1214/18-aoas1227},
interhash = {0e023415dfb15c6a5fdcfcf39f17d68d},
intrahash = {6bae66863ca15b930b79480349f8c888},
journal = {The Annals of Applied Statistics},
keywords = {PCA methods phylogenetics},
month = jun,
number = 2,
publisher = {Institute of Mathematical Statistics},
timestamp = {2021-05-27T19:04:49.000+0200},
title = {Adaptive {gPCA}: A method for structured dimensionality reduction with applications to microbiome data},
url = {https://doi.org/10.1214%2F18-aoas1227},
volume = 13,
year = 2019
}