schliessen

Filtern

 

Bibliotheken

Permutation tests for phylogenetic comparative analyses of high-dimensional shape data: What you shuffle matters

Evaluating statistical trends in high‐dimensional phenotypes poses challenges for comparative biologists, because the high‐dimensionality of the trait data relative to the number of species can prohibit parametric tests from being computed. Recently, two comparative methods were proposed to circumve... Full description

Journal Title: Evolution 2015-03, Vol.69 (3), p.823-829
Main Author: Adams, Dean C
Other Authors: Collyer, Michael L
Format: Electronic Article Electronic Article
Language: English
Subjects:
Publisher: United States: Blackwell Publishing Ltd
ID: ISSN: 0014-3820
Link: https://www.ncbi.nlm.nih.gov/pubmed/25641367
Zum Text:
SendSend as email Add to Book BagAdd to Book Bag
Staff View
recordid: cdi_proquest_miscellaneous_1664447653
title: Permutation tests for phylogenetic comparative analyses of high-dimensional shape data: What you shuffle matters
format: Article
creator:
  • Adams, Dean C
  • Collyer, Michael L
subjects:
  • Animals
  • Antioxidants
  • Biologists
  • Body Size
  • Cladistic analysis
  • Comparative analysis
  • Evolutionary biology
  • Genotype & phenotype
  • Geometric morphometrics
  • Head - anatomy & histology
  • Least-Squares Analysis
  • Linear Models
  • Models, Genetic
  • phylogenetic comparative method
  • phylogenetic generalized least squares
  • phylogenetic independent contrasts
  • Phylogenetics
  • Phylogeny
  • Research
  • Urodela - anatomy & histology
  • Urodela - genetics
  • Usage
ispartof: Evolution, 2015-03, Vol.69 (3), p.823-829
description: Evaluating statistical trends in high‐dimensional phenotypes poses challenges for comparative biologists, because the high‐dimensionality of the trait data relative to the number of species can prohibit parametric tests from being computed. Recently, two comparative methods were proposed to circumvent this difficulty. One obtains phylogenetic independent contrasts for all variables, and statistically evaluates the linear model by permuting the phylogenetically independent contrasts (PICs) of the response data. The other uses a distance‐based approach to obtain coefficients for generalized least squares models (D‐PGLS), and subsequently permutes the original data to evaluate the model effects. Here, we show that permuting PICs is not equivalent to permuting the data prior to the analyses as in D‐PGLS. We further explain why PICs are not the correct exchangeable units under the null hypothesis, and demonstrate that this misspecification of permutable units leads to inflated type I error rates of statistical tests. We then show that simply shuffling the original data and recalculating the independent contrasts with each iteration yields significance levels that correspond to those found using D‐PGLS. Thus, while summary statistics from methods based on PICs and PGLS are the same, permuting PICs can lead to strikingly different inferential outcomes with respect to statistical and biological inferences.
language: eng
source:
identifier: ISSN: 0014-3820
fulltext: no_fulltext
issn:
  • 0014-3820
  • 1558-5646
url: Link


@attributes
NO1
SEARCH_ENGINEprimo_central_multiple_fe
SEARCH_ENGINE_TYPEPrimo Central Search Engine
RANK2.2562091
LOCALfalse
PrimoNMBib
record
control
sourceidgale_opena
recordidTN_cdi_proquest_miscellaneous_1664447653
sourceformatXML
sourcesystemPC
galeidA411137545
sourcerecordidA411137545
originalsourceidFETCH-LOGICAL-15446-10e5d022f7552c93d6ef30a87c9e24a5d2ce6cebb71dad51f46623f0a740df740
addsrcrecordideNqNkl9vFCEUxSdGY2v1wS9gSHzxZbbA8GfHt6apVVOtTdT6Rli47FJnhhGY6n572d12TfRFSCAhv3O4XE5VPSd4Rso4htswI5S34kF1SDif11ww8bA6xJiwuplTfFA9SekGY9xy0j6uDmgBSCPkYTV-gthPWWcfBpQh5YRciGhcrbuwhAGyN8iEftSxILeA9KC7dYKEgkMrv1zV1vcwpKLWHUorPQKyOuvX6HqlM1qHqRxOznWAep0zxPS0euR0l-DZ3X5UfXlz9vn0bX1xef7u9OSiJpwxURMM3GJKneScmraxAlyD9VyaFijT3FIDwsBiIYnVlhPHhKCNw1oybF1Zjqr3O98wwqB9BDVG3-u4VkF7ZcvL1KZQ4zMo5py1DSbzlgu8aKUpBpI1ZN7QVmDGi9mrndkYw4-pdEn1PhnoOj1AmJIiQoqGb9r9H6hgjEnBm4K-_Au9CVMsjdxSnLblT2mhZjtqqTtQfnAhR23KtNB7EwZwvpyfsBKERvJtsS_ubKdFD3b_7vtPL8DxDjAxpBTBqdKFbQKKs-8UwWoTK1Vipbax-qP4We5a7x3_AdXZ18t7Rb1T-JTh116h43clZKlTXX88V1fsChP5Qahv-DePwNng
sourcetypeOpen Access Repository
isCDItrue
recordtypearticle
pqid1665291252
display
typearticle
titlePermutation tests for phylogenetic comparative analyses of high-dimensional shape data: What you shuffle matters
creatorAdams, Dean C ; Collyer, Michael L
creatorcontribAdams, Dean C ; Collyer, Michael L
descriptionEvaluating statistical trends in high‐dimensional phenotypes poses challenges for comparative biologists, because the high‐dimensionality of the trait data relative to the number of species can prohibit parametric tests from being computed. Recently, two comparative methods were proposed to circumvent this difficulty. One obtains phylogenetic independent contrasts for all variables, and statistically evaluates the linear model by permuting the phylogenetically independent contrasts (PICs) of the response data. The other uses a distance‐based approach to obtain coefficients for generalized least squares models (D‐PGLS), and subsequently permutes the original data to evaluate the model effects. Here, we show that permuting PICs is not equivalent to permuting the data prior to the analyses as in D‐PGLS. We further explain why PICs are not the correct exchangeable units under the null hypothesis, and demonstrate that this misspecification of permutable units leads to inflated type I error rates of statistical tests. We then show that simply shuffling the original data and recalculating the independent contrasts with each iteration yields significance levels that correspond to those found using D‐PGLS. Thus, while summary statistics from methods based on PICs and PGLS are the same, permuting PICs can lead to strikingly different inferential outcomes with respect to statistical and biological inferences.
identifier
0ISSN: 0014-3820
1EISSN: 1558-5646
2DOI: 10.1111/evo.12596
3PMID: 25641367
languageeng
publisherUnited States: Blackwell Publishing Ltd
subjectAnimals ; Antioxidants ; Biologists ; Body Size ; Cladistic analysis ; Comparative analysis ; Evolutionary biology ; Genotype & phenotype ; Geometric morphometrics ; Head - anatomy & histology ; Least-Squares Analysis ; Linear Models ; Models, Genetic ; phylogenetic comparative method ; phylogenetic generalized least squares ; phylogenetic independent contrasts ; Phylogenetics ; Phylogeny ; Research ; Urodela - anatomy & histology ; Urodela - genetics ; Usage
ispartofEvolution, 2015-03, Vol.69 (3), p.823-829
rights
02015 The Author(s).
1Copyright Wiley Subscription Services, Inc. Mar 2015
citedbyFETCH-LOGICAL-15446-10e5d022f7552c93d6ef30a87c9e24a5d2ce6cebb71dad51f46623f0a740df740
links
openurl$$Topenurl_article
thumbnail$$Usyndetics_thumb_exl
backlink$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/25641367$$D View this record in MEDLINE/PubMed
search
creatorcontrib
0Adams, Dean C
1Collyer, Michael L
title
0Permutation tests for phylogenetic comparative analyses of high-dimensional shape data: What you shuffle matters
1Evolution
addtitleEvolution
descriptionEvaluating statistical trends in high‐dimensional phenotypes poses challenges for comparative biologists, because the high‐dimensionality of the trait data relative to the number of species can prohibit parametric tests from being computed. Recently, two comparative methods were proposed to circumvent this difficulty. One obtains phylogenetic independent contrasts for all variables, and statistically evaluates the linear model by permuting the phylogenetically independent contrasts (PICs) of the response data. The other uses a distance‐based approach to obtain coefficients for generalized least squares models (D‐PGLS), and subsequently permutes the original data to evaluate the model effects. Here, we show that permuting PICs is not equivalent to permuting the data prior to the analyses as in D‐PGLS. We further explain why PICs are not the correct exchangeable units under the null hypothesis, and demonstrate that this misspecification of permutable units leads to inflated type I error rates of statistical tests. We then show that simply shuffling the original data and recalculating the independent contrasts with each iteration yields significance levels that correspond to those found using D‐PGLS. Thus, while summary statistics from methods based on PICs and PGLS are the same, permuting PICs can lead to strikingly different inferential outcomes with respect to statistical and biological inferences.
subject
0Animals
1Antioxidants
2Biologists
3Body Size
4Cladistic analysis
5Comparative analysis
6Evolutionary biology
7Genotype & phenotype
8Geometric morphometrics
9Head - anatomy & histology
10Least-Squares Analysis
11Linear Models
12Models, Genetic
13phylogenetic comparative method
14phylogenetic generalized least squares
15phylogenetic independent contrasts
16Phylogenetics
17Phylogeny
18Research
19Urodela - anatomy & histology
20Urodela - genetics
21Usage
issn
00014-3820
11558-5646
fulltextfalse
rsrctypearticle
creationdate2015
recordtypearticle
recordideNqNkl9vFCEUxSdGY2v1wS9gSHzxZbbA8GfHt6apVVOtTdT6Rli47FJnhhGY6n572d12TfRFSCAhv3O4XE5VPSd4Rso4htswI5S34kF1SDif11ww8bA6xJiwuplTfFA9SekGY9xy0j6uDmgBSCPkYTV-gthPWWcfBpQh5YRciGhcrbuwhAGyN8iEftSxILeA9KC7dYKEgkMrv1zV1vcwpKLWHUorPQKyOuvX6HqlM1qHqRxOznWAep0zxPS0euR0l-DZ3X5UfXlz9vn0bX1xef7u9OSiJpwxURMM3GJKneScmraxAlyD9VyaFijT3FIDwsBiIYnVlhPHhKCNw1oybF1Zjqr3O98wwqB9BDVG3-u4VkF7ZcvL1KZQ4zMo5py1DSbzlgu8aKUpBpI1ZN7QVmDGi9mrndkYw4-pdEn1PhnoOj1AmJIiQoqGb9r9H6hgjEnBm4K-_Au9CVMsjdxSnLblT2mhZjtqqTtQfnAhR23KtNB7EwZwvpyfsBKERvJtsS_ubKdFD3b_7vtPL8DxDjAxpBTBqdKFbQKKs-8UwWoTK1Vipbax-qP4We5a7x3_AdXZ18t7Rb1T-JTh116h43clZKlTXX88V1fsChP5Qahv-DePwNng
startdate201503
enddate201503
creator
0Adams, Dean C
1Collyer, Michael L
general
0Blackwell Publishing Ltd
1Society for the Study of Evolution
2Wiley Subscription Services, Inc
3Wiley
scope
0BSCLL
1CGR
2CUY
3CVF
4ECM
5EIF
6NPM
7BSHEE
87QG
97QL
107QP
117QR
127SN
137SS
147TK
157TM
167U9
178FD
18C1K
19FR3
20H94
21M7N
22P64
23RC3
247X8
25CLFQK
sort
creationdate201503
titlePermutation tests for phylogenetic comparative analyses of high-dimensional shape data: What you shuffle matters
authorAdams, Dean C ; Collyer, Michael L
facets
frbrtype5
frbrgroupidcdi_FETCH-LOGICAL-15446-10e5d022f7552c93d6ef30a87c9e24a5d2ce6cebb71dad51f46623f0a740df740
rsrctypearticles
prefilterarticles
languageeng
creationdate2015
topic
0Animals
1Antioxidants
2Biologists
3Body Size
4Cladistic analysis
5Comparative analysis
6Evolutionary biology
7Genotype & phenotype
8Geometric morphometrics
9Head - anatomy & histology
10Least-Squares Analysis
11Linear Models
12Models, Genetic
13phylogenetic comparative method
14phylogenetic generalized least squares
15phylogenetic independent contrasts
16Phylogenetics
17Phylogeny
18Research
19Urodela - anatomy & histology
20Urodela - genetics
21Usage
creatorcontrib
0Adams, Dean C
1Collyer, Michael L
collection
0Istex
1Medline
2MEDLINE
3MEDLINE (Ovid)
4MEDLINE
5MEDLINE
6PubMed
7Academic OneFile (A&I only)
8Animal Behavior Abstracts
9Bacteriology Abstracts (Microbiology B)
10Calcium & Calcified Tissue Abstracts
11Chemoreception Abstracts
12Ecology Abstracts
13Entomology Abstracts (Full archive)
14Neurosciences Abstracts
15Nucleic Acids Abstracts
16Virology and AIDS Abstracts
17Technology Research Database
18Environmental Sciences and Pollution Management
19Engineering Research Database
20AIDS and Cancer Research Abstracts
21Algology Mycology and Protozoology Abstracts (Microbiology C)
22Biotechnology and BioEngineering Abstracts
23Genetics Abstracts
24MEDLINE - Academic
25OpenAIRE
jtitleEvolution
delivery
delcategoryRemote Search Resource
fulltextno_fulltext
addata
au
0Adams, Dean C
1Collyer, Michael L
formatjournal
genrearticle
ristypeJOUR
atitlePermutation tests for phylogenetic comparative analyses of high-dimensional shape data: What you shuffle matters
jtitleEvolution
addtitleEvolution
date2015-03
risdate2015
volume69
issue3
spage823
epage829
pages823-829
issn0014-3820
eissn1558-5646
notesNSF - No. DEB-1257287
abstractEvaluating statistical trends in high‐dimensional phenotypes poses challenges for comparative biologists, because the high‐dimensionality of the trait data relative to the number of species can prohibit parametric tests from being computed. Recently, two comparative methods were proposed to circumvent this difficulty. One obtains phylogenetic independent contrasts for all variables, and statistically evaluates the linear model by permuting the phylogenetically independent contrasts (PICs) of the response data. The other uses a distance‐based approach to obtain coefficients for generalized least squares models (D‐PGLS), and subsequently permutes the original data to evaluate the model effects. Here, we show that permuting PICs is not equivalent to permuting the data prior to the analyses as in D‐PGLS. We further explain why PICs are not the correct exchangeable units under the null hypothesis, and demonstrate that this misspecification of permutable units leads to inflated type I error rates of statistical tests. We then show that simply shuffling the original data and recalculating the independent contrasts with each iteration yields significance levels that correspond to those found using D‐PGLS. Thus, while summary statistics from methods based on PICs and PGLS are the same, permuting PICs can lead to strikingly different inferential outcomes with respect to statistical and biological inferences.
copUnited States
pubBlackwell Publishing Ltd
pmid25641367
doi10.1111/evo.12596
tpages7