True and false discovery in DNA microarray experiments: Transcriptome changes in the hippocampus of presenilin 1 mutant mice

Travis Unger, Zeljka Korade, Orly Lazarov, David Terrano, Sangram S. Sisodia, Károly Mirnics

Research output: Contribution to journalArticle

14 Citations (Scopus)

Abstract

In transcriptome profiling experiments using DNA microarrays, it is critical to maximize putatively true data discovery while keeping the false discovery rate at acceptable levels. Using previously published and verified transcriptome datasets of mice with genetically altered PS1 physiology, we present a simple, robust, and system-specific assessment of type I and type II errors in two independent microarray experimental series. We provide evidence to suggest that for maximizing true discovery and minimizing false discovery, statistical criteria alone are inferior to statistical significance plus magnitude of change criteria. Furthermore, we found that, regardless of the exact criteria used for determining differential expression, different data extraction protocols give rise to different discovery and false discovery rates. In addition, a large proportion of expression differences were both dataset and analytical approach dependent. The data assessment methods presented and discussed in this manuscript can be easily carried out on any microarray dataset using basic spreadsheet functions as the only tool needed. Finally, we provide an in-depth analysis of the hippocampal transcriptome of ΔE9 hPS1 transgenic mice and mice with a conditional ablation of the PS1 gene.

Original languageEnglish (US)
Pages (from-to)261-273
Number of pages13
JournalMethods
Volume37
Issue number3
DOIs
StatePublished - Nov 1 2005

Fingerprint

Presenilin-1
Microarrays
Oligonucleotide Array Sequence Analysis
Transcriptome
Hippocampus
Gene Expression Profiling
DNA
Experiments
Spreadsheets
Physiology
Ablation
Transgenic Mice
Genes
Datasets
mouse presenilin 1

Keywords

  • Affymetrix
  • Alzheimer's disease
  • Brain
  • DCHIP
  • DNA microarray
  • Data mining
  • False discovery rate
  • Hippocampus
  • Knockout
  • Latin square
  • MAS5
  • Mutation
  • Normalization
  • Presenilin 1
  • RMA
  • Statistical significance
  • Transgenic

ASJC Scopus subject areas

  • Molecular Biology
  • Biochemistry, Genetics and Molecular Biology(all)

Cite this

True and false discovery in DNA microarray experiments : Transcriptome changes in the hippocampus of presenilin 1 mutant mice. / Unger, Travis; Korade, Zeljka; Lazarov, Orly; Terrano, David; Sisodia, Sangram S.; Mirnics, Károly.

In: Methods, Vol. 37, No. 3, 01.11.2005, p. 261-273.

Research output: Contribution to journalArticle

Unger, Travis ; Korade, Zeljka ; Lazarov, Orly ; Terrano, David ; Sisodia, Sangram S. ; Mirnics, Károly. / True and false discovery in DNA microarray experiments : Transcriptome changes in the hippocampus of presenilin 1 mutant mice. In: Methods. 2005 ; Vol. 37, No. 3. pp. 261-273.
@article{49be80a327aa4da4b47aa5eae0ef0c6c,
title = "True and false discovery in DNA microarray experiments: Transcriptome changes in the hippocampus of presenilin 1 mutant mice",
abstract = "In transcriptome profiling experiments using DNA microarrays, it is critical to maximize putatively true data discovery while keeping the false discovery rate at acceptable levels. Using previously published and verified transcriptome datasets of mice with genetically altered PS1 physiology, we present a simple, robust, and system-specific assessment of type I and type II errors in two independent microarray experimental series. We provide evidence to suggest that for maximizing true discovery and minimizing false discovery, statistical criteria alone are inferior to statistical significance plus magnitude of change criteria. Furthermore, we found that, regardless of the exact criteria used for determining differential expression, different data extraction protocols give rise to different discovery and false discovery rates. In addition, a large proportion of expression differences were both dataset and analytical approach dependent. The data assessment methods presented and discussed in this manuscript can be easily carried out on any microarray dataset using basic spreadsheet functions as the only tool needed. Finally, we provide an in-depth analysis of the hippocampal transcriptome of ΔE9 hPS1 transgenic mice and mice with a conditional ablation of the PS1 gene.",
keywords = "Affymetrix, Alzheimer's disease, Brain, DCHIP, DNA microarray, Data mining, False discovery rate, Hippocampus, Knockout, Latin square, MAS5, Mutation, Normalization, Presenilin 1, RMA, Statistical significance, Transgenic",
author = "Travis Unger and Zeljka Korade and Orly Lazarov and David Terrano and Sisodia, {Sangram S.} and K{\'a}roly Mirnics",
year = "2005",
month = "11",
day = "1",
doi = "10.1016/j.ymeth.2005.09.005",
language = "English (US)",
volume = "37",
pages = "261--273",
journal = "ImmunoMethods",
issn = "1046-2023",
publisher = "Academic Press Inc.",
number = "3",

}

TY - JOUR

T1 - True and false discovery in DNA microarray experiments

T2 - Transcriptome changes in the hippocampus of presenilin 1 mutant mice

AU - Unger, Travis

AU - Korade, Zeljka

AU - Lazarov, Orly

AU - Terrano, David

AU - Sisodia, Sangram S.

AU - Mirnics, Károly

PY - 2005/11/1

Y1 - 2005/11/1

N2 - In transcriptome profiling experiments using DNA microarrays, it is critical to maximize putatively true data discovery while keeping the false discovery rate at acceptable levels. Using previously published and verified transcriptome datasets of mice with genetically altered PS1 physiology, we present a simple, robust, and system-specific assessment of type I and type II errors in two independent microarray experimental series. We provide evidence to suggest that for maximizing true discovery and minimizing false discovery, statistical criteria alone are inferior to statistical significance plus magnitude of change criteria. Furthermore, we found that, regardless of the exact criteria used for determining differential expression, different data extraction protocols give rise to different discovery and false discovery rates. In addition, a large proportion of expression differences were both dataset and analytical approach dependent. The data assessment methods presented and discussed in this manuscript can be easily carried out on any microarray dataset using basic spreadsheet functions as the only tool needed. Finally, we provide an in-depth analysis of the hippocampal transcriptome of ΔE9 hPS1 transgenic mice and mice with a conditional ablation of the PS1 gene.

AB - In transcriptome profiling experiments using DNA microarrays, it is critical to maximize putatively true data discovery while keeping the false discovery rate at acceptable levels. Using previously published and verified transcriptome datasets of mice with genetically altered PS1 physiology, we present a simple, robust, and system-specific assessment of type I and type II errors in two independent microarray experimental series. We provide evidence to suggest that for maximizing true discovery and minimizing false discovery, statistical criteria alone are inferior to statistical significance plus magnitude of change criteria. Furthermore, we found that, regardless of the exact criteria used for determining differential expression, different data extraction protocols give rise to different discovery and false discovery rates. In addition, a large proportion of expression differences were both dataset and analytical approach dependent. The data assessment methods presented and discussed in this manuscript can be easily carried out on any microarray dataset using basic spreadsheet functions as the only tool needed. Finally, we provide an in-depth analysis of the hippocampal transcriptome of ΔE9 hPS1 transgenic mice and mice with a conditional ablation of the PS1 gene.

KW - Affymetrix

KW - Alzheimer's disease

KW - Brain

KW - DCHIP

KW - DNA microarray

KW - Data mining

KW - False discovery rate

KW - Hippocampus

KW - Knockout

KW - Latin square

KW - MAS5

KW - Mutation

KW - Normalization

KW - Presenilin 1

KW - RMA

KW - Statistical significance

KW - Transgenic

UR - http://www.scopus.com/inward/record.url?scp=27844594839&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=27844594839&partnerID=8YFLogxK

U2 - 10.1016/j.ymeth.2005.09.005

DO - 10.1016/j.ymeth.2005.09.005

M3 - Article

C2 - 16308155

AN - SCOPUS:27844594839

VL - 37

SP - 261

EP - 273

JO - ImmunoMethods

JF - ImmunoMethods

SN - 1046-2023

IS - 3

ER -