Extreme selection strategies in gene mapping studies of oligogenic quantitative traits do not always increase power

David B. Allison, Moonseong Heo, Nicholas J. Schork, Su Ling Wong, Robert C. Elston

Research output: Contribution to journalArticle

53 Citations (Scopus)

Abstract

It is well known that obtaining adequate statistical power to detect linkage to or association with genes for complex quantitative traits can be very difficult. In response, investigators have developed a number of power-enhancing strategies that consider restraints such as genotyping (and/or phenotyping) costs. In the context of both association and sib pair linkage studies of quantitative traits, one of the most widely discussed techniques is the selective sampling of phenotypically extreme individuals. Several papers have demonstrated that such extreme sampling can markedly increase power (under certain circumstances). However, the parenthetical phrase in the previous sentence has generally not been made explicit and it appears to be implied that the more phenotypically extreme the individuals, the more power one has. In this paper, we show by simulation that this is not true under all circumstances. In particular, we show that under oligogenic models, where some biallelic quantitative trait loci (QTLs) have markedly asymmetric allele frequencies and large mean displacement among genotypes, and others have less asymmetric allele frequencies and smaller mean displacement among genotypes, power to detect linkage to or association with the latter QTL can actually decrease by sampling more extreme sib pairs. This suggests that more extreme sampling is not always better. The 'optimal' sampling scheme may depend on both what one suspects the underlying genetic architecture to be and which of the oligogenic QTL one has greatest interest in detecting.

Original languageEnglish (US)
Pages (from-to)97-107
Number of pages11
JournalHuman Heredity
Volume48
Issue number2
DOIs
StatePublished - Mar 1998
Externally publishedYes

Fingerprint

Multifactorial Inheritance
Chromosome Mapping
Quantitative Trait Loci
Gene Frequency
Genotype
Research Personnel
Costs and Cost Analysis
Genes

Keywords

  • Association studies
  • Extreme sampling
  • Linkage
  • Oligogenic traits
  • Power
  • Quantitative trait loci
  • Quantitative traits
  • Selective sampling

ASJC Scopus subject areas

  • Genetics(clinical)

Cite this

Extreme selection strategies in gene mapping studies of oligogenic quantitative traits do not always increase power. / Allison, David B.; Heo, Moonseong; Schork, Nicholas J.; Wong, Su Ling; Elston, Robert C.

In: Human Heredity, Vol. 48, No. 2, 03.1998, p. 97-107.

Research output: Contribution to journalArticle

Allison, David B. ; Heo, Moonseong ; Schork, Nicholas J. ; Wong, Su Ling ; Elston, Robert C. / Extreme selection strategies in gene mapping studies of oligogenic quantitative traits do not always increase power. In: Human Heredity. 1998 ; Vol. 48, No. 2. pp. 97-107.
@article{9a367418a56b43a59376a27bfae50bc7,
title = "Extreme selection strategies in gene mapping studies of oligogenic quantitative traits do not always increase power",
abstract = "It is well known that obtaining adequate statistical power to detect linkage to or association with genes for complex quantitative traits can be very difficult. In response, investigators have developed a number of power-enhancing strategies that consider restraints such as genotyping (and/or phenotyping) costs. In the context of both association and sib pair linkage studies of quantitative traits, one of the most widely discussed techniques is the selective sampling of phenotypically extreme individuals. Several papers have demonstrated that such extreme sampling can markedly increase power (under certain circumstances). However, the parenthetical phrase in the previous sentence has generally not been made explicit and it appears to be implied that the more phenotypically extreme the individuals, the more power one has. In this paper, we show by simulation that this is not true under all circumstances. In particular, we show that under oligogenic models, where some biallelic quantitative trait loci (QTLs) have markedly asymmetric allele frequencies and large mean displacement among genotypes, and others have less asymmetric allele frequencies and smaller mean displacement among genotypes, power to detect linkage to or association with the latter QTL can actually decrease by sampling more extreme sib pairs. This suggests that more extreme sampling is not always better. The 'optimal' sampling scheme may depend on both what one suspects the underlying genetic architecture to be and which of the oligogenic QTL one has greatest interest in detecting.",
keywords = "Association studies, Extreme sampling, Linkage, Oligogenic traits, Power, Quantitative trait loci, Quantitative traits, Selective sampling",
author = "Allison, {David B.} and Moonseong Heo and Schork, {Nicholas J.} and Wong, {Su Ling} and Elston, {Robert C.}",
year = "1998",
month = "3",
doi = "10.1159/000022788",
language = "English (US)",
volume = "48",
pages = "97--107",
journal = "Human Heredity",
issn = "0001-5652",
publisher = "S. Karger AG",
number = "2",

}

TY - JOUR

T1 - Extreme selection strategies in gene mapping studies of oligogenic quantitative traits do not always increase power

AU - Allison, David B.

AU - Heo, Moonseong

AU - Schork, Nicholas J.

AU - Wong, Su Ling

AU - Elston, Robert C.

PY - 1998/3

Y1 - 1998/3

N2 - It is well known that obtaining adequate statistical power to detect linkage to or association with genes for complex quantitative traits can be very difficult. In response, investigators have developed a number of power-enhancing strategies that consider restraints such as genotyping (and/or phenotyping) costs. In the context of both association and sib pair linkage studies of quantitative traits, one of the most widely discussed techniques is the selective sampling of phenotypically extreme individuals. Several papers have demonstrated that such extreme sampling can markedly increase power (under certain circumstances). However, the parenthetical phrase in the previous sentence has generally not been made explicit and it appears to be implied that the more phenotypically extreme the individuals, the more power one has. In this paper, we show by simulation that this is not true under all circumstances. In particular, we show that under oligogenic models, where some biallelic quantitative trait loci (QTLs) have markedly asymmetric allele frequencies and large mean displacement among genotypes, and others have less asymmetric allele frequencies and smaller mean displacement among genotypes, power to detect linkage to or association with the latter QTL can actually decrease by sampling more extreme sib pairs. This suggests that more extreme sampling is not always better. The 'optimal' sampling scheme may depend on both what one suspects the underlying genetic architecture to be and which of the oligogenic QTL one has greatest interest in detecting.

AB - It is well known that obtaining adequate statistical power to detect linkage to or association with genes for complex quantitative traits can be very difficult. In response, investigators have developed a number of power-enhancing strategies that consider restraints such as genotyping (and/or phenotyping) costs. In the context of both association and sib pair linkage studies of quantitative traits, one of the most widely discussed techniques is the selective sampling of phenotypically extreme individuals. Several papers have demonstrated that such extreme sampling can markedly increase power (under certain circumstances). However, the parenthetical phrase in the previous sentence has generally not been made explicit and it appears to be implied that the more phenotypically extreme the individuals, the more power one has. In this paper, we show by simulation that this is not true under all circumstances. In particular, we show that under oligogenic models, where some biallelic quantitative trait loci (QTLs) have markedly asymmetric allele frequencies and large mean displacement among genotypes, and others have less asymmetric allele frequencies and smaller mean displacement among genotypes, power to detect linkage to or association with the latter QTL can actually decrease by sampling more extreme sib pairs. This suggests that more extreme sampling is not always better. The 'optimal' sampling scheme may depend on both what one suspects the underlying genetic architecture to be and which of the oligogenic QTL one has greatest interest in detecting.

KW - Association studies

KW - Extreme sampling

KW - Linkage

KW - Oligogenic traits

KW - Power

KW - Quantitative trait loci

KW - Quantitative traits

KW - Selective sampling

UR - http://www.scopus.com/inward/record.url?scp=0031909410&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031909410&partnerID=8YFLogxK

U2 - 10.1159/000022788

DO - 10.1159/000022788

M3 - Article

VL - 48

SP - 97

EP - 107

JO - Human Heredity

JF - Human Heredity

SN - 0001-5652

IS - 2

ER -