Behavioral GeneticsTopic #9

Gene Identification

Why Search for Behavioral Genes?

• Gene identification would confirm genetic effects

• Identify the physiological basis of behavior; identify interventions

• Classification of behavior and behavioral disorders

• Genotype-environment interaction

“Human genetics is now at a critical juncture. The molecular methods used to identify the gens underlying rare mendelian syndromes are failing to find the numerous genes causing more common non-mendelian diseases.”

Neil J. Risch (2000, Nature, 405:200)

Search Strategies (Where?)

• Non-targeted– genome-wide

• Targeted:– Chromosomal anomaly (e.g., partial trisomy 5 and SZ,

VCFS)– Biological-based hypotheses (e.g., DRD4 and Novelty-

seeking)– Animal models– Positive linkage result from a non-targeted search– Micro-array findings

Search Methods (How?)• Linkage analysis (w/i family association) Advantages:

– Feasible to implement genome-wide (systematic & comprehensive); don’t need to have hypotheses about location or mechanism

– relative to alternatives, low false positive rate

Disadvantages– Limited power: detect genes accounting for 5% or more

of the phenotypic variance– Limited resolution: tight linkage can be millions of

bases away– Never finds the gene, rather at best identifies a region

Mendel vs. Galton, Redux?

• Mendelian– Single gene

– Rare

– Strong phenotype-genotype correlation

• Galtonian– Multiple genes

– Common (?)

– Weak phenotype-genotype correlation

Positional cloning strategy has been very successful in mapping genes that are rare and have large phenotypiceffect, even if for common disease (BRCA-1, BRAC-2, APP, MODY-1,-2,-3)

Search Methods; cont.• Allelic Association: association between

allele status and phenotype in unrelated individuals. Population-level association

Stomach Cancer

- +

O

Not O(A,B,AB)

40

40

60

60

Allelic Association

• Advantages– In principle, very high statistical power– In principle, can identify causal agent

• Disadvantages– Need functional polymorphisms in a candidate

gene– Concern about false-positives due to mis-matching

cases and controls

Ethnic Group #1:- Hyper + Hyper

O 24

(.40)

16

(.40)

40

(.40)

Not O

36 24 60

60 40 100

- Hyper

+

Hyper

O 9

(.10)

1

(.10)

10

(.10)

Not O

81 9 90

90 10 100

Ethnic Group #2:

- Hyper + Hyper

O 33

(.22)

17

(.34)

50

(.25)

Not O

117 33 150

150 50 100

Combined:

The Confounding Role of Ethnicity

• In candidate gene studies, if– The phenotype (disorder) varies as a function of

ethnicity– The genetic polymorphism varies in frequency

as a function of ethnicity

• Then,– An artifactual association between genetic

polymorphism and disorder can be observed

Ebstein et al. (1996)

• Phenotype: Personality trait of novelty-seeking

• Sample: 124 young Israelis

• Genotype: Variable 48 base repeat sequence in Dopamine D4 Receptor gene (DRD4)

• Finding: 7 repeat allele associated with higher novelty seeking

DRD4

Results

Association Studies of DRD4and Novelty Seeking

Study

Trait

N

Effect Size

Ebstein et al. (1996) Benjamin et al. (1996)

Novelty Seeking Novelty Seeking

124 315

.50*

.39*

Malhotra et al. (1996) Ebstein et al. (1997)

Novelty Seeking Novelty Seeking

193 94

-.11 .23

Jonsson et al. (1997) Vandenbergh et al. (1997)

Impulsivity Novelty Seeking

126 200

.28 not sig

Gebhardt et al. (1997) Pogue-Giele et al. (1998)

Novelty Seeking Novelty Seeking

58 281

-.14 .04

Meta-Analysis

Association Studies of DRD4 and Substance Abuse

N Phenotype Assoc.?

George et al. (1993) 72 Alcohol Yes

Adamson et al. (1995) 226 Alcohol No

Chang et al. (1997) 127 Alcohol No

Li et al. (1997) 275 Heroin Yes

Geijer et al. (1997) 139 Alcohol No

Kotler et al. (1997) 251 Heroin Yes

Why the Inconsistent Results?• False positive due to poor matching

• Low power in samples of modest size

• DRD4 effect depends on genetic background (epistasis):– Experimental studies in mouse

– APOE and AD

• False positive due to low a priori likelihood

Blum et al. (1990)

• Phenotype: Alcoholism• Genotype: A1/A2 allele at DRD2• Sample: 35 alcoholics and 35 non-alcoholics

• Finding:NotAlc Alc

A1

A2

7(20%)

28

24(69%)

11

DRD2

If real, how could the association arise?

• Linkage Disequilibrium:– Non-random association of alleles at linked loci– Can lead to population associations between

non-functional genetic markers and phenotypes

s s s SO O O A

Non-Sz, O-blood type Mother Sz, A-blood type Dad

Suppose: 1. This is the original mutation 2. Distance is = .05 3. Everyone has 4 children

S

A

A A A A

I - 100%

II - 100%

O

III - 94%

IV - 89%

V - 85%

Linkage Disequilibrium Mapping

• Linkage disequilibrium will be a function of: – How tightly linked the two loci are– The number of generations since introduction

of the original mutational event– Whether the mutation has been introduced

more than once

• In short, it depends on the evolutionary history of the population

Linkage Disequilibrium

Challenges of Gene Identification

“of the 166 putative associations which have been studied three or more times, only 6 have been consistently replicated.”

•Hirschhorn et al. (2002). “A comprehensive review of genetic association studies. Genetics in Medicine 4:45-61.

“detection of linkage and positional cloning of specific disease-susceptibility loci remains elusive.”

-- Altmüller et al. (2001). AJHG, 69: 936-950

Common Disease-Common Variant (CDCV) Hypothesis

• The heritable basis of common, complex disease owe primarily to alleles that are:– Relatively common (i.e., not rare, e.g., > 10%)– Experience little selective pressure (i.e., only

disadvantageous when combined with other mutations)

– Ancient (i.e., introduced more than 5000 Gs or 100,000 yrs ago)

Therefore

• In outbred populations (e.g., the US), LD may not extend much beyond 3 kb a genome-wide LD study would require 500,000 markers !

• Founder or isolated populations may need fewer:– Small # of founders, little immigration, expansion

– Samii, Costa Rica, Quebec, Iceland, Japan

Are there other explanations for the DRD2-Alcoholism Association?

• Linkage Disequilibrium

• False positive owing to ethnic mismatching

DRD2 A1 allele frequencies

Is it really going to take genome-wide studies with 500,000 markers to map

complex diseases?• Validity of CDCV hypothesis

• Genome-wide association study of ‘functional’ SNPs (~ 60,000-100,000)

• HapMap project

Haplotypes• Allelic constitution of multiple loci on a

single chromosome

• Recombination is not random, but rather there are recombination “hot spots”

• This gives rise to blocks of DNA (haplotypes) where there is very little recombination w/i blocks but strong recombination between blocks

Cardon & Abecasis (2003). Using haplotype blocks to map the human genome.Trends in Genetics, 19: 135-140.

Although 9 markers in block 4, only 4 possible haplotypes, which

can be determined by only 3 markers

   Linkage

 Allelic Association

 Linkage

Disequilibrium

 Type of Association

 w/i Family between

marker and phenotype

 Pop assoc btw functl poly and phenotype

 

 Pop assoc btw marker and phenotype

 Interpretation

 Region implicated

 Gene or nearby gene

implicated

 Region implicated

 Genome-wide

 Yes, 300-400

markers

 Not now but maybe

in the future(~100,000)

 Not now but maybe

in future (~500,000)

 Resolution

 Typically w/i 8-10

centimorgans recombination

 Not relevant

 Typically (much) less than 1 centimorgan

 Preferred Sample

 Large pedigrees

Sib pairs

 Unrelated individuals

 Genetic isolates

if rare

 Sensitivity

 Relatively low

 Relatively High

 Relatively High

 Specificity

 Relatively High

 Relatively Low

 Relatively High (?)