Transcript ppt

Lecture 27 : Asscociation Genetics
April 21, 2014
Announcements
Final exam April 29 at 3 pm, 3306 LSB (computer lab)
Review session on Friday
Bring questions
Final lab on Wednesday
Course evaluations
Extra credit opportunity: earn up to 10 points for
lab report
Due at final exam
Last Time
Sequence data and quantification of variation
Sequence-based tests of neutrality
 Ewens-Watterson Test
 Tajima’s D
 Hudson-Kreitman-Aguade Test
 Synonymous versus Nonsynonymous substitutions
 McDonald-Kreitman
Today

Quantitative traits
Genetic basis
Heritability

Linking phenotype to genotype
QTL analysis introduction
Limitations of QTL
Association genetics
Mendelian trait
Individual
1
2
3
4
5
6
7
8
9 10
Allele A1
Allele A2
Genotype =
12 11 22 22 11 22 12 11 22 12
Quantitative trait
16
28
40
52
Height
64
76
88
Courtesy of Glenn Howe
Quantitative traits are polygenic
55
Students at
Connecticut
Agricultural
College, 1914
60
65
70
75
As the number of loci
controlling a trait
increases, the
distribution of trait
values in a population
becomes bell-shaped
80
85
Influence of Environment on Human Height
Mean = 67  2.7 in.
1914
Height vs GDP (1925-1949)
By Country
Mean = 70  3 in.
1996
Baten 2006
4:10
Schilling et al. 2002. Amer. Stat. 56: 223-229
6:5
Hartl and Clark 2007
Hartl, D. 1987. A primer of Population Genetics.

3 loci, 2 additive
alleles

Uppercase alleles
contribute 1 unit to
phenotype (e.g.,
shade of color)
The phenotype is the outward manifestation
of the genotype
=
+
Phenotype
Genotype
σ2 P
σ2G
Environment
σ2E
Courtesy of Glenn Howe
Types of genetic variance (σ2G)
 Additive
(σ2A): effects of individual alleles
 Dominance
locus
 Interaction
(epistasis)
(σ2D): effects of allele interactions within
(σ2I): effects of interactions among loci
σ2G = σ2A + σ2D + σ2I
Non-additive
Main cause for resemblance
between relatives
Heritability

Phenotype vs Genotype
Var(phenotype) = Var(genotype) + Var(environment)

Heritability:
Var(genotype) / Var(phenotype)

Two types of heritability
 Broad-Sense Heritability includes all genetic effects: dominance,
epistasis, and additivity
− For example, the degree to which clones or monozygotic twins
have the same phenotype
 Narrow-Sense Heritability includes only additive effects
− For example, degree to which offspring resemble their parents
Heritability (continued)

Characteristic of a trait measured in a particular population in a
particular environment

Best estimated in experiments (controlled environments)

Estimated from resemblance between relatives

The higher the heritability, the better the prediction of genotype
from phenotype (and vice versa)
h² = 0.5
h² = 0.1
P
h² = 0.9
P
G
P
G
http://psych.colorado.edu/~carey/hgss/hgssapplets/heritability/heritability1/heritability1.html
G
Identifying Genes Underlying Quantitative
Traits
 Many
individual loci are responsible for quantitative
traits, even those with high heritability
 Identification
of these loci is a major goal of breeding
programs
 Allows
mechanistic understanding of adaptive
variation
 Methods
usually rely on correlations between
molecular marker polymorphisms and phenotypes
Quantitative Trait Locus Mapping
A
B
C
A
B
C
Parent 2
a
b
c
X
HEIGHT
Parent 1
a
b
c









A
B
C
A
B
c
B
b
a
b
c
Bb
A
B
C
X
A A
b B
c c
a a
BB
c c
BB
F1
F1
BB
a A
B b
c c
bb
modified from D. Neale
A a
b b
c c
bb
AA
bB
cC
BB
bb
a
b
c
A A
B b
c c
Bb
a a
B B
c c
Bb
GENOTYPE
A a
b B
c c
Bb
a
B
c
BB
BB
Quantitative Trait Locus Analysis
 Step 1: Make a controlled cross to create a large family
(or a collection of families)
 Parents should differ for phenotypes of interest
 Segregation of trait in the progeny
 Step 2: Create a genetic map
 Large number of markers phenotyped for all progeny
 Step 3: Measure phenotypes
 Need phenotypes with high heritability
Step 1: Construct Pedigree

Cross two individuals with
contrasting characteristics

Create population with segregating
traits

Ideally: inbred parents crossed to
produce F1s, which are intercrossed
to produce F2s

Recombinant Inbred Lines created
by repeated intercrossing
 Allows precise phenotyping,
isolation of allelic effects
Grisel 2000 Alchohol Research & Health 24:169
Step 2: Construct Genetic Map

Number of recombinations between
markers is a function of map distance

Gives overview of structure of entire
genome

Anonymous markers are cheap and
efficient: AFLP, Genotyping by
Sequencing

Codominant markers much more
informative: SSR, SNP

Genotyping by Sequencing gives best of
both worlds: cheap, abundant,
codominant markers!
Step 3: Determine Phenotypes of Offspring
0.1

Phenotype must be segregating in
pedigree

Must differentiate genotype and
environment effects
 How?
0.5

0.9
Works best with phenotypes with
high heritability
Step 4: Detect Associations between Markers and
 Single-marker associations are
Phenotypes
simplest
 Simple ANOVA, correcting for multiple
comparisons
 Log likelihood ratio: LOD (Log10 of
odds)

LOD = log10
Pr(Data | QTL)
Pr(Data | noQTL)
If QTL is between two markers,
situation more complex
 Recombination between QTL and
markers (genotype doesn't predict
phenotype)
 'Ghost' QTL due to adjacent QTL

Use interval mapping or composite
interval mapping
 Simultaneously consider pairs of loci
across the genome
Step 5: Identify underlying molecular mechanisms
QTL
chromosome
Genetic Marker
QTG: Quantitative Trait Gene
QTN: Quantitative Trait Nucleotide
Adapted from Richard Mott, Wellcome Trust Center for
Human Genetics
QTL Limitations
 Huge
regions of genome underly QTL, usually hundreds
of genes

How to distinguish among candidates?
 Biased

toward detection of large-effect loci
Need very large pedigrees to do this properly
 Limited
genetic base: QTL may only apply to the two
individuals in the cross!
 Genotype
x Environment interactions rampant: some
QTL only appear in certain environments
Linkage Disequilibrium and Quantitative Trait Mapping


Linkage and quantitative trait locus (QTL)
analysis

Need a pedigree and moderate number of
molecular markers

Very large regions of chromosomes
represented by markers
Association Studies with Natural Populations
 No pedigree required
 Need large numbers of genetic markers
 Small chromosomal segments can be localized
 Many more markers are required than in
traditional QTL analysis
Cardon and Bell 2001, Nat. Rev. Genet.
2: 91-99
ancestral
chromosomes
G
T
HEIGHT
Association Mapping






*



TT
TC
GENOTYPE
CC
recombination
through
evolutionary
history
present-day
chromosomes
in natural
population
G
C
G
T
A
C
A
C
*
G
T
A
T
*
*
Slide courtesy of Dave Neale
Next-Generation Sequencing and Whole Genome
Scans

The $1000 genome is on the horizon
 Current cost with Illumina HiSeq
2000 is about $2000 for 10X depth




Thousands of human genomes have
now been sequenced at low depth
Can detect most polymorphisms with
frequency >0.01
True whole genome association
studies now possible at a very large
scale
Direct to Consumer Genomics: 23 &
Me and other genotyping services
http://www.1000genomes.org/
Commercial Services for Human Genome-Wide SNP
Characterization
NATURE|Vol 437|27 October 2005

Assay 1.2 million “tag SNPs” scattered across genome using Illumina
BeadArray technology

Ancestry analyses and disease/behavioral susceptibility
Identifying genetic mechanisms of simple vs. complex
diseases

Simple (Mendelian) diseases: Caused by a single major gene
 High heritability; often can be recognized in pedigrees
 Example: Huntington’s, Achondroplasia, Cystic fibrosis, Sickle Cell Anemia
 Tools: Linkage analysis, positional cloning

 Over 2900 disease-causing genes have been identified thus far: Human Gene Mutation
Database: www.hgmd.cf.ac.uk
Complex (non-Mendelian) diseases: Caused by the interaction between
environmental factors and multiple genes with minor effects
 Interactions between genes, Low heritability
 Example: Heart disease, Type II diabetes, Cancer, Asthma
 Tools: Association mapping, SNPs !!
 Over 35,000 SNP associations have been identified thus far:
http://www.snpedia.com
Slide adapted from Kermit Ritland
Complicating factor: Trait Heterogeneity
Same phenotype has multiple genetic mechanisms underlying it
Slide adapted from Kermit Ritland
Case-Control Example: Diabetes
 Knowler et al. (1988) collected data on 4920 Pima
and Papago Native American populations in
Southwestern United States
 High rate of Type II diabetes in these populations
 Found significant associations with Immunoglobin
G marker (Gm)
 Does this indicate underlying mechanisms of
disease?
Knowler et al. (1988) Am. J. Hum. Genet. 43: 520
Case-control test for association (case=diabetic, control=not diabetic)
Gm Haplotype
Type 2 Diabetes
present
absent
Total
present
8
29
37
absent
92
71
163
100
100
200
Total
Question: Is the Gm haplotype associated with risk of Type 2 diabetes???
(1) Test for an association
C21 = (ad - bc)2N
.
(a+c)(b+d)(a+b)(c+d)
= [(8x71)-(29x92)]2 (200)
= 14.62
(100)(100)(37)(163)
(2) Chi-square is significant. Therefore presence of GM haplotype
seems to confer reduced occurence of diabetes
Slide adapted from Kermit Ritland
Case-control test for association (continued)
Question: Is the Gm haplotype actually associated with risk of Type 2 diabetes???
The real story: Stratify by American Indian heritage
0 = little or no indian heritage;
8 = complete indian heritage
Index of indian
Heritage
0
4
8
Conclusion:
Gm
Haplotype
Percent
with
diabetes
Present
17.8
Absent
19.9
Present
28.3
Absent
28.8
Present
35.9
Absent
39.3
The Gm haplotype is NOT a risk factor for Type 2
diabetes, but is a marker of American Indian heritage
Slide adapted from Kermit Ritland
Population structure and spurious association

Assume populations are historically isolated

One has higher disease frequency by chance

Unlinked loci are differentiated between populations also

Unlinked loci show disease association when populations are lumped
together
Population with high
disease frequency
Gene flow barrier
Population with low
disease frequency
Alleles at neutral locus
Alleles causing
susceptibility to disease
Association Study Limitations

Population structure: differences between cases
and controls

Genetic heterogeneity underlying trait

Random error/false positives

Inadequate genome coverage

Poorly-estimated linkage disequilibrium
Association Analysis with a Mixed Model
effects of
background SNPs
phenotype
(response variable)
of individual i
effect of target SNP
Family effect
(Kinship
coefficient)
Population Effect (e.g., Admixture
coefficient from Structure or
values of Principal Components)
Implemented in the Tassel program (Wednesday in lab)