Transcript Trait Mapping - Nematode bioinformatics. Analysis tools and data

Trait Mapping
•Recombination Mapping
•SNP mapping
BIO520 Bioinformatics
Jim Lund
Why do we care about variations?
underlie phenotypic
differences
cause inherited
diseases
allow tracking human
history (ancient and
modern)
Traits
• Mendelian
– single locus, few alleles
– high penetrance, high expressivity
– eg color, enzyme, molecular, genetic
diseases (CF, hemophilia…)
• Quantitative
–
–
–
–
multiple allele, multilocus
variable penetrance, expressivity
epistasis, environmental effects
eg. blood pressure, weight, IQ...
Traits
How do we find their basis?
• Association of variance in trait with
variance in gene
• Genetic linkage
Basic Concepts
Parent 2
Parent 1
A
B
a
b
A
X
B
a
b
A B
a b
A B
a b
OR
a b
A B
A B
a b
High LD -> No Recombination
(r2 = 1) SNP1 “tags” SNP2
A b
A B
a B
A B
A b
a B
A B
A b
etc…
Low LD -> Recombination
Many possibilities
Mapping Issues
• Need many arbitrary, polymorphic
markers for dense map
– Molecular markers: RFLP, STS, SNP
• Need many progeny
– 100 progeny for 1 cM map
– 1000/0.1 cM map, 100 kb in mouse
• Map distance varies (the ratio of
kb/cM not constant)
– centromere suppression
– inversion suppression
Genetic crosses
• Model organisms, e.g. Fungi, no problem
• Humans
– rare woman who will bear >5, >10 children
– controlled breeding problematic
Alternate Mapping
• Pedigree analyses
– likelihood estimation
– The original method, now less
common
• Population-based mapping
– association studies
– linkage disequilibrium
Pedigree Analysis
• Likelihood Method (LOD scores)
• LOD  3-4, 1/1000 – 1/10000 odds
of linkage
– genome-wide p-value of p < .05
• Hard to extend to <1 cM
Cloning Human Genes
•
•
•
•
Positional
Positional/Candidate
Candidate Only
Functional
Complex diseases
Association mapping
• Disease gene: D, d
• Marker: M, m
M associated with D if
the probability of an individual having the disease given that they
have allele M is much greater than the chance of having the
disease if the individual has allele m. Written as: P(D|M) > P(D|m)
Linkage between the gene and marker increases the likelihood of
D
M1
M2
M3
M4
M5
M6
association.
Association can be caused by
–
–
–
–
Causation
Population subdivision
Statistical artifact
Linkage disequilibrium
Association Mapping
•Pedigree sampled
•Many Meiosis (>104)
•Limited by number of markers
M
r
D
2N generations
•Resolution: 10-5 Morgans (Kbases)
Gene Mapping & the
single mutation case
D
M
At time t
D
Now
M
Complicating factors
Major Disease Causing Mutation.
Minor Disease
Causing Mutation
+
+ has the disease.
+
+
Non-genetic cause
+
+
+
Incomplete penetrance
Oversampled
Alzheimers & Apolipoproteins E
Definition of QTL?
A quantitative trait locus (QTL) is the location of
individual or multiple loci that affects a trait that is
measured on a quantitative (linear) scale. Examples of
quantitative traits are blood pressure and grain yield
(measured on a balance). These traits are typically
affected by more than one gene, and also by the
environment. Thus, mapping QTL is not as simple as
mapping a single gene that affects a qualitative trait
(such as an inborn error of metabolism).
http://gnome.agrenv.mcgill.ca/tinker/pgiv/whatis.htm
QTLs-interesting traits
• Heritability often ~0.5
• Traits like:
– Heart disease
– Depression
– Type II diabetes
– High blood pressure
– Arthritis
– Most diseases!
QTLs-simple problems
• 30,000 markers
– P-value=0.01
– 299 false hits, 1 real one
– Correct for multiple testing
• 2 QTLS near one another
– “ghost” QTL between them
Factors that lead to success in
mapping QTLs
• Simple, easily quantified trait
• Genes of major effect
– distinct chromosomal loci
• Well-defined map
• Large numbers of progeny
– inbred
– outbred
Significance Thresholds by Permutation
Churchill and Doerge, 1994
1.Permute the data
(create the null hypothesis)
H0: there is no QTL in the tested interval
H1: there is QTL in the tested interval
2.Perform interval mapping
3. Repeat (1) and (2) many times
4.Choose Threshold
Human SNPs
• About 10 million SNPs exist in human populations
where the rarer SNP allele has a frequency of at
least 1%.
• A set of associated SNP alleles in a region of a
chromosome is called a "haplotype".
• SNPs are arranged in groups
– SNPs within groups show little recombination
– Nonrandom association of SNPs results in only a few
common haplotypes
– Patterns capture most of the variation in a region
• The HapMap will describe the common patterns of
genetic variation in humans.
• The HapMap Project will identify the associations
between SNPs and identify the SNPs that tag them
(tagSNPs).
SNPs identification methods
• Pairwise sequence comparison
• Deep resequencing
• High throughput mismatch detection
methods
– Denaturing high-performance liquid
chromatography (DHPLC)
– Single-strand Conformational
Polymorphism (SSCP)
HapMap
• Blocks of adjacent SNPs that show little
recombination are called haplotype blocks.
• Mean haplotype block length is tens of kb.
• HapMap project started examining 270
individuals from 4 ethnic groups.
• Now expanding to a more comprehensive
sample.
Characterization of haplotype blocks means
that fewer SNPs will need to be typed.
500,000 SNPs will identify 90% of
haplotype blocks.
HapMap Glossary
• LD (linkage disequilibrium): For a pair of SNP
alleles, it’s a measure of deviation from random
association (i.e., a measure of lack of
recombination). Measured by D’, r2, LOD
• Phased haplotypes: Estimated distribution of SNP
alleles. Alleles transmitted from Mom are in same
chromosome haplotype, while Dad’s form the
paternal haplotype.
• Tag SNPs: Minimum SNP set to identify a
haplotype. r2= 1 indicates two SNPs are
redundant, so each one perfectly “tags” the other.
HapMap Project
Phase 1
Phase 2
Phase 3
Samples
& POP
panels
269
samples
(4
panels)
270
samples
(4
panels)
1,115
samples
(11
panels)
Genotyp
ing
centers
HapMap
Internati
onal
Consorti
um
Perlege
n
Broad &
Sanger
Unique
QC+
1.1 M
3.8 M
(phase
1.6 M
(Affy 6.0
Phase 3 Samples
label
ASW*
CEU*
CHB
CHD
GIH
JPT
LWK
MEX*
MKK *
TSI
YRI*
population sample
African ancestry in Southwest USA
Utah residents with Northern and Western
European ancestry from the CEPH collection
Han Chinese in Beijing, China
Chinese in Metropolitan Denver, Colorado
Gujarati Indians in Houston, Texas
Japanese in Tokyo, Japan
Luhya in Webuye, Kenya
Mexican ancestry in Los Angeles, California
Maasai in Kinyawa, Kenya
Toscans in Italy
Yoruba in Ibadan, Nigeria
* Population is made of family trios
# samples
90
QC+ Draft 1
71
180
162
90
100
100
91
100
90
180
100
180
1,301
82
70
83
82
83
71
171
77
163
1,115
SNP databases
• dbSNP (NCBI)
– 12 million human SNPs
– 5 million validated SNPs
– http://www.ncbi.nlm.nih.gov/SNP/get_html.cgi?whichHtml=overview
• SNP frequency information
• Mapped to the current genome build
• HapMap (haplotypes)
How to use markers to find disease?
genome-wide, dense SNP marker map
• problem: genotyping cost precludes using millions of
markers simultaneously for an association study
• question: how to select from all available markers a
subset that captures most mapping information (marker
selection, marker prioritization)
• depends on the patterns of allelic association
(haplotypes) in the human genome
The promise for medical genetics
CACTACCGA
CACGACTAT
TTGGCGTAT
• within blocks a small number of SNPs
are sufficient to distinguish the few
common haplotypes  significant
marker reduction is possible
chromosome
blocks
• if the block structure is a general feature of human
variation structure, whole-genome association studies
will be possible at a reduced genotyping cost
• this motivated the HapMap project
Gibbs et al.
Nature 2003
The promise for medical genetics
•Discover genes contributing to complex
diseases
•Use these markers to test for inherited
disease risk
• Find SNPs associated with drug side effects
•Make drugs safer.
•Rescue drugs abandoned due to
significant side effects.
Pathway of Drug Development
• Lead or Target (Clinical
Candidate)
• Animal Model Testing
– Toxicity, Efficacy
• Phase I Pre-Clinical
(toxicity)
• Phase II (efficacy)
• Phase III (efficacy)
• NDA (new drug
application)
• $100M
2000
• $0.5M
100
• $0.5M
• $5M
• $50M
20
3
2
1
Why pharmacogenomics?
• Where do you find the next profitable
drug?
– The 19/20 drugs that failed AFTER phase 1,
but are still efficacious!
• How do you decrease the cost of clinical
trials?
– Don’t enroll people of the “wrong” genotype!
• Only give drugs to patients likely to
benefit and at a low genetic risk of side
effects!