Transcript Slide 1

Genetic Variations
Lakshmi K Matukumalli
Human – Mouse Comparison
Structural Variations
Ploidy
(Down’s Syndrome)
Inversions
Translocations
Segmental duplications
Molecular Variations
Single nucleotide polymorphisms
Short Indels
Simple sequence repeats
Copy number variants
Loss of heterozygosity
Microsatellite
(2-9 bp core repeat)
Minisatellite
(10-60 bp core repeat)
Copy number variants
Type of polymorphisms
Insertion/deletion
polymorphism (indel)
Single-nucleotide
Polymorphism (SNP)
T
C
Nonsynonymous
polymorphism
Synonymous
polymorphism
GAG Asp
GUG Val
GAU Asp
GAC Asp
5’
Untranslated
region
ATG
5’ Flanking
region
Promoter
TAACGG
TA GG
3’
Untranslated
region
End
3’ Flanking
region
Coding
Intron
Transcript
Coding
Choosing the Technology
Extent of Variation
(Human Genome)
> 5 million SNPs (dbSNP)
Recent genome analysis of diploid individual showed 4.1 million DNA variants,
encompassing 12.3 Mb.
- 3,213,401 single nucleotide polymorphisms (SNPs),
- 53,823 block substitutions (2–206 bp),
- 292,102 heterozygous insertion/deletion events (indels)(1–571 bp),
- 559,473 homozygous indels (1–82,711 bp),
- 90 inversions,
- Plus segmental duplications and copy number variations.
Non-SNP DNA variation accounts for 22% of all events, however they involve
74% of all variant bases. This suggests an important role for non-SNP genetic
alterations in defining the diploid genome structure.
Moreover, 44% of genes were heterozygous for one or more variants.
Importance of SNPs and other variants
 Study Genetic variation in diverse populations in any
species to





understand evolutionary origins and history,
estimate population size,
breeding structure, or life-history characters
Migration within and between sub-populations
Understand evolutionary basis for maintenance of
genetic variation and speciation.
 Applications
 Genetic association of traits
 Effects on gene expression (e.g., synonymous vs
nonsynonymous / TF binding sites)
 DNA finger printing or sample tracking
Fine Mapping with SNP Markers
Advantages of SNPs as genetic markers
as compared to microsatellites.
•High abundance
•Distribution throughout the genome
•Ease of genotyping
•Improved accuracy
•Availability of high throughput
multiplex genotyping platforms
SNP Discovery - Sanger sequencing (EST)
SNP Discovery - Diploids (heterozygous loci)
SNP-PHAGE (Software package)
SNP Pipeline for Haplotype Analysis and GEnbank
(dbSNP) submissions.
Important steps are








Primer development
Primer testing
Sequencing
Base calling,
Sequence assembly
Polymorphisms analysis
Haplotype analysis
GenBank submission of
confirmed polymorphisms
Primers
5’ amplicons
3’ amplicons
Sequence Variation
Application of Machine Learning
in SNP Discovery
Objective: Reduce human intervention by using expert annotated dataset
for training a Machine learning (ML) program and use it to differentiate
good/bad polymorphisms
Steps:
•Parameter Selection
•Parameter Optimization
•Testing
•Implementation.
Results:
Achieved substantial
improvement in the
accuracies as compared
to using only polybayes or
polyphred.
Inputs
Outputs
Machine
Learning
Program
Planning
Reasoning
Model
(Tree / Rules)
Training mode
Inputs
and
Model
(Tree /Rules)
Outputs
Testing/Prediction mode
SNP Discovery using next
generation sequencers
 Short sequences 23-35 bp long at a fraction of cost.
 Reduced Representation Sequencing
 Digest genomic DNA with restriction enzyme
Cost / Mb
Screen based on in silico digestion
ABI $880
 Size select based on
454 $160
 Repetitive DNA
 Number of fragments
Solexa $5
 Sequencing platform
 Allows “targeted” deep sequencing of pools of DNA
 Randomly distributed

SNP Discovery - Bioinformatics
 Strategies to maximize performance
 High quality score stringencies
 For each read
 At base for putative SNP
 Require single map location of a 23-bp “tag” (and 4-bp
restriction site)
 Allow only one single base pair difference match for a
putative SNP
 Reduces repeat content
 Reduces gene family/paralog false positives
 Require 2 copies of each allele – assembly can count
as 1
Predicted & Observed Minor Allele Frequency
Population Genetics
Population genetics is the study of the allele
frequency distribution and change under the
influence of the four evolutionary forces: natural
selection, genetic drift, mutation and gene flow. It
attempts to explain phenomena as adaptation and
speciation.
(www.wikipedia.org)
X
Variation
Population Genetics
Neutral theory : Rate at which new genetic
variants are formed is equal to the loss of
genetic diversity due to drift.
Genotypes : CT, CC, TT
C/T C/C T/T
Alleles : C and T
Genotyping of a population of 1000 individuals for a SNP resulted in
100, 500 and 400 genotypes for CC, CT and TT respectively
Genotype Frequencies: CC (0.1), CT (0.5) and TT(0.4)
Allele Frequencies: C (p) = (200+500)/2000 = 0.35 (minor allele -- MAF)
T (q) = (500+800)/2000 = 0.65 (major allele)
Hardy-Weinberg Equilibrium:
Expected genotype frequencies are p2, 2pq and q2
(122, 422 and 455)
HWE Deviations: Drift, Selection, Admixture etc.,
Fst
Useful to partition genetic variation into components:
within populations
between populations
among populations
Sewall Wright’s Fixation index (Fst is a useful index of
genetic differentiation and comparison of overall
effect of population substructure.
Measures reduction in heterozygosity (H)
expected with non-random mating at any one
level of population hierarchy relative to another
more inclusive hierarchical level.
Fst = (HTotal - Hsubpop)/HTotal
Fst ranges between minimum of 0 and maximum
of 1:
=0
 no genetic
differentiation
<< 0.5
 little genetic differentiation
>> 0.5
 moderate to great genetic
differentiation
= 1.0
 populations fixed
for different alleles
Genotype – Phenotype Association
(Significance of Haplotypes)
Haplotype inference
 The solution to the haplotype phasing problem is not
straightforward due to resolution ambiguity
 Computational and statistical algorithms for
addressing ambiguity in Haplotype Phasing:
1) parsimony
2) phylogeny
3) maximum-likelihood
4) Bayesian inference
Linkage disequilibrium (LD)
 Non-random association of alleles at two or more loci,
not necessary in the same chromosome.
 LD is generally caused by interactions between genes;
genetic linkage and the rate of recombination; random
drift or non-random mating; and population structure.
Let A and B be two loci segregating two alleles each;
a1 and a2 with frequencies p1 and p2 in A,
and b1 and b2 with frequencies q1 and q2 in B.
B1
B2
Total
A
A1
p11 = p1 q1 + D
p12 = p1 q2 - D
p1
B
A2
p21 = p2 q1 - D
p22 = p2 q2 + D
p2
q2
1
Total
q1
Linkage disequilibrium (cont)
D = p11 - p1q1
D depends on the allele frequencies at A and B.
D’ a scaled version of D:
D’ =
D
min(p1q1 , p2q2)
If D < 0
D
min(p1q2 , p2q1)
If D > 0
Linkage disequilibrium (cont)
 Squared correlation coefficient
r2 =
D2
p 1p 2q 1q 2
* The measure preferred by population geneticists
* Is independent of of allele frequencies
* Ranges between 0 and 1
* r2 = 1 implies the markers provide exactly the same
information
* r2 = 0 when they are in perfect equilibrium
2.4 Linkage disequilibrium (cont)
 Visualizing LD
2.4 Linkage disequilibrium (cont)
 Visualizing LD