2. Natural Selection in Humans
Sharareh Noorbaloochi
CS 374
Oct 10, 2006

3. Papers to be presented…
Science, 16 June 2006, Volume 312
PLoS Biology, March 2006, Volume 4

4. Overview Pursuit of natural selection Biological Background
Methods for detecting positive selection
Genome-wide studies
From candidate to function
In intro: motivation and why we care….put screen shot of each paper as well

5. Images from: Voight et al. 2006
Migration out of africa:
Images from: Voight et al. 2006

6. Adaptability of Modern Humans
Humans have undergone tremendous cultural and environmental changes during the last ~40-50 KY.
Spread around the world (migrate out of Africa 100KY)
Global warming trend since last ice age ~14 KYA
Transition from hunter to agricultural society (<~10KYA)
Increase in pathogen load due to greater
population density and proximity to livestock
The evolution of modern human populations has been
accompanied by dramatic changes in environment and
lifestyle. In the last 100,000 years, behaviorally modern
humans have spread from Africa to colonize most of the
globe. In that time, humans have been forced to adapt to a
wide range of new habitats and climates. Following the end of
the last ice age, 14,000 years ago, there was a major warming
event that raised global temperatures to roughly their current
levels. Further dramatic changes occurred with the transition
from hunter-gatherer to agricultural societies, starting about
10,000–12,000 years ago in the Fertile Crescent, and a little
later elsewhere. This was also a period marked by rapid
increases in human population densities. Increased population
density promoted the spread of infectious diseases, as
did the new proximity of farmers to animal pathogens [1,2].
Voight et al. (2006)

7. Pursuit of Natural Selection
Each of these kinds of changes likely resulted in powerful
selective pressures for new genotypes that were better suited
to the novel environments. Indeed, there are a number of
recent reports of genes that show signals of very strong and
recent selection in favor of new alleles: for example, in
response to malaria [3–5]; at the lactase gene in response to
dairy farming [6]; at a salt sensitivity variant in response to
climate [7]; and in genes involved in brain development [8,9].
To date, the best examples of recent selection in humans
have all been discovered in studies of candidate genes where
there was a prior hypothesis of selection. Hence, very little is
known about how widespread such signals are; nor is there
unbiased information about what kinds of genes or biological
processes are most involved in the adaptation of modern
humans. It is unclear whether these genes are the same kinds
of genes that were most important in the earlier evolution of
the Homo lineage, as identified from comparisons with
chimpanzees [10–12]. It is also not known to what extent
recent selective events have been geographically restricted, as
opposed to taking place in all populations. A number of
recent studies have detected more signals of adaptation in
non-African populations than in Africans [13–17], and some
of those studies have conjectured that non-Africans might
have experienced greater pressures to adapt to new environments
than Africans have.
In this study, we use newly

8. Candidate Genes: Selection
G6PD [malaria resistance] Tishkoff et al 01, Sabeti et al 02
Lactase [dairy farming] Bersaglieri et al 04
MCPH1, ASPM [brain development] Evans et al 05, Mekel-Bobrov et al 05
Spend 0.1 s on this slide

9. Some Facts In human beings, 99.9 percent of bases are the same.
Remaining 0.1 percent makes a person unique.
Different attributes / characteristics / traits
how a person looks,
diseases he or she develops.
These variations can be:
Harmless (change in phenotype)
Harmful (diabetes, cancer, heart disease, Huntington's disease, and hemophilia )
Latent (variations found in coding and regulatory regions, are not harmful on their own, and the change in each gene only becomes apparent under certain conditions e.g. susceptibility to lung cancer)
The genetic sequences of different people are remarkably similar. When the chromosomes of two humans are compared, their DNA sequences can be identical for hundreds of bases. c (Figure 1). One person might have an A at that location, while another person has a G, or a person might have extra bases at a given location or a missing segment of DNA. Each distinct "spelling" of a chromosomal region is called an allele, and a collection of alleles in a person's chromosomes is known as a genotype.

10. Human Genetic Variations
Two types of genetic mutation events for today:
Single base mutation which substitutes one nucleotide for another
-- Single Nucleotide Polymorphisms (SNP)
Insertion or deletion of one or more nucleotide(s)
--Tandem Repeat Polymorphisms
--Insertion/Deletion Polymorphisms
Structural variations also important
(copy numbers)
One of the Most common type of genetic variation
Differences in individual bases are by far the most common type of genetic variation.
These genetic differences are known as single nucleotide polymorphisms, or SNPs (pronounced "snips"). By identifying most of the approximately 10 million SNPs estimated to occur commonly in the human genome, the International HapMap Project is identifying the basis for a large fraction of the genetic diversity in the human species.

11. What is SNP ?
A SNP is defined as a single base change in a DNA sequence that occurs in a significant proportion (more than 1 percent) of a large population.
For example a SNP might change the DNA sequence
AAGGCTAA  ATGGCTAA.

12. SNP facts SNPs are found in coding and (mostly) noncoding regions.
Occur with a very high frequency
about 1 in 1200 bases on average.
approximately 10 million SNPs occur commonly in the human genome.
One person might have an A at that location, while another person has a G, or a person might have extra bases at a given location or a missing segment of DNA. Each distinct "spelling" of a chromosomal region is called an allele, and a collection of alleles in a person's chromosomes is known as a genotype.
Say that a spelling change in a gene increases the risk of suffering from high blood pressure, but researchers do not know where in our chromosomes that gene is located. They could compare the SNPs in people who have high blood pressure with the SNPs of people who do not. If a particular SNP is more common among people with hypertension, that SNP could be used as a pointer to locate and identify the gene involved in the disease.

13. Allele
Allele: Any one of a number of viable DNA codings occupying a given locus (position) on a chromosome.
Usually alleles are DNA sequences that code for a gene, but sometimes the term is used to refer to a non-gene sequence.
In a diploid organism, like humans, one that has two copies of each chromosome, two alleles make up the individual's genotype.
For example, all of the people who have an A rather than a G at a particular location in a chromosome can have identical genetic variants at other SNPs in the chromosomal region surrounding the A.

14. Haplotype
Haplotype is a set of SNPs on a single chromatid that are statistically associated.

15. SNP Maps Sequence genomes of a large number of people
Compare the base sequences to discover SNPs.
Generate a single map of the human genome containing all possible SNPs => SNP maps

16. SNP Maps

17. The HapMap Project
The DNA samples for the HapMap come from a total of 270 people:
Yoruba people in Ibadan, Nigeria (30 both-parent-and-adult-child trios),
Japanese in Tokyo (45 unrelated individuals),
Han Chinese in Beijing (45 unrelated individuals),
CEPH (European) (30 trios).
These numbers of samples will allow the Project to find almost all haplotypes with frequencies of 5% or higher.
Ascertainment Bias (not enough samples to look at lower frequencies than 5%)
The HapMap is a catalog of common genetic variants that occur in human beings. It describes what these variants are, where they occur in our DNA, and how they are distributed among people within populations and among populations in different parts of the world. The International HapMap Project is not using the information in the HapMap to establish connections between particular genetic variants and diseases. Rather, the Project is designed to provide information that other researchers can use to link genetic variants to the risk for specific illnesses, which will lead to new methods of preventing, diagnosing, and treating disease.
The DNA in our cells contains long chains of four chemical building blocks -- adenine, thymine, cytosine, and guanine, abbreviated A, T, C, and G. More than 6 billion of these chemical bases, strung together in 23 pairs of chromosomes, exist in a human cell. (See for basic information about genetics.) These genetic sequences contain information that influences our physical traits, our likelihood of suffering from disease, and the responses of our bodies to substances that we encounter in the environment.
The genetic sequences of different people are remarkably similar. When the chromosomes of two humans are compared, their DNA sequences can be identical for hundreds of bases. But at about one in every 1,200 bases, on average, the sequences will differ (Figure 1). One person might have an A at that location, while another person has a G, or a person might have extra bases at a given location or a missing segment of DNA. Each distinct "spelling" of a chromosomal region is called an allele, and a collection of alleles in a person's chromosomes is known as a genotype.

18. Hapmap, SNPs, Haplotype, Tag SNPs
The construction of the HapMap occurs in three steps.
(a) Single nucleotide polymorphisms (SNPs) are identified in DNA samples from multiple individuals.
(b) Adjacent SNPs that are inherited together are compiled into "haplotypes."
(c) "Tag" SNPs within haplotypes are identified that uniquely identify those haplotypes.
By genotyping the three tag SNPs shown in this figure, researchers can identify which of the four haplotypes shown here are present in each individual.
In a given population, 55 percent of people may have one version of a haplotype, 30 percent may have another, 8 percent may have a third, and the rest may have a variety of less common haplotypes. The International HapMap Project is identifying these common haplotypes in four populations from different parts of the world. It also is identifying "tag" SNPs that uniquely identify these haplotypes. By testing an individual's tag SNPs (a process known as genotyping), researchers will be able to identify the collection of haplotypes in a person's DNA. The number of tag SNPs that contain most of the information about the patterns of genetic variation is estimated to be about 300,000 to 600,000, which is far fewer than the 10 million common SNPs.
Once the information on tag SNPs from the HapMap is available, researchers will be able to use them to locate genes involved in medically important traits. Consider the researcher trying to find genetic variants associated with high blood pressure. Instead of determining the identity of all SNPs in a person's DNA, the researcher would genotype a much smaller number of tag SNPs to determine the collection of haplotypes present in each subject. The researcher could focus on specific candidate genes that may be associated with a disease, or even look across the entire genome to find chromosomal regions that may be associated with a disease. If people with high blood pressure tend to share a particular haplotype, variants contributing to the disease might be somewhere within or near that haplotype.

19. SNPs may / may not alter protein structure
Genetic variants that alter protein functions are usually deleterious  less likely to become common or fixated.
Synonymous: AKA silent mutation, are mutations that have no functional affect on the protein.
Non-synonymous: amino acid-altering mutations  sickle cell anemia
Degeneracy of Genetic Code
Synonymous
Non- Synonymous

20. Alleles, polymorphism,…
One of two or more alternative forms of a gene or DNA sequence at a specific chromosomal location. Example: A or a
Polymorphism
Arbitrary defined as existing when at least two alleles are in the population and the minor allele is at a 1% or greater frequency.
Variant
An allele below the 1% frequency is sometimes called a variant.
Fixation:
The process by which one allele increases in a population until all other alleles go extinct and the locus becomes monomorphic.
Simply: 100% frequency.

21. Haplotype, hitchhiking, selective sweep
An allele that rises to a high frequency through positive selection at a linked locus is said to be “hitchhiking”.
The reduction in diversity at loci linked to a recently fixed allele is dubbed as “selective sweep”.
Ascertainment bias:
Distortion in a dataset caused by the way the markers or samples are collected.

22. How does human history affect genetic variation?
A genome-wide survey of Linkage Disequilibrium
Linkage disequilibrium is a phenomenon whereby genetic variants are associated: people who have one variant tend to have a second variant as well.
Slide by: David Reich, Broad Institute

23. Variation Over time Emergence of Variations Over Time
Common Ancestor
Emergence of Variations Over Time
time
present
Variations in Chromosomes Within a Population
Mutation
Slide by: David Reich, Broad Institute

24. Linkage Disequilibrium (LD)
Linkage Disequilibrium (LD): Nonrandom association between alleles at two or more loci, not necessary on the same chromosome, due to their tendency to be coinherited because of reduced recombination between them.
Linkage disequilibrium (LD) is a term used in the study of population genetics for the non-random association of alleles at two or more loci, not necessarily on the same chromosome. It is not the same as linkage, which describes the association of two or more loci on a chromosome with limited recombination between them. LD describes a situation in which some combinations of alleles or genetic markers occur more or less frequently in a population than would be expected from a random formation of haplotypes from alleles based on their frequencies.
Linkage disequilibrium is caused by fitness interactions between genes or by such non-adaptive processes as population structure, inbreeding, and stochastic effects. In population genetics, linkage disequilibrium is said to characterize the haplotype distribution at two or more loci. Formally, if we define pairwise LD, we consider indicator variables on alleles at two loci, say I1,I2. We define the LD parameter δ as:
Here p1,p2 denote the marginal allele frequencies at the two loci and h12 denotes the haplotype frequency in the joint distribution of both alleles. Various derivatives of this parameter have been developed. In the genetic literature the wording "two alleles are in LD" usually means to imply . Contrariwise, linkage equilibrium, denotes the case δ = 0.
- show the formulas on pge 130.
Pathogenic: Relating to the causation of disease.

25. LD Continued
LD describes a situation in which some combinations of alleles or genetic markers occur more or less frequently in a population than would be expected from a random formation of haplotypes from alleles based on their frequencies.
Illustrative Example:
Consider 2 neighboring loci A and B with two alleles
A and a, and B and b- at each locus
If no association: p (AB) = pA x pB
If there is association: D = PAB – PAx PB
Track LD:
Dt = (1-r)t x D0
r is the recombination rate

26. Take each oval and enlarge them and make it interactive

27. What Determines Extent of LD?
2,000 gens. ago
Mutation
1,000 gens. ago
Time = present
Recombination is the key!
Slide by: David Reich, Broad Institute

28. Positive Natural Selection
Neutral Evolution
Versus
Positive Natural Selection

29. Neutral Evolution Generations Genetic Drift: slow process1 2 3 4 5 6 7 8 9 10
Most of the genome is thought to be evolving neutrally, that is most of the mutations don’t have any effect on the fittness. When a neutral mutation occurs, its frequency in the population changes randomly. It is a slow process, known as genetic drift.
Genetic Drift: slow process
Frequency of the neutral mutations in the population changes randomly.
Generations
Reproduced from Sabeti et al.

30. Positive Natural Selection1
Positive Selection:
A selective regime that favors the fixation of an allele that increases
the fitness of its carrier.
Fixation: The process by which one allele increases in a population
until all other alleles go extinct and the locus becomes monomorphic.
Simply: 100% frequency.
The situation is different when selection operates, when a mutation will make it more possible for an individual to survive and to produce, it can spread quickly,infact it can go all the way to fixation that is 100% frequency. 2 3 4 5 6 7 8 9 10
Generations
Reproduced from Sabeti et al.

31. Methods for detecting selection
Difference between species
High proportion of function altering mutations
Within-species variation
Low diversity
Excess of derived alleles
Differences between populations
Long unbroken haplotypes

32. Methods for detecting selection
Test 1: Function altering mutations
Age: many millions of years

33. P. C. Sabeti et al., Science 312, 1614 -1620 (2006)
Test 1: High proportion of function altering mutations
Excess of function-altering mutations in PRM1 exon 2
Over a prolonged period, positive selection can increase the fixation rate of beneficial function-altering mutations.
Signature detected by comparing rates of mutations
power limited: needs multiple selected changes before gene stands out from background neutral rate
Synonymous mutation
genetic variants that alter protein function are usually deleterious and are thus less likely to become common or reach fixation (i.e., 100% frequency) than are mutations that have no functional effect on the protein (i.e., silent mutations). Positive selection over a prolonged period, however, can increase the fixation rate of beneficial function-altering mutations (20, 21), and such changes can be measured by comparison of DNA sequence between species. The increase can be detected by comparing the rate of nonsynonymous (amino acid–altering) changes with the rate of synonymous (si-lent) or other presumed neutral changes, by comparison with the rate in other lineages, or by comparison with intraspecies diversity. One extreme example of this kind of signature is found in the gene PRM1, mentioned earlier, which has 13 nonsynonymous and 1 synonymous differences between human and chimpanzee
This signature can be detected over a large range of evolutionary time scales. Moreover, it focuses on the beneficial alleles themselves, eliminating ambiguity about the target of selection. Its power is limited, however, because multiple selected changes are required before a gene will stand out against the background neutral rate of change. It is thus typically possible to detect only ongoing or recurrent selection. In practice, when the human genome is surveyed in this manner, few individual genes will give statistically significant signals, after correction for the large number of genes tested. However, the signature can readily be used to detect positive selection across sets of multiple genes (25). For example, genes involved in gameto-genesis clearly stand out as a class having a high proportion of nonsynonymous substitutions (25–27).
Common Statistical test:
Ka/Ks test
Relative rate test
McDonald-Kreitman test
P. C. Sabeti et al., Science 312, (2006)

34. Ka/Ks test (Li et al. 1985) Ka/Ks > 1 Main idea:
Contrast two types of substitutions events.
Goal of the test:
calculate the synonymous rate (Ks) and the non-synonymous rate (Ka), at each codon site.
Purified (negative) selection: Ka decreases,
Ka/Ks < 1 is indicative of purifying selection.
Positive Selection: Ka increases (replacement of amino acid is beneficial to the organism)
Ka/Ks > 1
In rare cases where the opposite pattern is observed (Ka /Ks>1), "positive Darwinian selection" is usually invoked as a possible explanation. The idea is that in such positive sites, rapid replacement of an amino acid is advantageous to the organism; hence mutations are fixed at a rate higher than that of neutral mutations,

35. Within-species Tests Test 2: Test 3: Test 4: Sweep Signatures
Low diversity, many rare alleles
(age < 250,000 years)
Test 3:
Many high frequency derived alleles
(age < 80,000 years)
Test 4:
Long common haplotypes unbroken by recombination
(age < 30,000 years)
Things to clarify: low diversity, rare alleles, derived alleles, haplotypes

36. Within-species Test 5: population Difference
Age < 50,000 to 70,000 years
Example 1
Extreme population differences (PD) in FY*O allele frequency.
FY*O allele, which confers resistance to P. vivax malaria, is prevalent and even fixed in many African populations, but virtually absent outside Africa.

37. P. C. Sabeti et al., Science 312, 1614 -1620 (2006)
5 Signatures of Selection
Tests based on human genetic variation includes (last 4 tests):
Development of anatomically modern human
Migration out of Africa
Development of agriculture
These different signatures of selection probe different periods of the evolutionary history, for example, difference between species are persistent and they can let us look for selections that occurred many millions of years ago.
Test based on human genetic variation probe much shorter time periods, from a few tens to a few hundreds of thousand years. Much shorter but still long enough to include the development of anatomically modern humanm the migration out of Africa, and the development of agriculture.
Published by AAAS
P. C. Sabeti et al., Science 312, (2006)

38. Genome-wide Studies New data sets make genome surveys possible
Full sequence for human, chimpanzee, mouse
Dense surveys of human genetic variation
Newly available data sets have fundamentally changed how we study natural selection in humans moving toward genome wide surveys, these key data sets includes full sequence for humans and other species as well as dense survey of human genetic variation.

39. Between-species results
Limited power to detect selection at single genes
Powerful for functional classes of genes rapidly changing:
Sperm-related genes
Olfactory (sense of smell) receptors
The rich sequence data has made between species of function altering mutations possible, these studies have limited power to detect selection at individual genes but it provides insight in the evolution of functional classes of genes. Rapidly evolving classes of genes that are identified include sperm-related genes and genes that are involved in the olfaction (sense of smell)

40. Finding selective sweeps
Statistical tests:
Distinguish the pattern of genetic variation expected under neutrality from that expected under natural selection
Pick a statistical test to detect sweeps
In studies of selective sweeps and in all studies of genome-wide variation, scientists have developed statistical tests that distinguishes the pattern of genetic variation expected under neutrality from that expected under neutral selection, they then apply that statistical test to the who genome, identifying candidate regions for selection.
Scientist devleope these statistical tests by using theoretical expectations and simulations of neutral variations, If the neutral variations in the genome was completely predicted by simulations we could then in principle simply identify the variants that are outliers from this distribution as possible candidates of selection. The problem however is that We do not fully know the shape of the neutral distribution and how it’s affected by other factors such as demographic history.
Demographic: a single vital or social statistic of a human population, as the number of births or deaths.
Apply the statistic across the genome

41. Most likely candidate of selection
Finding selective sweeps
Problem
We do not fully know the shape of the neutral distribution and how it’s affected by other factors such as demographic history.
However, the best we can do:
use statistic based on simulations
apply it to empirical genome-wide data sets
Identify the loci in the extreme tail
Most likely candidate of selection
Therefore, the best we can do is to use the statistic based on our existing simulations and apply them to our empirical genome-wide sets, then identify the loci in the extreme tail, that would be the most likely candidate for natural selection

42. No Selection Positive Selection Test based on the relationship between
allele frequency and extent of linkage disequilibrium
Young alleles:
• low frequency
• long-range LD (long haplotypes)
No Selection
Old alleles:
• low or high frequency
• short-range LD
Positive Selection
Young alleles:
• high frequency
• long-range LD
Slide by: David Reich, Broad Institute

43. The signal of selection
Neutrality
Linkage Disequilibrium
(Homozygosity)
Positive Selection
Frequency of core haplotype
frequency
Slide by: David Reich, Broad Institute

44. Let us understand these methods better …

45. Methods for detecting selection
Within-species Tests
Test 2: Low genetic diversity/many rare alleles
Age < 250,000 years

46. Test 2: Low genetic diversity/many rare alleles
As allele increases in population frequency  variants at nearby locations
on the same chromosome (linked variants) rise in frequency.
Such so-called "hitchhiking"  "selective sweep”.
Most common type of variant used: SNPs
Low diversity and many rare alleles at the Kell blood antigen cluster
Common Statistical Test:
Tajima’s D
Hudson-Kreitman-Aguade (HKA)
Fu and Li’s D*
Reduction in genetic diversity (age <250,000 years). As an allele increases in population frequency, variants at nearby locations on the same chromosome (linked variants) also rise in frequency. Such so-called "hitchhiking" leads to a "selective sweep," which alters the typical pattern of genetic variation in the region. In a complete selective sweep, the selected allele rises to fixation, bringing with it closely linked variants; this eliminates diversity in the immediate vicinity and decreases it in a larger region. New mutations eventually restore diversity, but these appear slowly (because mutation is rare) and are initially at low frequency. Positive selection thus creates a signature consisting of a region of low overall diversity, with an excess of rare alleles.
Unlike excess functional changes, which involve differences between species, selective sweeps are detected in genetic variation within a species. The most common type of variant used is the single-nucleotide polymorphism (SNP). As an example, Akey et al. identified a 115-kb region containing four genes including the Kell blood antigen, which showed an overall reduction in diversity and more rare alleles in Europeans than expected under neutrality (Fig. 3) (28). Statistical tests commonly used to detect this signal include Tajima's D, the Hudson-Kreitman-Aguadé (HKA) test, and Fu and Li's D* (29–32).
On the basis of three different statistical tests, the 115-kb region(contatining four genes) shows evidence of a selective sweep in Europeans.
P. C. Sabeti et al., Science 312, (2006)

47. Methods for detecting selection
Within-species Tests
Test 3: High-frequency derived alleles
Age < 80,000 years

48. Test 3: Many high-frequency derived alleles
Derived alleles: non-ancestral alleles
Arise by new mutations
Typically lower allele frequencies than ancestral
However, in selective sweep, derived alleles linked to the beneficial alleles can hitchhike to high frequency.
Derived (that is, nonancestral) alleles arise by new mutation, and they typically have lower allele frequencies than ancestral alleles (33). In a selective sweep, however, derived alleles linked to the beneficial allele can hitchhike to high frequency. Because many of these derived alleles will not reach complete fixation (as a result of an incomplete sweep or recombination of the selected allele during the sweep), positive selection creates a signature of a region containing many high-frequency derived alleles. A good example of this kind of signature is the 10-kb region around the Duffy red cell antigen (FY), which has an excess of high-frequency derived alleles in Africans, thought to be the result of selection for resistance to P. vivax malaria (Fig. 4) (34, 35). The most commonly used test for derived alleles is Fay and Wu's H (36).
FIGURE: The 10-kb region near the gene has far greater prevalence of derived alleles (represented by red dots) than of ancestral alleles (represented by gray dots)
Figure: Excess of high-frequency derived alleles at the Duffy red cell antigen (FY) gene (34)
P. C. Sabeti et al., Science 312, (2006)

49. Methods for detecting selection
Within-species Tests
Test 5: Differences between populations
Age < 50,000 to 70,000 years

50. Test 5: Differences between populations
Geographically separate populations are subject to distinct environmental or cultural pressures  change of allele frequency in one populations and not the other.
Can only arise when populations are at least partially isolated reproductively.
For humans, after the major human migrations out of Africa some 50,000 to 70,000 years ago.
Weakness of the test: similar to other population genetic signatures, distinguishing between genuine selection and the effect of demographic history (especially population bottleneck) on genetic variation can be hard.
Common Statistical Tests:
FST
Pexcess
Differences between populations (age <50,000 to 75,000 years). When geographically separate populations are subject to distinct environmental or cultural pressures, positive selection may change the frequency of an allele in one population but not in another. Relatively large differences in allele frequencies between populations (at the selected allele itself or in surrounding variation) may therefore signal a locus that has undergone positive selection. For example, the FY*O allele at the Duffy locus is at or near fixation in sub-Saharan Africa but rare in other parts of the world, an extreme case of population differentiation (Fig. 5) (34, 38). Similarly, the region around the LCT locus demonstrates large population differentiation between Europeans and non-Europeans, reflecting strong selection for the lactase persistence allele in Europeans (6). Commonly used statistics for population differentiation include FST and pexcess (39–41).
Reduction in size of a single, previously larger, population and a loss of prior diversity.

51. Extreme population Difference
Example 1
Extreme population differences (PD) in FY*O allele frequency.
FY*O allele, which confers resistance to P. vivax malaria, is prevalent and even fixed in many African populations, but virtually absent outside Africa.

52. Extreme population Difference
Example 2:
Region around LCT locus demonstrates large PD between Europeans and non-Europeans  Strong selection for lactase persistence allele in Europeans.
LCT

53. Genome-wide Survey using Tests: Low diversity and population separation
Several studies have now produced genome-wide distributions for varying tests of selection and identified the outliers of the distribution. Here, we present plots of diversity versus population differentiation.
Signal of selective sweep occur here in regions of low diversity and high population differentiation. These surveys have made it possible to find many novel candidates, also they also allow us to re-examine previously published candidates in comparison with genome-wide distributions. Some candidates remain among the strongest signals such as lactase and imino-globiline A (IgA) while others do not.
Outliers: low diversity with high population differentiation

54. A Little Break?

55. Interesting fact: Pardis Sabeti is a rock star!

56. Back to work now…

57. Methods for detecting selection
Within-species Tests
Test 4: Long Haplotypes
Age < 30,000 years

58. Recent past Present A short while later… Mb 0.1 0.2 0.3 Advent of
a beneficial allele
A short while later…
Recent past
Present Mb
0.1
0.2
0.3
Under positive selection, a selected allele may rise in prevalence rapidly enough that recombination does not substantially break down the association with alleles at nearby loci on the ancestral chromosome. Such a collection of alleles in a chromosomal region that tend to occur together in individuals is termed a haplotype. Selective sweeps can produce a distinctive signature that would not be expected under neutral drift—namely, an allele that has both high frequency (typical of an old allele) and long-range associations with other alleles (typical of a young allele). The long-range associations are seen as a long haplotype that has not been broken down by recombination. For example, the lactase persistence allele at the LCT locus lies on a haplotype that is common ( 77%) in Europeans but that extends largely undisrupted for more than 1 million base pairs (1 Mb) (Fig. 6) (6), much farther than is typical for an allele of that frequency. This signature can be detected with the long-range haplotype (LRH) test, haplotype similarity, and other haplotype-sharing methods (42–45). Developing such tests is an area of vigorous current investigation (46, 47).
Model: Maynard Smith and Haigh, 1974, Simulation by SelSim, Coop and Spencer, 2004

59. Haplotype Core Haplotypes gene 1 2 3 4 5
Adjacent SNPs that are inherited together are compiled into "haplotypes."
Slide by: David Reich, Broad Institute

60. Long-range multi-SNP haplotypes
Core
markers
gene
C/T A/G A/G C/T C/T C/T
Long-range markers 5 3 2 1 4
Decay of LD
Slide by: David Reich, Broad Institute

61. Powerful general approach for detecting selection3 2 1 4
Slide by: David Reich, Broad Institute

62. Long-range multi-SNP haplotypes
gene
C/T
A/G
Core
markers
Long-range markers T C A G
18% G C T
35%
75% 3
EHH estimates the level of haplotype splitting due to recombination and mutation at extended regions on both sides of a specified core region.
100%
Decay of homozygosity (probability, at any distance, that any two haplotypes that start out the same have all the same SNP genotypes)
Slide by: David Reich, Broad Institute

63. Test Statistic: Extended Haplotype Homozygosity (EHH)
(A) Decay of haplotypes in a single region in which a new selected allele (red) is sweeping to fixation, replacing the ancestral allele(blue).
(B) Decay of haplotype homozygosity for ten replicate simulations.
Right side: derived alleles are favored
sigma=2Ns= 250
Figure 1. Decay of EHH in Simulated Data for an Allele at Frequency 0.5
(A) Decay of haplotypes in a single region in which a new selected allele (red, center column) is sweeping to fixation, replacing the ancestral allele
(blue). Horizontal lines are haplotypes; SNP positions are marked below the haplotype plot using blue for SNPs with intermediate allele frequencies
(minor allele .0.2), and red otherwise. For a given SNP, adjacent haplotypes with the same color carry identical genotypes everywhere between that
SNP and the central (selected) site. The left- and right-hand sides are sorted separately. Haplotypes are no longer plotted beyond the points at which
they become unique.
(B) Decay of haplotype homozygosity for ten replicate simulations. When the core SNP is neutral (s¼0; left side) the haplotype homozygosity decays at
similar rates for both ancestral and derived alleles. When the derived alleles are favored (s¼2Ns¼250; right side), the haplotype homozygosity decays
much slower for the derived alleles than for the ancestral alleles. The discrepancy in the overall areas spanned by these two curves forms the basis of
our text for selection (iHS).
DOI: /journal.pbio g001
Haplotype homozygosity (HH) is an effective measure of linkage disequilibrium (LD) for more than 2 markers.
If we want to see, how LD breaks down with increasing distance to a specified core region, we can calculate HH in a stepwise manner as extended HH (EHH, Sabeti et al. 2002).
HH is evaluated as HH = [sum(pi2) - 1/n] / [1 - 1/n] with pi being the relative haplotype frequency and n the sample size.
EHH estimates the level of haplotype splitting due to recombination and mutation at extended regions on both sides of a specified core region.
In combination with the core haplotype frequency it may also serve as an indicator of recent positive selection. Frequent core haplotypes with an unusually high long-range LD are supposed to be positively selected. The various core haplotypes can serve as internal controls.

64. Simulations of decay of haplotype homozygosity
Have to explain EHH here.
Sharareh: As shown in the figure, when an allele rises rapidly in frequency due to strong selection, it tends to have high levels of haplotype homozygosity extending much further than expected under a neutral model. hence in plots of EHH versus distance, the area under the EHH curve wil usually be much greater for a selected allele than for a neutral allele.
What is sigma?
10 simulations: SelSim
Haplotype Homozygosity: Sabeti et al. 2002

65. iHS: Measures the extent of haplotypes along alleles at a given SNP
Derived Allele
Ancestral Allele
EHH
Perhaps triage this slide?
0.05
Genetic Distance
iHHA : iHH with respect to Ancestral core allele.
iHHD : iHH with respect to Derived core allele.

66. iHS Score Useful for variants that have not yet reached fixation.
Large negative iHS: derived allele has swept up in frequency
Large positive iHS: an ancestral alleles hitchhike with the selected sites.
Hence, both cases are considered interesting!

67. The Data: Hapmap Project
860,000 SNPs genome-wide
60 unrelated individuals
European (CEPH): CEU
Nigerians from Ibadan (Yoruba): YRI
89 unrelated individuals
Han Chinese from Beijing and Japanese from Tokyo: ASN

68. |iHS|
|iHS|
|iHS|
-iHS

69. Lines of Evidence: Selection
Enrichment of signal in genic relative to non-genic regions (p < 10-20)
Replication of previously published candidates
LCT (Bersagleri et al 2004, Coelho et al 2005)
ADH (Osier et al 2002)
17q23inv (Stefanson et al 2005)
CYP3A5 (Thompson et al 2004)
Ch. 11 Olfactory gene Cluster (Gilad et al, 2003)
Correlates with departures in the frequency spectrum (Fay and Wu, 2000)

70. SYT1, Yoruba Haplotype Decay at SYT1, iHS = -4.7
Binds Ca2+; implicated in release
of neurotransmitters [OMIM]
Plot of high iHS scores on
Chromosome 12
Haplotype
Decay at
SYT1,
iHS = -4.7

71. SPAG4, East Asians Haplotype Decay at SPAG4, iHS = -3.1
Interacts with ODF27, gene found
in mammalian sperm tails [OMIM]
Plot of high iHS scores on
Chromosome 20
Haplotype
Decay at
SPAG4,
iHS = -3.1
Describe scatter plot; perhaps cut one of these slides (if time constraints)

72. How much do regions overlap across populations?

73. Distribution of Regions Across Populations

74. Do signals of selection correlate with known biological processes?

75. Enriched Ontological Categories
Olfaction [CEU/YRI]
Gametogenesis/Fertilization [ASN/CEU]
MHC-I related Immunity [CEU/YRI]
Metabolism [All]

76. Other Interesting Stories...
Skin Pigmentation
MYO5A, OCA2, DTNBP1, TYRP1, SLC24A5* [CEU]
Sugar Metabolism
MAN2A1 [ASN/YRI]; SI [ASN]; LCT [CEU]
Processing of Fatty Acids
SLC27A4, PPARD [CEU]; LEPR [ASN]; NCOA1 [YRI]
Add MCPHX – XX; Tell stories, don’t say where they are from (already mentioned it).

77. Conclusion on iHS Method
Pervasive signals of positive selection across the human genome
Both population specific and signals shared between populations
Strong evidence of selection in Africa (unlike other reports)
Putative medical relevance because:
Have phenotypic consequences
Differences between populations

78. Happlotter

79. That's all folks!