Human population genetics

Rough outline of human evolution What makes us human? Variation in modern humans Relevance to human health

Relationship to other primates

Our closest living relative appears to be the chimpanzees. . .

Pan troglodytes (chimpanzee)

Pan paniscus (bonobo)

Humans, chimps, and gorillas

Chen & Li1 sequenced 53 regions randomly chosen short regions from humans, chimps, and gorillas. They then estimated 53 phylogenetic trees. . .

C 31

G 10

G 12

1 Am.

J. Hum. Genet. 2001

Gene trees and species trees

If species are separated by enough time for gene lineages to coalesce within species, then gene trees will reect species trees. Coalescence times within species depend on population sizes, geographic structure, etc.

Gene trees and species trees

If species are not separated by enough time for gene lineages to coalesce within species, then gene trees will often not reect species trees. The human-chimp split is borderline: coalescence times within the species seem to lie between 0.5 and 1 Myr; the species may only be separated by 45 Myr.

Perspective on polymorphism
two human genomes dier from each other about once every 1,000 base pairs chimps dier from each other once every 200 base pairs fruit ies dier from each other once every 100 base pairs Californian intertidal copepods (Suzanne Edmands) may dier from each other every 10 bp a human and a chimp genome dier from each other about once every 100 base pairs

Most dierences have no eect

There is no direct relationship between the amounts of genetic and phenotypic dierentiation. Of the 109 103 = 106 dierences between two humans, most mean nothing!

Aside: The Neutral Theory

Formulated in the late 1960s in response to the polymorphism data that was then becombing available: Protein sequencing revealed an approximately linear relationship between divergence times and sequence divergence: the molecular clock (Zuckerkandl & Pauling, 1965) Protein electrophoresis (Lewontin & Hubby, 1966) revealed substantial variation within species: 1550% of enzyme-coding genes were polymorphic with more than one common allele

The molecular clock

The Neutral Theory of Molecular Evolution

Kimura (1968): most mutations are deleterious and are rapidly eliminated a very small number of mutations are favorable and are rapidly xed most of the variation that we observe within species is selectively neutral, and is governed by the interplay of mutation and genetic drift by extension, most of the dierences between species are simply due to the random xation of mutations

Synonymous and nonsynonymous sites

Evidence for the Neutral Theory

Further evidence for the Neutral Theory

The importance of gene regulation

The low level of dierention between humans and chimps on the amino-acid level led King & Wilson2 to propose that most of the dierences were regulatory We still know next to nothing about the genetic dierences between chimps and humans Recently, Bailey et al.3 found that there are more genetic dierences than previously believed in terms of segmental duplications. . .

2 Science 3 Science

1975 2002

Types of mutations
single base-pair substitutions insertions and deletions inversions translocations duplications

How can we nd the dierences that matter?

A geneticist would either: make some crosses, or; carry out a mutational screen (for the chimp mutation. . . ) The former makes a lot more sense than the latter, but neither is feasible. Instead, we have to rely on indirect evidence traces of selection in the pattern of polymorphism and divergence either in candidate loci or in genome-wide screens

Traces of selection
Selection can leave traces two main dierent ways4 : rapid selection for a new function that requires many functional changes can result in a detectable excess of functional changes whenever an allele is swept through a population by selection, linked neutral polymorphisms hitch-hike along possibly leaving a detectable trace in the regional pattern of overall polymorphism

4 Reviewed by Kreitman, Annu. Rev. Genomics & Hum. Genet. 2000, vol. 1

Example: very rapid evolution of reproductive proteins

The per-site rate of non-synonymous substitution (Ka ) is typically much lower than the per-site rate of synonymous substitution (Ks ) Ka /Ks > 1 is a sure sign of directional selection (but is an extremely conservative test) Examples: sea-urchin sperm binding protein and lysin, Drosophila male reproductive proteins female reproductive proteins in mammals also various resistance alleles

The McDonald-Kreitman test

A more powerful test would utilize within-species polymorphism as well. A very simple test that does this is the McDonald-Kreitman test, which utilizes the fact that, under neutrality, polymorphism and divergence are both proportional to u, the neutral mutation probability. Consider, for example, G6pd from D. melanogaster and D. simulans: type of dierence xed polymorphic 21 2 26 36

type of change non-synonymous synonymous

Selective sweeps
The methods mentioned will detect neither simple amino-acid substitutions nor regulatory changes To detect these, we need to look for the footprint of selection in the overall pattern of polymorphism
Advantageous variant

Selection

Teosinte to corn: < 10,000 years; ve genes?

teosinte

maize

maize with tb1 mutation

A sweep in the tb1 promotor region?

Gene trees reect the sweep

a) tree based on the 1,729-bp transcribed region b) tree based on the 1,143-bp 5' non-transcribed region

Selective sweeps are the Holy Grail

No successes in humans yet We can only expect to detect sweeps that were recent enough and strong enough to change the local genealogical structure

How related are we?

Lets compare you to your brother or sister. . . about 50% of your genomes will be identical by descent from mom or dad for the rest, you may well be more closely related to someone from Africa (or Sweden, or the Brazilian rain forest) than to each other on average, you will be slightly more closely related to someone from your ethnic group on average, the time back to a common ancestor is 800,000-1,000,000 years the level of polymorphism reects this. . .

We may all be more closely related for some loci

0.0015

0.001

p
0.0005

selective sweep?
0 0 5 10 15 position Mb 20 25

A gene for language?

The ability to to develop articulate speech relies on capabilities, such as ne control of the larynx and mouth, that are absent in chimpanzees and other great apes FOXP2 is the rst gene relevant to the human ability to develop language Mutations in FOXP2 appear to cause a (dominant) disorder characterized by severe articulation diculties accompanied by linguistic and grammatical impairment

 Enard et al.5 studied variation at FOXP2 within humans, and also sequenced it in several primates The pattern of substitutions suggests recent selection

5 Nature

2002

 There are also signs of selection in the pattern of polymorphism overall diversity is not decreased, but: there is an excess of rare alleles there are too many non-ancestral alleles at high frequency It has been suggested (in particular by Richard Klein, an anthropologist at Stanford University) that something dramatic happened in human evolution about 50,000 years ago, possibly due to improved communication, and that this could be due to a single mutation. . . (I nd the notion of a single gene for culture ridiculously simplistic)

How many genes?

Early hominid evolution

bipedal, with small brains and large protruding faces with large canines fossils are all found in Southern or Eastern Africa, and are assigned to Ardipithecus and Australopithecus start appearing about 4.4 mya the famous Lucy (Australopithecus afarensis) is perhaps 3 million years old between 3 and 2.5 mya, there appears to be many species, some robust, some gracile robust forms disappear roughly 1 mya; some gracile species gives rise to. . .

The genus Homo

Fragmentary fossils as early as 2.5 mya; also rst evidence of manufactured stone tools Several species around 2 mya in East Africa: H. habilis, H. rudolfensis, and H. erectus H. erectus appears to have been completely bipedal, and was about the size of a modern human, but with a smaller brain H. erectus is found world-wide: Indonesia 1.61.8 mya; China 1.11.9 mya; Georgia 1.7 mya H. erectus appears to have been much more carnivorous than other hominids

But. . .
This summer, Brunet et al.6 , reported a remarkable 6 million year old fossil from Chad: Sahelanthropus tchadensis looks like a chimp from behind, but. . . a 1.75 million year old australopith from the front. . . This nding wreaks havoc with many tidy scenarios for hominid evolution. . .
6 Nature

2002

Too many stories, too little data. . .

It is important to recognize that: whereas all living things have ancestors, not all fossils have descendants only a small fraction of all species that ever existed will have left fossils Tavar et al.7 estimated that only 7% of all primate species e that ever existed have left fossils. They also estimated that actual divergence times between species will typically be at 70% older than the age of the oldest known fossil. . .
7 Nature

2002

Speciation model

Conclusion
Things are probably far more complex than suspected Famous fossils like Lucy are probably not ancestral to us (almost certainly not in the genetic sense) I suspect the same will be true for modern human origins. . .

The origin of modern humans

Archaic H. sapiens start appearing 400,000-500,000 years ago: they generally have brain sizes similar to modern humans, but rather more robust skulls Neanderthals are an example of archaic humans How modern humans evolved from H. erectus is extremely controversial. . .

Israel, 90,000 years ago

France, 23,00027,000 years ago

Australia, 9,00013,000 years ago

Out of Africa

Mitochondrial Eve
Africa Europe Asia Africa Europe Asia Africa Europe Asia 0

Africans non-Africans

0.5

1 migration Africa Schematic version of the human mtDNA tree a) Out-of-Africa Model Africa b) Multiregional Model Africa c) Candelabra Model million years ago

A single gene tree does not tell us much. . .

Variation in modern humans

Humans have less polymorphism than other primates There is typically more variation in African samples (exceptions seem to involve selection) There is typically less LD in African samples

Race
Next time you hear or read There is no biological basis for race, please ask what no basis means. . . Race, in the sense of genetically based phenotypic dierences between humans that are not geographically randomly distributed, clearly exist Race, in the sense of pure lines, certainly does not In terms of genetic markers, there is much more variation within groups than between groups . . . however, this may not be all that relevant Human variation is continuous. . .

What is going on?

A biased selection of current topics: The Human Haplotype Map Project Common Disease Common Variant? Predicting drug response: Genaissance Pharmaceuticals Should ethnicity be taken into account? Genotyping Iceland: deCode Genetics

The Human HapMap Project

A genome-wide screen for disease association is too expensive

There are plenty of markers: 1.8 million SNPs are available from the private/public SNP Consortium8 However, typing a SNP currently costs $0.30. . . . . . so typing a single individual for half a million SNPs would cost $150,000. . . . . . and a typical epidemiological study involves hundreds or thousands of individuals
8 http://snp.cshl.org/

Population genetics to the rescue?

As seen here in the chromosome 21 data discussed earlier, the pattern of LD is blocky: this is expected under the standard coalescent model, but the eect is probably enhanced by heterogeneity in the recombination rate along the chromosomes.

The HapMap Project

If we knew the pattern of LD, it may be possible choose markers intelligently Many unanswered questions: Will markers chosen in one sample work in another? Are genome-wide screens a waste in any case? $100 million have been allocated to nding out9 . . .

9 http://www.nhgri.nih.gov/About

NHGRI/Der/variat.htm

Common Disease Common Variant?

Genome-wide screens for association are unlikely to work unless there are major allelesalleles that explain a large fraction of the variation in the population Mendelian diseases are typically rare: the target is common diseases Are common diseases due to common alleles, or are they due to many alleles (each of which has a tiny eect)? If there are many alleles, are there also many loci? If there are many loci, do they interact epistatically?

Predicting drug response

A very large number of drugs fail testing not because they dont work, but because they have signicant side eects in a small subset of the population It may well be that this often has a relatively simple genetic basis Genaissance Pharmaceuticals10 is trying to identify haplotype markers short fragments containing several SNPs that can be used to predict drug response They are completely sequencing the coding regions of 10,000 genes in a sample of 96 individuals
10 http://www.genaissance.com

Preliminary results from 313 genes were published by Stephens et al.a The distribution of variation among ethnic groups agree with other results.
a Science

2001

Genes, Race, and Disease

A recent debate concerned the use of race in epidemiology: Wilson et al.11 published a study showing that drug response was better predicted by genetic markers than by ethnicity. In other words, let the markers decide the grouping, and ignore what you think people are, or what they say they are. Risch et al.12 replied that this was complete nonsense:
11 Nature 12 Genome

Genet. 2001 Biology, 2002

 the results in question was an artifact of lumping New Guineans and Chinese as Asians with better samples and more markers, markers and self-proclaimed ethnicity are likely to agree much better (cf. Noahs results) in any case, self-proclaimed ethnicity should always be taken into account, because it is likely to reect the environment race is not only a matter of genes

deCode Genetics
A dierent approach to common diseases is taken by deCode Genetics13 . They argue that the Devil is in the Details: The common diseases public health problems such as stroke, heart disease, and Alzheimers result from the interplay of multiple genes and environmental and health factors Unraveling this complexity requires the ability to gather and correlate detailed information on disease and genetic variations across as large a group of people as possible: a population.
13 http://www.decode.com

 To do this eciently it is also critical to have accurate and comprehensive genealogical records, the only means for tracing how the genetic components of disease travel between generations of vast extended families. A population with all three kinds of data genetic, disease, and genealogical is needed. . .

Genotyping Iceland
Iceland was founded by Norwegian vikings (and probably Celtic slaves) about 1100 years ago. Islendingabok The Book of Icelanders contains more than 95% of all those who have lived in Iceland since the rst census in 1703, and stretches back to the 9th century. In all, it contains more than half of all Icelanders who have ever lived. Iceland has a well-functioning national health service, with excellent records DNA samples and disease data are being collected from approximately 80,000 volunteers: 1/3 of the adult population is participating in deCodes projects, as are more than 90% of those over 65.

Controversial stu
deCODE holds a license from the government of Iceland to build and commercialize the Icelandic Health Sector Database (IHD). Now under construction, the IHD will assemble anonymized, encrypted data from patient records from Icelands national health service in a secure computer system. The IHD will enable users, including doctors in Icelands national health system, to conduct population analyses of longitudinal healthcare data and trends on, for example, clinical measurements and basic lifestyle information, disease diagnoses, treatments and outcomes. deCODE believes that this system will provide valuable resource for better understanding the environmental components that, along with genetic factors, lie behind the onset of common diseases.

Publications 2002
PRINTABLE PAGE 2002-12-06 09:32:01 -0800

2002
AA Hicks, et al. A susceptibility gene for late-onset idiopathic Parkinson's disease Annals of Neurology (Vol. 52 No. 5) Gislason T, et al. Familial Predisposition and Cosegregation Analysis of Adult Obstructive Sleep Apnea and the Sudden Infant Death Syndrome American Journal of Respiratory and Critical Care Medicine (Vol. 166) Stefansson H, et al. Neuregulin 1 and Susceptibility to Schizophrenia. American Journal of Human Genetics (Vol. 71 No. 4) Hakonarson H, et al. A Major Susceptibility Gene for Asthma Maps to Chromosome 14q24. American Journal of Human Genetics (Vol. 71 No. 3) Gudbjartsson T et al. A population-based familial aggregation analysis indicates genetic contribution in a majority of renal cell carcinomas. International Journal of Cancer (Vol. 100 No. 4) Kong A, et al. A high-resolution recombination map of the human genome Nature Genetics (Vol. 33 No. 3) Kristjansson K, Manolescu A, Kristinsson A, Hardarson T, Knudsen H, Ingason S, Thorleifsson G, Frigge ML, Kong A, Gulcher JR, Stefansson K. Linkage of essential hypertension to chromosome 18q. Hypertension (Vol. 39 No. 6) Gretarsdottir S et al. Localization of a Susceptibility Gene for Common Forms of Stroke to 5q12. American Journal of Human Genetics (Vol. 70 No. 3) Gudmundsson G et al. Localization of a Gene for Peripheral Arterial Occlusive Disease to Chromosome 1p31. American Journal of Human Genetics (Vol. 70 No. 3) Stefansson H, Geirsson RT, Steinthorsdottir V, Jonsson H, Manolescu A, Kong A, Ingadottir G, Gulcher J, Stefansson K. Genetic factors contribute to the risk of developing endometriosis. Human Reproduction (Vol. 17 No. 3) Gudjonsson JE, Karason A, Antonsdottir AA, Runarsdottir EH, Gulcher JR, Stefansson K, Valdimarsson H. HLA-Cw6-Positive and HLA-Cw6-Negative Patients with Psoriasis Vulgaris have Distinct Clinical Features. The Journal of investigative dermatology (Vol. 118 No. 2).
Copyright deCODE genetics, Inc. 1995-2002

http://www.decode.com/main/view.jsp?branch=19134&e196OriginBranchID=3627&e196Colnumber=2&e196Rownumber=4&oldbranch=3627

Page 1