Download Human Genetic Variation, Shared and Private

PERSPECTIVES
GENETICS
Human Genetic Variation,
Shared and Private
Analyses of rare and common variants in the
genome reveal another layer of human genetic
variation within and among populations.
Ferran Casals and Jaume Bertranpetit
CREDIT: C. BICKEL/SCIENCE
T
he development of agriNot all loss-of-function variants
culture and livestock in
will have the same phenotypic
the transition from the
effect, with some better tolerated
Paleolithic to the Neolithic era
because of genetic and functional
some 10,000 years ago heralded
redundancy. Because the relationa demographic explosion in our
ship between a mutation and its
species that is still ongoing today.
functional effect is not straightParadoxically, this rapid popuforward, more sophisticated prelation growth, made possible by
dictive tools are needed that conimproved living conditions, may
sider aspects such as the role and
be responsible for an excess of
position in gene interaction netdamaging variants in our genome.
works and the final phenotypic
Two papers in this issue, by Teneffect given possible alternative
nessen et al. (1) on page 64 and
pathways that may hide the deleNelson et al. (2) on page 100,
terious or lethal effect of a mutareport deep-resequencing analtion. Predictions will have to conyses of all the human proteinsider the effect on organisms, not
coding genes (the exome) and of
only on protein structure.
202 genes that are putative drug
The studies by Tennessen et
targets in thousands of individual. and Nelson et al. sound two
als, respectively. The studies show
important warnings regarding
that most of the genetic variants
the experimental design of assooccur at very low frequencies in
ciation analyses for rare varihuman populations, which accu- Human genetic variation. Proportion of shared and unshared (private) variants ants: Large samples sizes will
mulate an excess of potentially between the African-American and the European-American populations [data be required (sometimes too large
from (1)].
harmful mutations.
for actual studies), and replicaUntil recently, the snapshot
tion across populations will be
of human genetic variation was mainly and function (1–6) and, hence, are poten- limited, as most of the rare variants will be
restricted to variants at frequencies above tially involved in disease. The proportion of specific to each population. Genome-wide
~1 to 5%, because of technological and cost rare, functional nucleotide changes greatly association studies, which include up to sevlimitations. As such, medical and population exceeds the expected number predicted by eral million polymorphic positions throughgenetics studies have been based on infor- classical population genetics analysis. This out the genome, have provided insight into
mation derived from common variants called seems to be due to the inefficiency of natural complex disorders by accounting for up to
polymorphisms (and is the reason that stud- selection in removing these variants because 10 to 20% of their genetic component (9).
ies have focused on single-nucleotide poly- of the dramatic population growth in the last The rest of the heritability remains to be
morphisms, or SNPs). Now, next-genera- few thousand years (3, 4), which could lead explained, and rare variants are among the
tion DNA sequencing technologies allow to an important reduction of fitness in future alternative proposed factors (10, 11). Conthe analysis of thousands of samples and societies (7).
tingency tables comparing the allele frethe characterization of less frequent genetic
This mutational burden in the population quency in control groups versus patients do
variation. This variation is younger than that is also seen at the individual level. Each per- not have enough statistical power because
studied previously and harbors informa- son carries some 35 nonsense substitutions of the low frequency of rare variants, even
tion about more recent demographic events: that disrupt the formation of functional pro- for very large samples. Much effort is being
There is an excess of rare variants (at a fre- teins. This number increases to 100 when all invested in developing statistical tools to
quency q ≤ 0.5%) in relation to the exist- the loss-of-function mutations are consid- check for association of rare variants with
ing models due to the explosive population ered, with 20 substitutions in a homozygous complex disease, usually considering an
growth of recent epochs (1–6).
state (8), while 10 to 15% of the individuals aggregate effect of rare variants (12), which
Low-frequency variants are enriched are compound heterozygotes for nonsense is of statistical interest but has poor biomedifor changes that alter protein sequence variants in at least one gene. These figures cal interpretation. Thus, samples comprising
raise the question of predicting the func- thousands of individuals still have reduced
tional impact of DNA sequence variants not statistical power even to detect large genetic
IBE–Institute of Evolutionary Biology (UPF-CSIC), Univeronly at the local level of protein structure and effects, leading to the need for even bigger
sitat Pompeu Fabra, Barcelona, Catalonia, Spain. E-mail:
[email protected]
function but also at the level of the organism. sample sizes (1).
www.sciencemag.org SCIENCE VOL 337 6 JULY 2012
Published by AAAS
39
PERSPECTIVES
Replication of the association across populations is crucial in genetic studies. In general, allele frequencies for common variants
do not differ significantly across populations,
especially within the same continent (13). But
this picture becomes different for rare variants (14), as many will be population-specific. The sharing of rare variants is reduced
to about 10 to 30% among populations in different continents (see the figure) and to 70 to
80% among populations in the same continent (1, 2). Consequently, those variants are
more prone to show geographic stratification
(due not to a predisposition to disease but to
geographical differences), which may lead to
false-positives (1, 15) and predicts a lack of
replication across populations. Reduced allele
sharing across populations points out the need
for population-specific catalogs of variants, as
well as careful collection of information about
ancestry and correction for stratification. The
genetic analysis of complex diseases through
the comparison of cases and controls, which
has long been a widely used study design, may
be approaching a conundrum that may not be
solved just by increasing sample sizes and
doing the most detailed type of analysis, the
comparison of gene sequences.
Studies based on whole-genome
sequences in thousands of individuals over
the next few years should help to clarify the
role of rare variants and their distribution in
the genome and across human populations;
such studies will represent an analytical and
a computational challenge. But first, a better understanding of the functional impact
of non–protein-coding variation, including
regulatory regions, is needed as well as the
development of tools to carefully functionally
annotate genome variants. These steps would
allow prioritizing of candidate variants and
genes, both in the filtering approaches for rare
Mendelian diseases and in the assignment of
different weights in aggregate tests for complex disease studies and the interpretation
of personalized genomic risk profiles. Ulti-
mately, we will need to go back to basic biology to link genome variation and phenotypes.
References and Notes
1. J. A. Tennessen et al., Science 337, 64 (2012); 10.1126/
science.1219240.
2. M. R. Nelson et al., Science 337, 100 (2012); 10.1126/
science.1217876.
3. A. Coventry et al., Nat. Commun. 1, 131 (2010).
4. A. Keinan, A. G. Clark, Science 336, 740 (2012).
5. Y. Li et al., Nat. Genet. 42, 969 (2010).
6. G. T. Marth et al., Genome Biol. 12, R84 (2011).
7. M. Lynch, Proc. Natl. Acad. Sci. U.S.A. 107, 961 (2010).
8. D. G. MacArthur et al., Science 335, 823 (2012).
9. P. M. Visscher et al., Am. J. Hum. Genet. 90, 7 (2012).
10. G. Gibson, Nat. Rev. Genet. 13, 135 (2011).
11. T. A. Manolio et al., Nature 461, 747 (2009).
12. J. Asimit, E. Zeggini, Annu. Rev. Genet. 44, 293 (2010).
13. G. Barbujani, A. Magagni, E. Minch, L. L. Cavalli-Sforza,
Proc. Natl. Acad. Sci. U.S.A. 94, 4516 (1997).
14. S. Gravel et al., Proc. Natl. Acad. Sci. U.S.A. 108, 11983
(2011).
15. I. Mathieson, G. McVean, Nat. Genet. 44, 243 (2011).
Acknowledgment: F.C. holds a BP-DGR fellowship from
AGAUR (Generalitat de Catalunya).
10.1126/science.1224528
PHYSICS
Two Atoms Announce Their
Long-Distance Relationship
The quantum entanglement of two atoms
20 m apart, verified by a secondary process,
can enable long-distance quantum
communications.
Jürgen Volz and Arno Rauschenbeutel
冷Ψ – 冭
Vienna Center for Quantum Science and Technology, Atominstitut, Vienna University of Technology, Stadionallee 2, 1020 Wien, Austria.
E-mail: [email protected]; arno.rauschenbeutel@
tuwien.ac.at
40
Entangled
Atom 1
Atom 2
6 JULY 2012 VOL 337 SCIENCE www.sciencemag.org
Published by AAAS
an entangled two-photon Bell
state like |Ψ–〉 is detected, the
atoms are also projected onto
an entangled state. The success
of the process is signaled by the
“click” of the Bell state detector.
CREDIT: P. HUEY/SCIENCE
E
ntanglement is a counterintuitive communication, the generation of entangle- many advantages, such as the exponential
property of quantum mechanics that ment over macroscopic distances must be speed-up of certain algorithms (2) or the posputs it at odds with our classical view “heralded”—a separate signal must verify sibility of detecting, with certainty, the presof the world. For example, an entangled state that the entangled state was created. On page ence of an eavesdropper in a communicaof two photons can be created such that each 72 of this issue, Hofmann et al. (1) describe tion channel (3). For this purpose, informasingle photon is unpolarized—it will always the accomplishment of this goal with two tion is encoded in “quantum bits” (qubits)—
pass through a polarization analyzer (e.g., a individual atoms held in laser traps and sepa- the basic quantum information carriers that
filter) with 50% probability no matter how rated by a distance of 20 m.
can, in contrast to classical bits, also exist in
the analyzer is oriented. However, once one
Quantum-based communication and a superposition of the logical states “0” and
of the two photons has passed through its information-processing schemes promise “1”. Experiments making use of these effects
analyzer, the second photon becomes
Telling on entangled atoms.
A
Optical fiber
Optical fiber
polarized, and its orientation will then
The experimental sequence
Bell state
Laser
Laser
be orthogonal to the first. Such effects
measurement
developed by Hoffman et al.
pulse
pulse
are not only interesting from a fundafor “heralding” the entanglemental point of view: If the two entanment of distant atoms is illusgled particles are shared between two
trated. (A) Each atom is excited
Photon 1
Photon 2
distant parties, the perfect quantum
by a short laser pulse and emits
a photon that is entangled with
Atom 1
Atom 2
correlations can be used to realize
the final atomic state. These
a so-called quantum channel over
photons are guided through
which quantum information can be
Optical fiber
Optical fiber
optical fibers to a Bell state
B
transmitted. However, to exploit these
Click!
measurement setup. (B) Once
schemes in long-distance quantum