22 4 JANUARY 2019 • VOL 363 ISSUE 6422 sciencemag.org SCIENCE
Kelso is one of many researchers who are
turning troves of genetic and medical data on
living people into windows on human evolu-
tion. In addition to unearthing archaic DNA,
the studies are pinpointing genes that natu-
ral selection may now be winnowing out of
the gene pool and other genes—for example
those linked to fertility—that it may be fa-
voring. Among the most fruitful of the data
sources is the UKB, which makes its data ac-
cessible to researchers, no matter where they
are and what their field. Its giant database
is “a magical new resource that [will] help
us answer a whole bunch of hard questions
we’re struggling with now because all of the
data has been under lock and key,” says
population geneticist Jeremy Berg, a
postdoc at Columbia University. “It is a
step beyond other databases.”
For the UKB architects, who designed
it for biomedical research, the evolu-
tionary discoveries are an unexpected
bonus. “No one was thinking about
Neanderthal traits when we designed
the protocol,” says molecular epidemio-
logist Rory Collins of the University of
Oxford in the United Kingdom, who is
principal investigator of the UKB. “The
experiment [is] working well beyond
people’s expectations.”
NEANDERTHALS SNEAKED INTO the UKB
in 2013, when Harvard University pop-
ulation geneticist David Reich was in
Oxford to give a talk. His host, Oxford
geneticist Peter Donnelly, was oversee-
ing the design of chips to identify genes
of interest in blood samples like those
in the UKB. Donnelly asked Reich
whether he’d like to add Neanderthal
variants to a custom chip used to geno-
type the UKB participants; that would allow
Reich and others to fish for rare Neanderthal
variants in half a million people. “David was
very enthusiastic,” Donnelly recalls.
Soon after, Reich and his postdoc, Sriram
Sankararaman, emailed Donnelly a wish list
of variants to add to the chip: 6000 relatively
rare alleles likely to come from Neander-
thals. Their calculations suggested the UKB
was big enough to include enough carriers
of these variants so researchers could probe
the function of the genes. “Imagine 1% of the
population has a Neanderthal variant,” says
Sankararaman, now a computational geneti-
cist at the University of California (UC), Los
Angeles. “If you’re looking at half a million
people, you’re looking at enough copies of
that variant in enough individuals [5000] so
you can detect subtle effects.”
At the same time, computational bio-
logist Tony Capra at Vanderbilt University in
Nashville had the same bright idea to search
for Neanderthal DNA in a large database. He
used proprietary electronic records of 28,
Americans. His team was the first to publish,
reporting Neanderthal DNA variants that
raise the risk of depression, skin lesions,
blood clots, and other disorders in people
today (Science, 12 February 2016, p. 648). In-
spired by Capra’s study, Kelso jumped in, be-
coming the first to use UKB data to publish
Neanderthal gene variants in living people.
Her results suggest that although some Ne-
anderthal gene variants may have been op-
timal for active lives outdoors in prehistoric
Europe, they may be problematic for people
now, who live mostly indoors in artificial
light and get less exercise.Groups led by Kelso and Sankararaman
are now looking for links between Nean-
derthal DNA and traits in genotyped data
from 500,000 people—the total UKB data
set, which was released in July 2017. Al-
ready, they are learning that Neanderthal
alleles help cause baldness and mental ill-
ness and boost certain immune functions,
Sankararaman says. Meanwhile, another
team has found variants that help explain
why modern humans’ heads are round, in
contrast to the elongated, football-like shape
of Neanderthal skulls (Science, 14 Decem-
ber 2018, p. 1229). Those researchers plan
to combine forthcoming MRI brain scans
of 100,000 UKB participants with genetic
data to probe the genetic basis of brain dif-
ferences between us and our extinct cousins.
Capra says when it comes to scanning
and understanding DNA from Neander-
thals, the UKB cohort offers even more ana-
lytical power than the medical databases he
used, because it covers “a broader range ofpsychiatric and lifestyle traits.” Those rich
data have also made the UKB a hunting
ground for clues to evolutionary changes
that have shaped people’s genomes in the
past few generations—and may even be do-
ing so today.A FEW YEARS AGO, Molly Przeworski of Co-
lumbia University and Joe Pickrell of the
New York Genome Center in New York City
met for lunch near Columbia’s campus. Talk
turned to aging and Alzheimer’s disease.
Pickrell had been writing a blog, where he
had discussed studies showing that between
the ages of 70 and 85, carriers of the ApoE
allele, which boosts the risk of Al-
zheimer’s and cardiovascular disease,
died at about twice the rate of non-
carriers. The pair wondered whether
other gene variants affect survival so
dramatically—and whether natural
selection is weeding them out.
When it comes to natural selection
in humans, most studies have only
been able to detect dramatic cases
thousands or millions of years ago
in genes of known function. Now,
Pickrell and Przeworski wondered
whether they could detect genetic vari-
ants that affect survival today—and
whether natural selection in recent gen-
erations has been weeding out harmful
ones or favoring beneficial ones.
To do this, they realized they’d need
data on DNA as well as on traits like
participants’ age at death. For statis-
tical confidence, they’d need a giant
sample size—at least 100,000—to de-
tect how the frequency of common al-
leles varied in people of different ages.
Databases like the UKB were the an-
swer. “We suddenly realized that the some of
these databases were large enough to let us
study selection in contemporary humans,”
Przeworski says.
They soon got access to genetic and health
data on 57,696 people in the Resource for
Genetic Epidemiology Research on Aging
database at Kaiser Permanente in Oakland,
California, and 117,648 individuals in the
UKB’s 2015 data release. They sorted partici-
pants into 5-year age intervals, and looked
at the frequency of many alleles, including
ApoE4 , in each age group, as well as how the
variants correlated with 42 traits potentially
associated with early death or long life, such
as cardiovascular disease, cholesterol levels,
asthma, age at puberty, and menopause.
Nearly all the variants they examined
persisted at the same frequency even into
old age, suggesting they had no large ef-
fect on survival. That implies natural se-
lection has efficiently weeded out harmful
variants, even if they act only in old age— PHOTO: RONNY BARR/MAX PLANCK INSTITUTE FOR EVOLUTIONARY ANTHROPOLOGYNEWS | FEATURES
Janet Kelso f shed for Neanderthal gene variants in the UK Biobank.Published by AAASon January 3, 2019^http://science.sciencemag.org/Downloaded from