The Global Virome Project

Abstract

Expanded viral discovery can improve mitigation

Full text

sciencemag.org SCIENCE
By Dennis Carroll, Peter Daszak,
Nathan D. Wolfe, George F. Gao,
Carlos M. Morel, Subhash Morzaria,
Ariel Pablos-Méndez, Oyewale Tomori,
Jonna A. K. Mazet
Outbreaks of novel and deadly viruses
highlight global vulnerability to
emerging diseases, with many having
massive health and economic impacts.
Our adaptive toolkit—based largely
on vaccines and therapeutics—is often
ineffective because countermeasure develop-
ment can be outpaced by the speed of novel
viral emergence and spread. Following each
outbreak, the public health community be-
moans a lack of prescience, but after decades
of reacting to each event with little focus
on mitigation, we remain only marginally
better protected against the next epidemic.
Our ability to mitigate disease emergence
is undermined by our poor understanding
of the diversity and ecology of viral threats,
and of the drivers of their emergence. We de-
scribe a Global Virome Project (GVP) aimed
to launch in 2018 that will help identify the
bulk of this viral threat and provide timely
data for public health interventions against
future pandemics.
Nearly all recent pandemics have a viral
etiology with animal origins, and with their
intrinsic capacity for interspecies transmis-
sion, viral zoonoses are prime candidates for
causing the next great pandemic (1, 2). How-
ever, if these viruses are our enemy, we do not
yet know our enemy very well. Around 263
viruses from 25 viral families are known to
infect humans (3) (see the figure), and given
the rate of discovery following identification
of the first human virus (yellow fever virus in
1901), it is likely many more will emerge in
the future (4). We estimate, from analysis of
recent viral discovery data (5), that ~1.67 mil-
lion yet-to-be-discovered viral species from
key zoonotic viral families exist in mammal
and bird hosts—the most important reser-
voirs for viral zoonoses (supplementary text).
By analyzing all known viral-host relation-
ships (3, 6), the history of viral zoonoses (7),
and patterns of viral emergence (1), we can
reasonably expect that between 631,000
and 827,000 of these unknown viruses have
zoonotic potential (supplementary text). We
have no readily available technological coun-
termeasures to these as-yet-undiscovered vi-
ruses. Furthermore, the rate of zoonotic viral
spillover into people is accelerating, mirror-
ing the expansion of our global footprint and
travel networks (1, 8), leading to a nonlinear
rise in pandemic risk and an exponential
growth in their economic impacts (8).
PROMISING PILOT, CHALLENGING SCALE
Since 2009, the U.S. Agency for International
Development (USAID) has conducted a large-
scale pilot project, spanning more than 35
countries over 8 years at a cost of around
$170 million, to evaluate the feasibility of
preemptively mitigating pandemic threats.
Other previous studies had begun to conduct
targeted viral discovery in wildlife (9), and
develop mitigation strategies for the emer-
gence of avian flu, for example. However, the
USAID Emerging Pandemic Threats (EPT)
PREDICT project is the first global-scale co-
ordinated program designed to conduct vi-
ral discovery in wildlife reservoir hosts, and
characterize ecological and socioeconomic
factors that drive their risk of spillover, to
mitigate their emergence in people (10).
Working with local partners and govern-
ments, wildlife and domestic animals and
at-risk human populations in geographic
hotspots of disease emergence (1) are sam-
pled, and viral discovery conducted. A strat-
egy to identify which novel viruses are most
at risk of spillover has been developed (11),
and further work is conducted on these to
characterize them prior to, or in the early
stages of, spillover. Metadata on the ecology
of wildlife–livestock–human transmission in-
terfaces, and on human behavioral patterns
in communities, are concurrently analyzed
so that strategies to reduce spillover can be
developed (supplementary text). To date,
EPT PREDICT has discovered more than
1000 viruses from viral families that con-
tain zoonoses, including viruses involved in
recent outbreaks (12), and others of ongoing
public health concern (13). The focus of EPT
PREDICT on capacity building, infrastruc-
ture support, training, and epidemiological
analysis differs substantially from the GVP’s
emphasis on large-scale sampling and viral
discovery. However, to discover the bulk of
the projected remaining 1.67 million un-
known viruses in animal reservoirs and char-
acterize the majority of 631,000 to 827,000
viruses of highest zoonotic potential requires
overcoming some challenges of scale.
The first challenge is cost. To estimate this,
we analyzed data on field sampling and labo-
ratory expenditures for viral discovery from
(5, 10), and estimates of unknown viral diver-
sity in mammalian and avian hosts (supple-
mentary text). We estimate that discovery of
all viral threats and characterization of their
risk for spillover, using currently available
technologies and protocols, would be ex-
tremely costly at over $7 billion (supplemen-
tary text). However, previous work shows
that viral discovery rates are vastly higher in
the early stages of a sampling program, and
that discovering the last few, rare, viruses is
extremely costly and time-consuming owing
to the number of samples required to find
INFECTIOUS DISEASES
The Global Virome Project
Expanded viral discovery can improve mitigation
The list of author affiliations is provided in the supplementary
materials. Email: daszak@ecohealthalliance.org
POLICY FORUM Scientists prepare to collect a blood sample from
a Rousettus sp. fruit bat in Thailand to test for novel
viruses. The Global Virome Project aims to identify
and characterize the majority of currently unknown
viruses in key wildlife groups, including rodents,
nonhuman primates, and bats.
872 23 FEBRUARY 2018 • VOL 359 ISSUE 6378
PHOTO: © 2018 ECOHEALTH ALLIANCE
DA_0223PolicyForum.indd 872 2/21/18 11:05 AM
Published by AAAS
on March 15, 2018 http://science.sciencemag.org/Downloaded from

SCIENCE sciencemag.org
INSIGHTS
them (5) (supplementary text). We used data
on rates of viral discovery (5) to estimate that
the substantial majority of the viral diversity
from our target zoonotic reservoirs could be
discovered, characterized, and assessed for
viral ecology within a 10-year time frame
for ~$1.2 billion (16% of total costs for 71%
of the virome, considering some fixed costs)
(fig. S1). Those viruses remaining undiscov-
ered will, by the nature of sampling bias to-
ward more common host species, represent
the rarest viruses with least opportunity
for spillover, and therefore reduced public
health risk. Their discovery would require
exponentially greater sampling effort and
funding that could be better spent on coun-
termeasures for the more likely threats (sup-
plementary text).
Stakeholders from Asia, Africa, the
Americas, and Europe, spanning industry,
academia, intergovernmental agencies, non-
governmental organizations (NGOs), and
the private sector, began meeting in 2016
to design a framework for the governance,
management, technical operation, and scope
of the GVP. Key efforts include developing
finance streams; establishing a transparent,
equitable implementation strategy; design-
ing data- and sample-sharing protocols; de-
veloping laboratory and metadata platforms;
targeting of host taxa and sampling sites;
analyzing return on investment; forming
collaborative field and laboratory networks;
developing risk characterization frameworks
for viruses discovered; designing a strategy
to assess and mitigate risk behaviors that
facilitate viral emergence; and planning in-
country capacity building for sustainable
threat mitigation. Funding has been identified
to support an initial administrative hub, and
fieldwork is planned to begin in the first two
countries, China and Thailand, during 2018.
With outputs intended to serve the global
public good, the GVP is developing a transpar-
ent and equitable strategy to share data, viral
samples, and their likely products, including
benefits derived from future development of
medical countermeasures. These build on the
Nagoya Protocol to the Convention on Bio-
logical Diversity and the Pandemic Influenza
Preparedness Framework, negotiated by the
World Health Organization (WHO). The in-
ternational collaborative nature and global
ownership of the project should help leverage
funding from diverse international donors,
including government agencies focused on
national virome projects or on international
development projects in other countries, and
private-sector philanthropic donors focused
on technology and big science.
The diversity of tasks required to conduct
the GVP should reduce the potential for it to
divert funds from current public health pro-
grams. For example, discrete work streams on
targeted sampling of wildlife, on bioinformat-
ics, and on behavioral risk analysis fall within
the focus of current scientific research pro-
grams in a range of donor agencies. Govern-
ments and corporations with specific remits
and geographic responsibilities have been
approached to finance subprojects relevant
to their sectors (e.g., capacity development,
surveillance of specific taxa, geographically
focused activities, medical countermeasure
development, training, surveillance, and
technological platforms). In addition, lead-
ers in China and a number of countries have
begun developing national virome projects as
part of the GVP, leveraging current research
funding to include GVP sites.
Technological challenges include safe field
sampling in remote locations and cost-effec-
tive laboratory platforms that can be stan-
dardized in low-income settings. To achieve
these goals, existing national, regional, and
international networks will need to be en-
hanced and expanded within standardized
sampling and testing frameworks. Existing
networks of field biologists from environ-
ment ministries, academic institutions, and
conservation and health NGOs may assist in
surveillance. National science and technol-
ogy agencies, regional One Health platforms,
transboundary disease surveillance net-
works, Institut Pasteur laboratories, WHO,
United Nations Food and Agricultural Or-
ganization, and the World Organization for
Animal Health collaborating, and reference
centers and viral discovery laboratories, in-
cluding USAID EPT PREDICT, are currently
involved in planning these activities around
a decade-long sampling and testing time
frame. A monitoring and evaluation strat-
egy is being developed based on analysis of
viral discovery rates against predicted viral
diversity, to identify when to halt surveil-
lance and testing as the GVP progresses.
Stakeholders will also tackle the challenge
of how to decide when enough potentially
dangerous viruses have been discovered in a
host species or region to call for action to re-
duce underlying drivers of emergence (e.g.,
hunting and trading of a wildlife reservoir).
Laboratory platforms developed by
USAID EPT PREDICT have proven capacity
to identify novel viruses and are relatively
inexpensive and reliable, being based on
polymerase chain reaction using degenerate
primers that target a range of viral families of
known zoonotic potential. However, scaling
up to a full global virome project will require
discovery of three orders of magnitude more
viruses in a similar time frame. Technologi-
cal solutions will be needed to increase the
speed and efficiency, and reduce the cost,
of sequence generation. These will likely in-
clude next-generation sequencing and other
unbiased approaches to identify evolution-
arily distinct viral clades.
A key challenge is how to assess the po-
tential for novel viruses to infect people or
become pandemic (14). The EPT PREDICT
project (11) and others (2, 6) have developed
preliminary zoonotic risk characterization
frameworks based on viral and host traits and
the ecological and demographic characteris-
tics of the sampling site. These approaches
will be used in the GVP to triage novel viruses
for further characterization to assess their
zoonotic capacity (supplementary text). In vi-
tro receptor binding analyses coupled with in
vivo models have proven useful in this capac-
ity for some viral families [e.g., coronaviruses
(13)]. Although this is not yet feasible for all
potentially zoonotic viral clades, applying
these techniques to a larger viral data set as
the GVP progresses will allow validation of
risk frameworks and may increase our capac-
ity to predict zoonotic potential. However,
advancing these goals will require new col-
laboration among lab virologists, epidemiolo-
gists, and modelers, innovative approaches to
field-testing the boundaries of virus-host re-
lationships, and support across agencies that
often fund separate virology, public health,
evolutionary biology, and biodiversity model-
ing initiatives.
INVESTMENTS, RETURNS
The cost of the GVP represents a sizable
investment and, even if a large number of
potential zoonoses are discovered, only a
minority is likely to have the potential to
cause large-scale outbreaks and mortality in
people (1, 2, 7). However, given the high cost
of single epidemic events, data produced by
the GVP may provide substantial return on
investment by enhancing diagnostic capacity
in the early stages of a new disease outbreak
or by rapidly identifying spillover hosts, for
example. Recent analysis of the exponentially
rising economic damages from increasing
rates of zoonotic disease emergence sug-
gests that strategies to mitigate pandemics
would provide a 10:1 return on investment
(1, 8). Even small reductions in the estimated
costs of a future influenza pandemic (hun-
dreds of billions of dollars) or of the previous
SARS (severe acute respiratory syndrome)
epidemic ($10 billion to $30 billion) could be
substantial. The goal of the GVP is to improve
efficiency in the face of these increasing viral
“...the GVP goals…improve
capacity to detect, diagnose,
and discover viruses in
vulnerable populations…”
23 FEBRUARY 2018 • VOL 359 ISSUE 6378 873
DA_0223PolicyForum.indd 873 2/21/18 11:05 AM
Published by AAAS
on March 15, 2018 http://science.sciencemag.org/Downloaded from

sciencemag.org SCIENCE
GRAPHIC: V. ALTOUNIAN/SCIENCE
INSIGHTS |
POLICY FORUM
spillover rates by enhancing (not replacing)
current pandemic surveillance, prevention,
and control strategies. If we were to invest
only in surveillance for known pathogens
(our current business-as-usual strategy), our
calculations suggest we would protect our-
selves against less than 0.1% of those viruses
that could conceivably infect people, even us-
ing the lower bounds of our uncertainty for
our viral estimates (i.e., 263 viruses known
from humans out of 263,824 unknown poten-
tial zoonoses; supplementary text).
The potential benefits of the GVP may be
enhanced to maximize public health benefits
(supplementary text) by (i) optimizing sam-
pling to target species most likely to harbor
“missing zoonoses” (6), or to target emerging
disease hotspot regions most likely to propa-
gate major disease outbreaks (1); (ii) using
human and livestock syndromic surveillance
to identify regions for wildlife sampling prox-
imal to repeated outbreaks of severe influ-
enza-like-illnesses, fevers of unknown origin,
encephalitides, livestock “abortion storms,”
and other potential emerging disease events;
(iii) initially targeting RNA viruses, which
caused 94% of the zoonoses documented
from 1990 to 2010; and (iv) fostering econo-
mies of scale and adoption of technological
innovation as the GVP ramps up. This in-
cludes use of laboratories that can facilitate
regional sample processing, development of
centralized bioinformatics platforms, and
improved logistics for sample collection and
transport. We also expect the cost of testing
and sequencing to decrease as technology
is enhanced, much as the development of
next-generation sequencing reduced genetic
sequencing costs by up to four orders of mag-
nitude in a decade.
The accelerated pace of viral discovery un-
der the GVP will make the virological, phy-
logenetic, and modeling approaches used in
pandemic preparedness more data-rich, and
likely more effective. For example, having the
sequence data for thousands, rather than a
few, viruses from a single family could extend
vaccine, therapeutic, or drug development to
a wider range of targets, leading to broad-
spectrum vaccines and other countermea-
sures. Identification of novel viruses may be
useful to programs like the Coalition for Epi-
demic Preparedness Innovations (CEPI) in
assessing the breadth of action of candidate
vaccines and therapeutics, and in expand-
ing their efficacy. More broad-scale preven-
tion approaches could provide immediate
return on investment prior to vaccine and
countermeasure development, which would
require substantial investment and time. For
example, metadata on viral reservoir host
identity, geography, seasonality, proximity to
people, and drivers of emergence will refine
our mechanistic understanding of spillover
and enhance published models of emerging
infectious diseases risk (1, 6). Identification
of novel viruses in hunted, traded, or farmed
wildlife species could be used to enhance
bio security in markets and farming systems,
reducing public health risk, increasing food
security, and assisting in conservation of
hunted species. The presence of hosts har-
boring high-risk novel viruses in proximity to
human populations may allow targeted fol-
low-up to examine evidence of spillover and
design intervention strategies (supplemen-
tary text). Ultimately, the benefits of the GVP
may include enhancing our understanding of
viral biology, such as drivers of competition
or cooperation among viruses within hosts,
genomic underpinnings of host-virus coevo-
lution, processes underlying deep evolution
of viral clades, and the identification of novel
viral groups (15).
The regions targeted by the GVP are
largely highly biodiverse, rapidly developing
countries in the tropics, which often have low
capacity to deal with public health crises (1).
The expanded laboratory capacity, field sam-
pling, and data generation intrinsic to the
GVP goals will therefore improve capacity
to detect, diagnose, and discover viruses in
vulnerable populations within regions most
critical to preventing future pandemics. This
enhanced capacity may also help improve
diagnosis and control for endemic diseases,
as well as the portion of the virome that re-
mains undiscovered.
The Human Genome Project in the 1980s
catalyzed technological innovation that dra-
matically shortened the time and cost for its
completion, and ushered in the era of per-
sonalized genomics and precision medicine.
The GVP will likely accelerate development
of pathogen discovery technology, diagnos-
tic tests, and science-based mitigation strat-
egies, which may also provide unexpected
benefits. Like the Human Genome Project,
the GVP will provide a wealth of publicly
accessible data, potentially leading to dis-
coveries that are hard to anticipate, per-
haps viruses that cause cancers and chronic
physiological, mental health, or behavioral
disorders. It will provide orders-of-magni-
tude more information about future threats
to global health and biosecurity, improve
our ability to identify vulnerable popula-
tions, and enable us to more precisely target
mitigation and control measures to foster
an era of global pandemic prevention. j
REFERENCES AND NOTES
1. K. E. Jones et al., Nature 451, 990 (2008).
2. J. L. Geoghegan, A. M. Senior, F. Di Giallonardo, E. C.
Holmes, Proc. Natl. Acad. Sci. U.S.A. 113, 4170 (2016).
3. A. M. K ing, M. J. Ada ms, E. B. C arsten s, E. J. Lefkow itz, Vir us
Taxonomy: Classification and Nomenclature of Viruses:
Ninth Report of the International Committee on Taxonomy
of Viruses (Elsevier, Philadelphia, 2012).
4. M. Woolhouse, F. Scott, Z. Hudson, R. Howey, M. Chase-
Topping, Philos. Trans. R. Soc. Lond. B Biol. Sci. 367, 2864
(2012).
5. S. J . Anth ony et a l., mBio 4, e00598-13 (2013).
6. K. J . Oli val et al., Nature 546, 646 (2017).
7. N. D. Wolfe, C . P. Dun avan, J. Di amond , Nature 447, 279
(2007).
8. J. Pike , T. L. Bog ich, S. E lwood , D. C. Finno ff, P. Daszak,
Proc. Natl. Acad. Sci. U.S.A. 111, 18519 (2014).
9. J. F. Drexler et al., Nat. Commun. 3, 796 (2012).
10. data.predict.global
11 . S. S. Morse et al., Lancet 380, 1956 (2012).
12. G. Gr ard et al., PLOS Pathogens8, e1002924 (2012).
13 . X . -Y. G e et al., Nature 503, 535 (2013).
14. J. L. G eoghe gan, E. C. H olmes , Open Biol. 7, 170189 (2017).
15. C. X. Li et al., eLife 4, e05378 (2015).
ACKNOWLEDGMENTS
P.D., N.D.W., and J.A.K.M. are funded by USAID EPT PREDICT.
We ackn owled ge S. J. Ant hony, C. J. Chr isma n, Y. Feferholt z, T.
Gold stein , C. K. John son, D. Na barro, K . J. Olival , N. Ross , E. Rubi n,
R. Waldman, B. Watson, C. Zambrana-Torrelio, attendees of the
Rockefeller Foundation–funded Bellagio Center Global Virome
Project Workshop August 2016 (www.globalviromeproj ect.org/
about/), members of the GVP Core Group and Steering
Committee, and co-leads of the GVP Working Groups for their
help refining the concept and this manuscript.
SUPPLEMENTARY MATERIALS
www.sciencemag.org/content/359/6378/872/suppl/DC1
10.1126/science.aap7463
Of these 111 viral families,
the GVP will target 25 containing
viruses known to infect
(or to have substantial risk
of infecting) people.
Of these 1.67 million viruses, an
estimated
631,00 to 827,000
likely have the capacity to infect
people.
In these 25 families, an estimated
1.67 million unknown viruses
exist in mammals and
birds—hosts that represent 99%
of the risk for viral emergence.
111 viral families have been
discovered globally to date.
874 23 FEBRUARY 2018 • VOL 359 ISSUE 6378
GVP targeting strategy
The project will capitalize on economies of scale
in viral testing, systematically sampling
mammals and birds to identify currently unknown,
potentially zoonotic viruses that they carry.
DA_0223PolicyForum.indd 874 2/21/18 11:05 AM
Published by AAAS
on March 15, 2018 http://science.sciencemag.org/Downloaded from

The Global Virome Project
Oyewale Tomori and Jonna A. K. Mazet
Dennis Carroll, Peter Daszak, Nathan D. Wolfe, George F. Gao, Carlos M. Morel, Subhash Morzaria, Ariel Pablos-Méndez,
DOI: 10.1126/science.aap7463
(6378), 872-874.359Science
ARTICLE TOOLS http://science.sciencemag.org/content/359/6378/872
MATERIALS
SUPPLEMENTARY http://science.sciencemag.org/content/suppl/2018/02/21/359.6378.872.DC1
REFERENCES http://science.sciencemag.org/content/359/6378/872#BIBL
This article cites 13 articles, 5 of which you can access for free
PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions
Terms of ServiceUse of this article is subject to the
is a registered trademark of AAAS.Science
licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. The title
Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights reserved; exclusive
(print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience
on March 15, 2018 http://science.sciencemag.org/Downloaded from