The open-source model pioneered by the software industry has potential applications to drug research and development. Open-source initiatives allow scientists to collaborate freely across organizations to solve shared problems. While there are significant differences between software and drug development, public-private partnerships have adapted open-source principles by crowdsourcing research through an outsourced network. These partnerships operate on small budgets by vetting proposals and outsourcing research, pursuing projects from early discovery through clinical trials. Over 20 such partnerships have been established and have attracted over $1.5 billion in funding to develop treatments for neglected diseases.

1. Nature Reviews Drug Discovery | AOP, published online 18 August 2006; doi:10.1038/nrd2131
PERSPECTIVES
Subversion help over a million people
OUTLOOK
collaborate on more than 100,000 projects.
Can open-source R&D reinvigorate But other areas, such as life sciences, have
spawned open-source initiatives of their own.
The impetus to create open-source soft-
drug research? ware often comes from developers looking for
challenge. They agree on an attractive project,
Bernard Munos form a team and produce a ‘bare-bones’ pro-
gram with basic functionality. Then, they offer
Abstract | The low number of novel therapeutics approved by the US FDA in recent it at no cost on the Internet under a public-
years continues to cause great concern about productivity and declining innovation. domain license (there are many different
Can open-source drug research and development, using principles pioneered by types of open-source license; some, notably
the ‘copyleft’ or General Public License (GPL),
the highly successful open-source software movement, help revive the industry?
require those who download a program to
share any improvements they make). If the
Open-source research, which started as a involved? What are the better targets? Once project draws interest, others add features and
counterculture movement in the software these questions are answered, laboratory and post their code on the project’s webpage for
industry 15 years ago, has since grown into a clinical studies can be outsourced to institu- fellow programmers to critique. New code of
business model whose best-known product, tions with the requisite capacity through the sufficient quality is added to the authorized
Linux, has become a credible alternative to help of matchmaking software. version of the program.
Microsoft’s Windows. Now, with biology The resulting model is a hybrid in which Open-source’s chief benefit is to cross-
increasingly becoming an information- a part of R&D is open-sourced while the fertilize minds and tap creativity quickly,
orientated science, some have suggested rest is outsourced. To function, however, cheaply and on a scale that is beyond the reach
that what worked for software might be part it needs strong project leadership and of scientists working in the ‘ivory towers’ of
of the answer to the spiralling cost of drug expertise in the minutia of drug R&D, academia or behind the ‘corporate moats’
R&D. With this in mind, this article exam- which mostly exist in big pharmaceutical of industry. Hollingsworth1,2 has shown that
ines the relevance to pharmaceutical R&D firms. This suggests that, far from being innovation spikes when diverse minds interact
of the open-source model developed by the a threat to conventional drug R&D, frequently in an unstructured manner. By
software industry. In this context, open- open-source could be a way to leverage drawing talent from all around the world,
source no longer refers to source code, but big pharma’s capabilities in order to tackle open-source research takes these dynamics
instead to the open origin of contributors. challenges that the blockbuster model can- to a new scale. And by making innovation
Open-source R&D has already made not address economically, such as neglected immediately available to all, it speeds up the
inroads into bioinformatics and research diseases. As pharmacogenomics takes hold, accumulation and application of knowledge.
tools for drug hunters. However, key it might also be a way to address market Outsiders are often puzzled by the open-
differences between software and biology, niches that cannot support blockbusters. source idea. Why would anyone work for
such as regulatory requirements, have free? Simply put, because some people value
limited its application to drug development. A brief primer on open-source non-cash compensation more than money.
Nevertheless, in the past 5 years a new Open-source R&D is a novel approach to They volunteer their expertise to satisfy ideal-
breed of organizations called public–private research that lets scientists join hands freely ism or curiosity, seek new challenges, hone
partnerships (PPPs) have adapted the across organizations, disciplines and borders skills, build a reputation or enhance careers.
open-source concept and combined it with to solve problems in which they share an Feldman3 quotes the example of Australian
outsourcing to create a new, low-cost interest. The movement’s icon is Linux, the programmers who, within hours of Netscape’s
business model, which they have applied operating system started in the early 1990s by release of its browser code, attached an
with encouraging results to the discovery student Linus Torvalds, who used the nascent ‘add-on’ to enable secure internet transactions.
of new treatments for neglected diseases. Internet to circulate it to fellow computer No money changed hands, but the authors
Advances in data mining, visualization enthusiasts. Soon they were busy adding fea- received respect from the programming com-
and networking now make it feasible to go tures and improving the code, with Torvalds munity and the satisfaction of turning out an
one step further. It is possible to offer scien- overseeing the process. Fifteen years later, elegant and useful piece of software.
tists a computerized toolbox that lets them this grassroots experiment has blossomed Companies are learning to use open-
harness the creativity of numerous volunteers into a new culture that is spreading to other source to their advantage, and many now
to address the key questions that are holding disciplines. It is most prominent in computer allow their employees to participate on com-
back innovation. For example, what is the software development, for which dedicated pany time. They might use it to gain market
aetiology of a disease? What are the pathways websites such as such as SourceForge or share against entrenched competitors, or
NATURE REVIEWS | DRUG DISCOVERY A DVA NCE ON L I N E PU BL ICAT ION | 1

2. PE R S PEC T I V E S
Exploratory Discovery Preclinical Clinical development the achievements of others, and errors can
be patched without requiring the rewrite of
Lead Lead
identification optimization
Transition Phase I Phase II Phase III the whole program. With drugs, one care-
less worker can compromise years of work
costing tens of million of dollars.
OZ + PQP Chlorproguanil-
Isoquine Finally, the two industries follow different
PSAC Dihydrofolate Novel RBx11160/ dapsone
(improved
antagonist reductase macrolides
aminoquinoline)
OZ277 (Lapdap) intellectual property regimes. Software is
+ piperaquine -artesunate (CDA)
protected by copyrights that arise automati-
Pf enoyl-ACP New 4(1H)- cally as code is written, even if nothing is
AQ-13 new Paediatric
reductase dicationic pyridones
(Fab i) molecules Backups
aminoquinoline coartem filed. Drug research is protected by patents
that are costly to file and maintain, and
Pf protein for which meeting the legal standards that
Falcipain
Cyclofarnesyl farnesyl- Pyronaridine–
(cysteine define innovation is much harder.
sequiterpenes transferase artesunate
protease)
(Pf-PFT)
Next Entantio- Open-source biomedical research
generation selective
8-amino- EuArtekin (dihydroartemisinin–piperaquine) Early efforts. Despite these differences, the
antimalarials
quinolines open-source idea has entered biomedical
Novel
research6. The first inroads were made in bio-
imidazolidine informatics7,8 , as might have been expected.
-diones These efforts resulted in a collection of pro-
grams such as Biojava, BioPerl, BioPython,
MMV active support ended MMV/GSK portfolio New projects to be added
Bio-SPICE, BioRuby and Simple Molecular
Figure 1 | Portfolio of the Medicines for Malaria Venture. PSAC, plasmodial surface anion Mechanics for Proteins9, and inspired other
channel. initiatives such as the Human Genome
Project, the SNP Consortium, the Alliance
to entice developers to create applications has been accumulated. That knowledge for Cellular Signaling, BioForge, GMOD
for their product, possibly in the hope of acquisition can take years or decades, with no and Massachusetts Institute of Technology’s
turning it into a ‘platform’. Some of them way to know at the outset whether the store BioBricks (some of these have the transpar-
have been quite successful at turning open- of knowledge at hand is nearly sufficient or ency and feel of open-source, although the
source into profits. Red Hat, for instance, will require years of painstaking additional resources needed to get involved do not allow
has attained a US$5-billion market cap from research before innovation can thrive. all volunteers to participate; however, we still
selling support services for Linux. Software development is also simpler: call them ‘open-source’).
it spans only a few disciplines and has no
Can it work for drugs? equivalent to clinical trials. For the most An old idea. One could argue that there has
If biomedical scientists could adapt the part, a single programmer can master all the long been an active, if invisible, collabora-
open-source model, it could make a huge skills needed to write a program from start tive process akin to open-source in drug
difference to such projects as developing to finish. By contrast, drug development development, as, for some diseases, half
drugs for neglected diseases, for which requires coordination of multiple specialties of all prescriptions are for off-label uses10 .
needs are great but funds are scarce4 . Only with little overlap. Biomedical knowledge, Somehow, physicians share their ideas and
10% of R&D resources are spent on illnesses which grows at the rate of 1,000 publications experiences informally to uncover novel
that represent 90% of the burden of disease. per day, must be peer-reviewed and repli- uses for existing medicines. For instance,
Open-source drug R&D might not change cated before it is accepted. All this is slow oncologists routinely use drugs approved for
that equation, but could make it possible and enormously expensive. one kind of cancer to treat other types. In a
to get much more from that 10%. Drug R&D can go off-track more easily recent study, DeMonaco11 found that 59% of
There are, however, significant barriers to than software programming. Biologists drug therapy innovations were discovered
the deployment of open-source approaches can get mired in the complexity of biology by practicing clinicians via field discovery.
to drug R&D5. One is economic. All it takes without ever making much progress towards The way by which physicians uncover these
to write open-source software is a laptop and a drug — chemists handed the wrong target new indications is quick and inexpensive
an internet connection. With drug research, cannot do much good no matter how hard compared with Phase III trials. From an eco-
someone must pay for laboratory expenses they try; inadequate toxicology can derail nomic and medical standpoint, there would
and clinical trials. And the costs are high, at a compound late in development, or even be merit in exploiting these clinical observa-
more than US$800 million for the discovery after launch. One misstep along the way can tions and sharing them with physicians as a
and development of a novel drug by most render all downstream work useless. complement to, or replacement for, some of
estimates. In contrast to drug developers, software the traditional clinical development.
Research dynamics between the two indus- publishers are lightly regulated. They do
tries also differ. Software development does not need FDA approval. The quality stand- Public–private partnerships. Taking a different
not have a discovery phase. Once the objective ards they face are far less onerous than the approach, a new kind of organization, known
is set, programmers set to work and make minutia of Good Laboratory Practice as the public–private partnership (PPP), has
steady progress towards their goal. By contrast, (GLP), Good Clinical Practice (GCP) and recently developed a clever virtual business
drug discovery cannot flourish until a certain Good Manufacturing Practice (GMP). model that emulates the collaborative features
amount of knowledge about the target disease One sloppy programmer seldom jeopardizes of the open-source concept12 . An example is
2 | A DVA NCE ON L I N E PU BL ICAT ION www.nature.com/reviews/drugdisc

3. PE R S PEC T I V E S
Box 1 | The Medicines for Malaria Venture
markets (for example, to treat travellers).
Its alliance with GlaxoSmithKline supports
25 scientists funded equally by the partners.
CEO
The Initiative on Public–Private
Partnerships for Health reckons that there
Chief Finance Officer and are about 24 PPPs engaged in drug and
Administrative Assistant
Donor Relations vaccine R&D (TABLE 1). Most of them were
created in the past 7 years and share a com-
Communication
Human Resources Officer mon profile13. First, they focus on neglected
and Advocacy Officer diseases. Second, they operate as virtual drug
companies, with a small staff getting project
Chief Scientific Officer Director, International Ops ideas from outside, vetting them through a
committee of experts and outsourcing R&D
to a network of institutions. Third, they man-
Director, Drug Discovery Director, Clinical age growing portfolios of projects ranging
and Technology Development from discovery through to Phase III trials.
Fourth, they have been able to function on
Contracts Officer Director, Clinical lean budgets with a cumulative spending
Development that seldom exceeds US$50 million. This
makes them attractive vehicles to fund
Associate Director, research in areas that are not economical for
Administrative Assistant
Clinical Development
traditional drug R&D.
Management team By the end of 2005, PPPs had attracted
The Medicines for Malaria Venture (MMV) is run by a staff of 13. Its CEO reports to a Board of 12 funding in excess of US$1.5 billion.
Directors who represent funding organizations. An Expert Scientific Advisory Committee, which Foundations have given about US$1.15 billion
includes chemists, biologists, clinicians, malariologists and drug development experts, advises on (with the Gates Foundation alone contributing
project selection and research strategy. The Management Team’s responsibilities are to: more than US$950 million), governments
• Encourage the submission of research proposals US$244 million and private entities US$36
• Select proposals and negotiate with partners million. In addition, donors have committed
• Set up project-management teams and monitor progress
another US$3.5 billion which will be
• Organize manufacturing and marketing
• Earn appropriate returns from marketed products
disbursed as needed by The Global Fund to
• Raise funds Fight AIDS, Tuberculosis and Malaria.
• Communicate with government agencies
Backers
Open-source versus alliance networks. It can
• Gates Foundation be argued that the 25,000 alliances wrought
• BHP Billiton by the 8,000 pharma and biotech companies
• ExxonMobil over the past 15 years add up to a vast open-
• Global Forum for Health Research innovation system that mimics the collabo-
• International Federation of Pharmaceutical Manufacturers Associations rative features of the open-source model.
• Netherlands Minister for Development Cooperation Some scholars have countered, however, that
• Rockefeller Foundation alliances are less effective than open-source
• Swiss Agency for Development and Cooperation research at promoting innovation. This is
• United Kingdom Department for International Development
because open-source networks are richer in
• United States Agency for International Development
• World Bank
‘weak links’ (loose relationships), whereas
• Wellcome Trust alliances pride themselves on the strength
• World Health Organization of the connections between partners.
DeBresson14 has shown that weak links bring
novel ideas into the fray whereas strong links
the Medicines for Malaria Venture (MMV), R&D to a network of 300 scientists at 40 insti- tend to reinforce orthodoxies.
which was established in 1999 to discover and tutions (universities, big pharma, biotechs
develop new, affordable antimalarial drugs. and research institutes). Funding comes from PPPs and big pharmas. TABLE 2 lists some of
Established as a nonprofit entity with a staff public and philanthropic partners (BOX 1). the projects and organizations coordinated
of only 13 people, it has assembled a portfolio After each step, the Scientific Advisory by PPPs. As can be seen, GlaxoSmithKline
of 19 projects ranging from discovery to Committee reviews the data and decides features prominently, with Bristol-Myers
Phase III (FIG. 1). whether to proceed or terminate the project. Squibb, Novartis, Bayer, Sanofi-Aventis and
MMV gets its projects through open MMV’s cumulative spend from 2000 through Ranbaxy involved to a lesser extent.
calls — anyone with an idea can contribute. 2005 is about US$100 million, 90% of which
An Expert Scientific Advisory Committee funded actual research. MMV plans to out- Lessons learned
reviews the submissions and selects the source manufacturing to low-cost partners, PPPs have advantages and drawbacks
projects that will be funded. Each is managed sell drugs at cost to developing countries, and compared with traditional R&D15. Advantages
by a project manager who outsources the market them through partners in developed include the following.
NATURE REVIEWS | DRUG DISCOVERY A DVA NCE ON L I N E PU BL ICAT ION | 3

4. PE R S PEC T I V E S
Table 1 | Public–private partnerships engaged in drug and vaccine development Speed. Lean PPPs can decide quickly, partly
because they do not have layers of committees
Name Focus Year created
to satisfy. In addition, because they tap their
Aeras Global TB Vaccine Foundation Tuberculosis 1997 partner’s unused capacity, they can advance
BIO Ventures for Global Health Biotech drugs for neglected 2004 swiftly as there is often a qualified laboratory
diseases somewhere that can do the work without
Consortium for Industrial Collaboration in Development of new 1995 having to wait in someone else’s queue.
Contraceptive Research contraceptives
Contraceptive Research and Development Improving reproductive health in 1986 There are also some disadvantages to PPPs,
developing countries which include the following.
Dengue Vaccine Project Dengue fever 1989
Funding. US$5 billion has been committed
Drugs for Neglected Diseases Initiative Sleeping sickness, visceral 2003
leishmaniasis, Chagas disease
to PPPs ($1.5 billion disbursed). However,
despite the thriftiness of PPPs, there is
European Malaria Vaccine Initiative Malaria 1998 concern that these funds will be stretched
Gates Foundation–UNC Partnership for African trypanosomiasis, 2000 as more projects move into late, expensive
Development of New Drugs leishmaniasis clinical development.
Global Alliance for TB Tuberculosis 2000
Global Microbicide Project New microcides for women 2000 Sustainability. PPPs have not demon-
strated the sustainability of their business
Human Hookworm Vaccine Initiative Hookworm 2000
model. Some of their projects come from
Infectious Disease Research Institute Tuberculosis, leishmaniasis, 1993 companies that had shelved them because
Chagas disease, malaria, leprosy of insufficient commercial prospects. To
and Buruli ulcer
survive, PPPs will need to replenish their
Institute for One World Health Visceral leishmaniasis, cutaneous 2000 portfolios. There are also worries that, in
leishmaniasis, Chagas disease,
paediatric secretory diarrhoea some areas of science, the pool of contribu-
tors might be too thin to perform the work
International AIDS Vaccine Initiative AIDS 1996
that must be done.
International Partnership for Microbicides HIV 2002
Japanese Pharmaceutical, Ministry of Malaria 1999 TABLE 3 shows that the PPP R&D model
Health, WHO Malaria Drug Partnership has worked reasonably well. Some of this
Lapdap Antimalarial Product Development Malaria 1998 success comes from targeting low-hanging
fruits in diseases that have long been
Lassa Fever Initiative Lassa fever 2001
neglected, but it also suggests that the PPP
Malaria Vaccine Initiative Malaria 1999 model can be a potent tool in finding new
Medicines for Malaria Venture Malaria 1999 cures. Whether the PPP business model
Meningitis Meningitis 2001 becomes a transformational force or
remains a non-threatening niche depends
Microbicides Development Programme HIV 2001
on how it ultimately performs against
Pediatric Dengue Vaccine Initiative Dengue 2001 traditional pharmaceutical R&D. To
PneumoADIP Pneumococcal vaccines 2004 succeed, it must go beyond tools and soft-
UNC, University of North Carolina; WHO, World Health Organization. ware and tackle large projects where it will
rival the big firms that are helping it today.
Yet, this rivalry need not be a zero-sum
Agility. Virtual R&D makes it easier to Risk sharing. The open-innovation game. On the contrary, there is a place for
terminate projects that no longer look model of PPPs makes it easier for scientists collaborative and proprietary research in
promising. The project manager does not to collaborate on pre-commercial research drug R&D, just as in software16 . If open-
have to deal with entrenched advocates such as biomarkers or cell signalling. source drug R&D takes hold, what will
manoeuvering to save their project or move probably emerge is not the replacement of
it underground. Affordability. PPPs lower the critical mass one model by another, but an ecology in
required to be a pharmaceutical company. By which big pharma, biotech and collabora-
Creativity. PPPs enable experts from dif- leveraging external expertise and capabilities, tive research compete and collaborate at
ferent countries, specialties and styles of they allow small organizations to do much of the same time, feeding off each other
thought to leverage each other’s ideas. They what was once the domain of large companies. synergistically, while moving towards
harness the problem-solving skills of a much therapies along their own distinctive paths.
greater population than is typically available Impact. PPPs engage scientists in developing
to traditional research organizations. nations who have first-hand experience in A template for open-source drug R&D
many neglected diseases. It helps them build Can the PPP model succeed beyond
Focus. PPPs focus on one or few diseases. their clinical research capacity, which in turn neglected diseases? To answer this, it helps to
This helps them build deep expertise for better leverages the effectiveness of their public break down drug R&D into knowledge-based
decisions (for example, target selection). health systems. activities and rule-based tasks.
4 | A DVA NCE ON L I N E PU BL ICAT ION www.nature.com/reviews/drugdisc

5. PE R S PEC T I V E S
Knowledge-based work requires lots Table 2 | Public–private partnerships and their partners
of intelligence and intuition, but little
infrastructure. Examples include identifying Project Industry partner University/public Other PPP
health partner
targets, understanding metabolic networks,
and designing clinical trials or computer- Medicines for Malaria Venture
ized disease models. It is about scientists Improved 4-aminoquinoline GlaxoSmithKline University of Liverpool
leveraging each other’s ideas, and using Farnesyl transferase inhibitors Bristol-Myers Squibb University of Washington
tools to gain deeper insights that might lead
Manzamine derivatives University of Mississipi
to breakthroughs. This work is ideally
suited to the open-source model. Cysteine protease inhibitors GlaxoSmithKline UCSF
Rule-based work requires physical assets Fatty acid biosynthesis Texas A&M, A. Einstein,
(laboratories, equipment, patients and so on) inhibition Jacobus
and money. It is tightly scripted and must Pyridone GlaxoSmithKline
conform to rigid regulatory requirements.
New di-cationic molecules Immtech University of North
It is about organization, discipline and Carolina
implementation. Examples include toxicol-
Dihydrofolate reductase Biotec Thailand
ogy studies, Chemistry Manufacturing and inhibition
Controls (CMC) studies, and the conduct
of clinical trials. Rule-based work is ideally Artesunate derivatives GlaxoSmithKline, TDR, DNDi
Shin Poong
suited to outsourcing, and much of it is
already outsourced to contract research Artemisone Bayer University of Hong Kong
organizations. Synthetic peroxide Ranbaxy University of Nebraska
This division of labour suggests a busi- Intravenous artesunate Walter Reed
ness model template in which part of the
Coartem in infants Novartis TDR
R&D value chain is open-sourced, while
the rest is outsourced, with the following TB Alliance
features. Pyridones and quinolizines Taejon, Yonsei
Isoniazid analogue Wellesley College
Template features: operating principles
PA-824 NIH, Johns Hopkins
Open-sourcing. The open-source part of
our model should allow anyone who can Mocifloxaxin Bayer CDC, Johns Hopkins
contribute to join. Volunteers should be Institute for One World Health
able to log on to a website, find the page(s) Paromomycin TDR
that matches their area of expertise, peruse
challenges to be solved, review others’ con- Azole Yale
tributions, download computerized tools TDR
and start working towards contributions of Miltefosine Zentaris
their own. As they progress, they can pub-
Oral eflornithine Aventis
lish their findings in scientific journals and
CDC, Centers for Disease Control and Prevention; DNDi, Drugs for Neglected Diseases Initiative; NIH,
discuss their insights in on-line forums. National Institutes of Health; TDR, UNICEF–UNDP–World Bank–WHO Special Programme for Research
Over time, the better ones will gain authority and Training in Tropical Diseases.
and become the de facto leaders of their
open-source community.
is a group of senior executives that rules Projects origination. There is a permanent
Outsourcing. Work to be outsourced on important operational issues such open call for new projects. Scientists are
should be posted on a website for all to see. as fundraising, budgets, project funding, invited to submit ideas online for review by
Scientists and organizations qualified for the key hires and selection of partners. the SAC.
job can bid, and the sponsor picks the best It also approves recommendations from
candidate for each task. the Scientific Advisory Committee. The Portfolio management. The SAC is
Scientific Advisory Committee (SAC) is a responsible for maintaining an adequate and
Template features: procedures group of external experts from academia balanced pipeline of projects.
Governance. Three decision-making bodies and industry. It sets R&D strategy, proposes
provide leadership and guidance: the Board new projects, reviews existing ones and Project management. Each project is man-
of Directors, the Steering Committee recommends termination of those that no aged by a Project Team led by a member
and the Scientific Advisory Committee. longer deserve support. of the organization, and staffed by external
The Board of Directors includes senior experts in drug discovery, clinical research
executives and outsiders who represent Scope. This template calls for focusing and regulation. The Project Team is respon-
shareholders and stakeholders. It approves on single diseases or related illnesses. sible for developing the budget and timeline,
strategy and ensures that management An organization working on unrelated overseeing outsourced tasks and ensuring
performance is consistent with the organi- diseases should establish separate websites compliance with GxP. One of its crucial
zation’s mission. The Steering Committee for each one. duties is selecting what will be open-sourced
NATURE REVIEWS | DRUG DISCOVERY A DVA NCE ON L I N E PU BL ICAT ION | 5

6. PE R S PEC T I V E S
Table 3 | Public–private partnerships lower the critical mass required to discover and develop new cures
Organization Focus Staff Pipeline
Number of Discovery PK Clinical Cumulative
projects spending through
2005 (US$ million)
Medicines for Malaria 13 19 12 1 6 103
Malaria Venture
TB Alliance Tuberculosis 18 12 8 1 3 20
Drugs for Neglected Trypanosomiasis, visceral 36 20 9 4 7 20
Disease Initiative leishmaniasis, Chagas disease
OneWorld Health Leishmaniasis, malaria, Chagas 40 5 1 3 1 ?
disease, diarrhoeal diseases
International AIDS HIV/AIDS 169 6 – 1 5 120
Vaccine Initiative
Malaria Vaccines Malaria 32 10 4 2 4 ?
Initiative
Source: Annual reports. PK, pharmacokinetics.
and what will be outsourced. The project can lead to a patentable invention will need to post tasks, and experts to register their
leader is accountable for generating the data to exercise caution with disclosures until the skills, and Innocentive, an online problem-
used to decide whether to fund the next step. invention has been reduced to practice and solving tool that lets a company post a
Commitment to a project is limited to the patent applications have been filed, just as challenge with a reward: whoever finds the
current step, until the data warrants commit- would be necessary in a traditional research solution gets the money.
ting funds for the next one. Open-sourced setting. It is generally accepted that open
tasks are posted on the project’s website, communication promotes advancement Template features: costs
each on its own page, and outsourced ones of science, but needs to be balanced by the PPPs have been able to function on very low
are posted on a companion matchmaking need to protect the rights of inventors. budgets for several reasons (TABLE 4). First,
website such as Innocentive, or Scienteur. The same applies to open-source activities. they have few people, low overhead costs
Outsourcing bids are reviewed by the Project and no fixed assets. They rely on someone
Team, which issues recommendations Template features: tools else’s unused capacity, and the market seems
to the SAC. The discovery toolbox. As of February 2006, to price such capacity at marginal instead
349 genomes have been published and of full cost. Second, they outsource much of
Intellectual property ownership. There is another 1,575 are being sequenced. A new their work where it is cheaper to do so and
often a misperception that open-source generation of smart, computerized tools do most of their trials in developing coun-
initiatives are hostile to patents and bent is becoming available to mine data, comb tries. Third, they concentrate on infectious
on putting discoveries in the public domain. the literature, map metabolic networks, diseases for which costs are lower. Fourth,
The reality is more nuanced. Most open- perform in silico modelling, visualize bind- they receive in-kind donations.
source activities occur at a pre-commercial ing sites, identify chemical leads, design
R&D stage, when the ideas and hypotheses molecules and predict toxicity. These tools Will it work?
debated fall short of the legal standards that should be packaged into a convenient Despite the promise of open-source drug
define inventions in patent law. They are an toolbox, together with access to major R&D, both its pioneers, and the veterans
on-going scientific conversation that can be databases, and offered to volunteers willing of open-source software, point to several
likened to a global instant-messaging system to contribute their expertise. potentially troublesome issues that could
linking scientists interested in a topic. In that affect the success of the open-source model.
sense, open-source is no more threatening Outsourcing software. Several programs
to patents than other forms of scientific already exist to match projects with talent Availability of talent. Typical open-source
publishing. A scientist who engages in that and capacity. Two examples are Scienteur, projects do not require a large number of
conversation and comes up with an idea that a free e-marketplace that allows companies contributors. Data from the software indus-
try suggests that the ideal number ranges
from 6 to 20 people. Yet much of the drug
Table 4 | R&D costs for public–private partnerships (US$ million)
R&D expertise resides in an industry that
Stage MMV TB Alliance DNDi IAVI Big Pharma has a strong proprietary culture. Employees
Discovery and PK 8.3 18.6 16.2 20.0 26.0 are routinely asked to assign their intellectual
Phase I 1.6 0.6 Unpublished 2.0 15.2 output, including that created on their own
time, to their employers17. This could stifle
Phase II 1.2 3.4 Unpublished 5.0 23.5
talent supply in key areas. Two developments,
Phase III 9.5 22.6 Unpublished 30.0 86.3 however, might give open-source drug R&D
Total clinical 12.2 26.6 24.2 37.0 125.0 the permanent talent pool it needs. First,
Source: REF. 19. DNDi, Drugs for Neglected Diseases Initiative; IAVI, International AIDS Vaccine Initiative; thousands of highly trained pharmaceutical
MMV, Medicines for Malaria Venture. scientists are nearing retirement and might
6 | A DVA NCE ON L I N E PU BL ICAT ION www.nature.com/reviews/drugdisc

7. PE R S PEC T I V E S
welcome the opportunity to put their skills enough. As Darren Carroll, former CEO of 3. Feldman, R. The open-source biotechnology movement:
is it patent misuse? Minn. J. L. Sci. Tech. 6,1 (2004).
to good use. Second, drug companies might Innocentive, puts it, “If you build it, they will 4. Cukier, K. N. Community property: Open-source
be persuaded to ease restrictions on their not come!”. It takes a sustained effort to get the proponents plant the seeds of a new patent landscape.
Acumen 1, 54–60 (2003).
employee’s involvement. There is indeed little word out and build trust with stakeholders. 5. Rai, A. Open and collaborative research: a new model
conflict of interest in being a cancer scientist It also takes a leader who can connect with for biomedicine. Duke Law School, Legal Studies
Research Paper Series, Research Paper 61 (2004).
by day and an anthrax researcher at night, people, understand their motivation and foster 6. Maurer, S. New institutions for doing science: from
and firms might gain valuable goodwill from trust. Linux attracts thousands of contributors databases to open biology. Presented at University of
Maastricht, November 24–25 (2003).
letting employees seek cures for diseases in because they identify with Torvalds’ ideals and 7. DeLano, W. L., The case for open-source software in drug
which they have no interest. trust him to do the right thing. Open-source discovery. Drug Discov. Today10, 213–217 (2005).
8. Geldenhuys, W. J., Gaasch, K. E., Watson, M.,
drug R&D must build such leaders. Allen, D. D. & Van der Schyf, C. J. Optimizing the use
Availability of data and standards. Open- of open-source software applications in drug discovery.
Drug Discov. Today 11, 127–132 (2006).
source scientists cannot accomplish much Web interface. The design of the project’s 9. Eisenmenger, F., Hansmann, U. H. E., Hayryan, S. &
unless they can access data. Biological data website is crucial. It must be engaging and Hu, C. An enhanced version of SMMP — open-source
software package for simulation of proteins. Computer
is plentiful and getting richer, with terabytes appeal to visitors’ curiosity. They must be Phys. Comm. 174, 422–429 (2006).
of genomic and metabolic data continuously able to quickly find the pages that match 10. An open-source shot in the arm. The Economist (12
June 2004).
being added to the pool. Chemical and their interests, download the toolbox, and be 11. DeMonaco, H. J., Ali, A. & Von Hippel, E. The major
structural data, on the other hand, are more ‘up-and-playing’ in minutes. role of clinicians in the discovery of off-label drug
therapies. MIT Sloan Working Paper 4552-05 (2005).
scarce. In addition, the formats used to 12. Maurer, S. M., Rai, A. & Sali A. finding cures for
handle these data are still evolving. Biologists Quality assurance/quality control. When tropical diseases: is open-source the answer? PLoS
Med. 1, e56 (2004).
use a reasonably small number of them, but something as complex as drug R&D gets 13. Gardner, C. & Garner, C. Technology Licensing to
chemists are further from such consensus. parceled out around the world, quality nontraditional partners: non-profit health product
development organizations for better global health.
Both the lack of standards and the scarcity of assurance can become an issue. Oversight, Industry Higher Education 19, 241–247 (2005).
data in certain areas can cause problematic due-diligence, audits, good practices and 14. DeBresson, C. & Amesse, F. Networks of innovators:
a review and introduction to the issue. Res. Policy 20,
choke points in an open-source R&D effort. prior experience can be used to ensure quality. 363–379 (1991).
International Organization for Standardization 15. Nwaka, S. & Ridley, R. Virtual drug discovery and
development for neglected diseases through public–
Availability of tools. Open-source scientists standards could also help in the future. private partnerships. Nature Rev. Drug Discov. 2,
need open-source tools to practice their 919–928 (2003).
16. Hope, J. Open-Source Biotechnology. Ph.D. Thesis,
craft. Until recently, such tools were plentiful Selectivity. Not all projects will be equally Australian National Univ. (2004).
in bioinformatics, but less so in chemistry, suitable. Cancer might draw contributors, 17. Stahl, M. T. Open-source software: not quite endsville.
Drug Discov. Today 10, 219–222 (2005).
which has long been dominated by com- but hair loss might not. 18. Collins, F. S. The NIH Roadmap: new pathways to
mercial software. This is changing. The 2004 discovery — empowering small molecule research.
Office of Portfolio Analysis and Strategic Initiatives
launch of PubChem has brought online a Conclusion: a new ecology of drug R&D? (National Institutes of Health, Bethesda, 2006).
powerful suite of tools that allows scientists Is there still room for big pharma in open- 19. Towse, A. & Renowden, O. in Combating Diseases
Associated with Poverty: Financing Strategies for
to connect chemical information with source R&D? One must stress that ‘virtual’ Product Development and the Potential Role of
biomedical research and clinical informa- does not mean ‘leaderless’. To succeed, Public–Private Partnerships (eds Widdus, R. &
White, K.) [online], <http://www.globalforumhealth.
tion in an unprecedented way. Non-profit open-source R&D will need deep expertise org/filesupld/ippph_cd/06.PDF> (Initiative on
scientists can now access small-molecule in the minutia of drug R&D, which today Public–Private Partnerships for Health, London,
2004).
high-throughput screening, chemistry and resides overwhelmingly in the pharma-
Acknowledgements
informatics on a scale previously available ceutical industry. There might be many I thank A. Tashjian (Harvard School of Public Health and
only to industry. They can even get grants to volunteers, but they must be shepherded Harvard Medical School), B. Smith (Center for Biosecurity,
University of Pittsburgh Medical Center) and M. Munos
turn their online discoveries into assays for towards a goal. Such stewardship is a core (Gardner Carton & Douglas) for valuable feedback on
high-throughput screening18 . Other tools competency of pharmaceutical companies. previous versions of the manuscript.
such as eMolecules, Jmol or the Chemistry Our model is not a substitute for them, but Competing interests statement
B.M. works for Eli Lilly & Co., which has sponsored the
Development Kit are adding powerful a way to leverage their capabilities to tackle Scienteur and Innocentive ventures mentioned in this article.
chemical search and visualization capabilities unmet medical needs, such as the diseases of This declaration of potential competing financial interests is
also available in the Web version.
to the open-source scientist’s toolbox. poverty, orphan diseases and niche markets.
Pharmaceutical companies stand to gain
FURTHER INFORMATION
Intellectual leadership. Just as putting ingre- from co-opting the open-source model and SourceForge: http://sourceforge.net/
dients into a vat does not necessarily cause allowing it to flourish in ‘coopetition’ with BioForge: www.bioforge.net
PubChem: http://pubchem.ncbi.nlm.nih.gov
them to react, connecting smart people online traditional R&D, to handle the diseases or eMolecule: www.emolecules.com
does not guarantee they will produce anything R&D steps for which it is best suited. Jmol: http://jmol.sourceforge.net
CDK: http://almost.cubic.uni-koeln.de/cdk
valuable. In both cases, a catalyst is needed. Emboss: http://emboss.sourceforge.net/
For open-source drug R&D, the presence of Bernard Munos is at Eli Lilly & Co., Lilly Corporate Medicines for Malaria Venture: www.mmv.org
Center, 1085, Indianapolis, Indiana 46285, USA. The Initiative on Public-Private Partnerships for Health:
a subgroup of highly innovative contributors e-mail: bhmunos@stanfordalumni.org www.ippph.org
who can tune in the on-going conversation Global Fund to Fight AIDS, Tuberculosis and Malaria:
doi:10.1038/nrd2131 www.theglobalfund.org
and fuel it with their own creative insights acts Published online 18 August 2006 Aeras, Global TB Vaccine Foundation: www.aeras.org
as such a catalyst. Without it, the conversation Drugs for Neglected Diseases Initiative: www.dndi.org
1. Hollingsworth, J. R. & Hollingsworth, E. J. in Practicing Global Alliance for TB: www.tballiance.org
could remain shallow and fizzle out. Interdisciplinarity (eds Weingart, P. & Stehr, N.) Institute for One World Health: www.oneworldhealth.org
215–244 (Univ. Toronto Press, Toronto, 2000). International AIDS Vaccine Initiative: www.iavi.org
2. Hollingsworth, J. R. in Creating a Tradition of Malaria Vaccine Initiative: www.malariavaccine.org
Momentum. Enticing people to join is a Biomedical Research (ed. Stapleton, D.) 17–63 Access to this interactive links box is free online.
challenge. A good website helps, but it’s not (Rockefeller Univ. Press, New York, 2004).
NATURE REVIEWS | DRUG DISCOVERY A DVA NCE ON L I N E PU BL ICAT ION | 7