Is the availability of genetic information dangerous?

Point your web browser to the National Center for Biotechnology Information (NCBI), and you will find a treasure trove of genetic information. No fee or password is required to copy and paste the genetic code for Ebola Zaire, one of the most lethal strains of the Ebola virus, or variola, the virus that causes smallpox. According to the NCBI website, sequences of more than 130 billion base pairs (the building blocks for DNA) were available in April 2006, and the amount is expanding exponentially over time.

Is this widespread availability of genetic information a problem? It certainly could be. In the hands of a biologist intending to cause harm, the sequence information available on the internet could be the first step in bringing back the 1918 flu virus, for example. Alternatively, someone could make a currently circulating virus, such as Ebola, without getting the virus from a laboratory or an ongoing outbreak. With a detailed road map, they could construct it using chemical reagents and widely available technology. DNA synthesis technologies are getting better and faster, enabling greater swaths of genetic material to be more accurately produced.

Yet, should anything be done about these freely available sequences of potentially dangerous information? In my opinion, no. There are two main reasons why. Firstly, the information is already out there, and it is difficult to take something back in the digital age. But this begs the question, is there any genetic sequence that one can imagine that has not yet been posted that should not be freely accessible?

Secondly, the information is useful. For example, Siga Technologies's investigational new drug ST-246 was developed through an examination of the genetic sequences of poxviruses, including smallpox. The drug inhibits a protein essential for virus assembly, a protein found in multiple poxviruses but not in humans. This drug was successfully used to treat a boy who had developed eczema vaccinatum, a rare but serious complication of the smallpox vaccine that causes skin rashes in people with skin conditions such as eczema or atopic dermatitis. The boy had been exposed to the vaccine, which is a live virus, from contact with his father, who had been vaccinated against smallpox before deployment to Iraq.

Could all of this genetic information be available to those who could use it wisely, without letting everyone see? Not likely. It is difficult to imagine how one would decide who is allowed access, considering that the pursuit of biological sciences (and drugs and vaccines) is global. The information is needed in both the public and private sectors. If one were to block certain pathogens from access, it wouldn't help those who might be interested in developing countermeasures against them. Blocking access to agents that could be used as weapons would also isolate the biodefense research community, which would not be good for the quality of the science. Given the serendipitous nature of biology, blocking sequences could block scientists from understanding the significance of their work.

The question of whether genetic information for pathogens should be accessible has been considered by the National Research Council, in their report "Seeking Security: Pathogens, Open Access, and Genome Databases." They came to the conclusion that the current open access policy should not change, and that it is not practical or productive to block access. They recommend--and I strongly agree--that genetic information should be exploited fully to defend against infectious diseases of all types, whether they occur naturally or are the result of bioterrorism. Scientists should also be aware that misuse of this information is possible, perhaps even likely, and that defeating these bugs (and the misuse of them) will only happen through their efforts.

Is the availability of genetic information dangerous? Certainly. But so is driving a car, flying to the moon, or falling in love. We do a lot of dangerous things, simply because the benefits outweigh our concerns.

So, how dangerous is it to have sequence data available? As Gigi Kwik Gronvall wrote, it is "potentially dangerous," i.e. there is the possibility for misuse. But how likely is the misuse by terrorists? I would argue, not very. Sequence data require knowledge and background information--or, as Cornell Professor Kathleen Vogel terms it "context"--to be translated into useful information.

Besides context, successful gene synthesis from sequence data requires tacit knowledge--the type of knowledge that is necessary to turn a recipe into a cake. Tacit knowledge is gained by repetition and often goes unnoticed because most people simply do things as they were taught to do them. The transfer of tacit knowledge requires long-term intensive contacts between people.

Keeping this in mind, I would argue that it is not easy to synthesize a virus, not even for experts. It would be even more difficult to create a bacterium from scratch. There are roads to developing a pathogen that are much easier to travel than gene synthesis, such as the ones alluded to by Jens Kuhn.

The other participants of this roundtable seem to share my view that sequence data should remain freely available. However, I sense some hesitation in their comments about whether certain very specific data such as sequences of eradicated diseases or artificial pathogens should be made public. This is a dangerous concept to entertain. Work on eradicated or artificial pathogens has a particularly high potential for misuse. If a state carried out such work and the results were kept hidden, suspicions would certainly arise among other states. Here, it is particularly useful to put yourself in the shoes of states that are not close allies. How would they interpret your behavior? If Russia, China, Iran, or India interpreted U.S. research of this sort as the first steps toward an offensive capacity, they would likely begin or increase their activities along the same lines.

If we agree that genetic data should remain freely available, what limits on the use of these data are useful and possible? Here are two possibilities: Companies that synthesize genetic sequences recently started to screen sequence orders for "dangerous" bits. The assumption is that suspicious aggregations of orders for "dangerous" sequences, such as virulence factors or consecutive parts of the smallpox genome, would raise an alarm. This would make things more difficult for the "mid-level bad guy" who lacks the know-how and funds to synthesize sequences from scratch. Another option is to track the global distribution of certain equipment. If we knew, for instance, where all the high-speed gene synthetizers were located, we could keep a keen eye on how they were being used.

The other participants in this roundtable have (rightly) focused not on the access to genetic sequence information, but what it is possible to do with that information. In one scenario, sequence information could be used to recreate pathogens from scratch, through gene synthesis technology. (A recent report, "Synthetic Genomics: Options for Governance," outlines three basic sets of policy options for dealing with the dark side of gene synthesis technology and gives the advantages and disadvantages of each course of action.)

I would like to pick up the topic by addressing a question that Jens Kuhn asked: "Is gene synthesis an 'easy' technology that, with the proper access to machinery and reagents, could truly be used by nonprofessionals (e.g. criminals and terrorists) to create microbial genomes in the nearest future? Or does the technology require a level of sophistication and financial support that will only be available to professionals?" Iris Hunger wrote that "it is not easy to synthesize a virus, not even for experts." And Leonid F. Ryabikhin wrote that developing bioweapons, in general, requires "a high-level of scientific and technological know-how. Numerous other obstacles make this task extremely complicated." The general feeling seems to be that biological weapons, particularly pathogens synthesized from scratch, are just too hard to make to be a real threat, particularly from individuals or small groups.

If that is the case, and the technical obstacles are in fact a good deterrent toward bioweapons development--terrific! But, frankly, I'm not that optimistic, and I believe that the urgency of the problem requires that much more be done to mitigate an attack, such as preparing hospitals and investing in more rapid production of medical countermeasures, etc.

Why don't I think that we can rest easy? Just a few reasons: Firstly, expertise can be bought. Al Qaeda apparently paid $1.5 million to acquire highly enriched uranium--thankfully, it turned out it was not as advertised. (For more information on this incident, please see this report.) But members of Al Qaeda thought they could purchase the expertise and the material required to create a nuclear device, and experts believe that this is feasible and worrying. In comparison, it would likely be less expensive to acquire the skills and reagents to build a biological weapon, even a pathogen synthesized from scratch.

Secondly, advanced technology will become more accessible. Before I went to graduate school, I worked as a lab technician, making short stretches of oligonucleotides for researchers at Memorial Sloan-Kettering Cancer Center. What I did, day in, day out, was very routine and required only a little training. But 10 years earlier synthesizing a short stretch of oligonucleotides could form the basis of a PhD thesis. An incredibly useful technology like gene synthesis will likely become accessible to the scientific masses even more rapidly--10 years from now, it is likely that it will be found in labs everywhere.

Thirdly, high levels of scientific sophistication can be found globally. Gene synthesis technology, and other complementing technologies that could be used to make a biological weapon, are not out of the range of biotech companies, research institutions, and individual scientists in many parts of the world. Of course, it wouldn't be easy to make a biological weapon. But with persistence, it is doable, especially when the people doing the work are not breaking new scientific ground and know that it is physically possible that they can accomplish their goals.

I still think it is a good idea to keep sequence information freely available, because the benefits to security, to the development of medical countermeasures, and to scientific advancement outweigh the risks--but there are risks.

My fellow discussants repeatedly allude to the same reason to explain why we need to keep sequence data public: Not doing so would harm the advancement of science and would prevent the development of top-quality medicine, including for biodefense purposes. This implies that they would limit the availability of this information, were it not for this "do no harm to science" rule. I would argue, instead, that limits are inappropriate because, quite simply, they do not work. They will not prevent access: A dedicated terrorist will find a way to get the information he or she wants. But even if limits prevented access, they would not remove the threat. If we remove sequence data for traditional biological weapons agents such as smallpox from the public arena, who says that an enemy will not use regular influenza as a weapon? Plus, limits are not manageable globally. Who is to decide what information has to be limited? Who is and who is not allowed to have access to restricted information?

If scientists are serious about not wanting certain data available publicly, they need to think before they act. If it is likely that they will produce knowledge that they do not want to share with others, they simply should not produce it. The concept is preventive arms control. Scientists have to ask themselves whether it is worth it to take certain paths of research. In other words, they need to make a risk-benefit assessment before starting a project. Once a research project has been conducted, it should be published in full.

On a second point: the price of expertise. Both Gigi Gronvall and Jens Kuhn say that expertise can be bought, and therefore it is not a serious limiting factor in bioweapons development. History, however, teaches otherwise. The Japanese cult Aum Shinrikyo had trained scientists and a lot of money, and they did not successfully create a bioweapon. Al Qaeda certainly has a lot of money. In the past, they could convince a PhD-level Pakistani microbiologist to provide them with information, but tacit knowledge was not for sale. Other experts that could be bought, such as bioweapons experts from the former Soviet Union, are also not for hire. Most of them are in Western research projects--mostly civilian--not in Iran, Syria, or North Korea.

Lastly, both Leonid Ryabikhin and Jens Kuhn ask to "continue to explore possible ways to control bioinformation" and to "broaden our thinking." Well, do it! Please! I would like to hear new ideas; it does not seem to be easy. There is a lack of innovation about how to control the use of the diversified, globalized, and dual-use technologies, equipment, and knowledge in the life sciences.

In my view, a key principle has to be that we do not try to prevent certain activities from happening, but that we simply watch what people do. If someone or some country is engaged in a suspicious activity that might indicate bioweapon development--and this will be rare--governments and civil society should ask questions about intentions, whether the suspicious activities are taking place in the United States, Germany, Iran, or Malawi; or whether they are taking place in government, industry, or military facilities.

Scientists and governments need to be able to enter into a proper dialogue about these types of activities. Knowing that a gene synthesizer is located somewhere tells you about capabilities, but only through dialogue will we be able to determine its intended use.

I strongly disagree with Iris Hunger's assertion that history teaches us that biological expertise can't be purchased. One obvious counterargument is that history is full of achievements that were thought to be unlikely, even impossible, that is, until someone proved otherwise. This is particularly the case in the realm of inventions (and weapons) stemming from scientific advancements. But there is also the matter of proof: While most former bioweapons scientists may not have switched allegiances for money, is there evidence that none of them have? What about ordinary scientists who had no ties to bioweapons programs--is it possible that not one of thousands of scientists could be bought?

I also wonder about how truly important the tacit knowledge of weaponization is to a prospective bioweaponeer. Could direct information from a bioweapons program help a trained scientist design a weapon? I have no doubt that it could. But could the lack of direct experience be overcome with trial and error, a background in laboratory work, and perhaps knowledge gleaned from everyday biological controls (such as spraying certain bacteria on crops to repel insects) or biologics pharmaceutical manufacturing? It seems foolhardy to bet otherwise. After all, people who have made bioweapons had to learn at some point, and presumably, technological advances in biology have made that learning curve less steep.

In this round of comments, I would like to hear what my fellow discussants think about several specific proposed control mechanisms for science. Some have been taken from the "Synthetic Genomics: Options for Governance" report I mentioned in a previous comment. I invite my fellow discussants to suggest additional methods that they would either consider or reject to control the misuse of scientific research:

• requiring gene firms or oligonucleotide manufacturers to screen orders for potentially dangerous sequences;
• requiring gene firms or oligonucleotide manufacturers to store their orders, in case attribution becomes necessary;
• requiring licensing of those who order sequences from gene firms or oligonucleotide manufacturers, or requiring licensing scientists more generally.

Screening orders for potentially dangerous sequences is technically possible, and could prevent or mitigate an attack. However, it would be very important to establish what would be done if a suspicious sequence is detected, and if further investigation reveals that it was ordered by a researcher who has no obvious reason to order the sequence. I don't think it would be useful to involve the FBI for every sequence order that seems suspect, particularly as judgments are unlikely to be cut and dry. It would also be important to minimize the burden on the private companies that would be responsible for screening orders and reporting them; screening would be done at a cost, and in a global marketplace, it could place firms at a competitive disadvantage. Similarly, a reporting structure would need to minimize the burden on the company. For example, it may prove more advantageous to not report if the firm is shut down while an investigation commences. If those measures are in place, requiring gene firms to store their orders appears to be a minimal burden, and could help in an investigation.

I have sympathy with those who think that every scientist in a lab should be licensed, because many professions require licensing, and why should science be different? For example, engineers, electricians, hair stylists, and manicurists require a license to work. I also understand the rationale of scientists who oppose this concept and point to the degree on their wall and say, "See this PhD? This is my license." In the end, however, I think that a licensing regime would be very expensive and should be reserved for work in Biosafety Level-3 and Biosafety Level-4 laboratories. In this high-containment environment, researchers have access to, and have the potential to be exposed to diseases that may be immediate public health dangers. The public relies on these scientists to handle biological agents safely. A licensing regime may help to ensure that, at the very least, these workers receive safety training. I look forward to hearing others' opinions!

Two more thoughts on the issue of expertise: Expertise is important in building a bioweapon, but there are different types of expertise. As Jens Kuhn says, expertise in weaponization is rare. Expertise in synthesizing certain viruses is not. In talking about access to expertise we therefore need to clarify which expertise we mean.

Answering Gigi Kwik Gronvall's first set of questions: Of course, one of a thousand scientists can be bought, and there is no proof that none of the Russian bioweapons experts has moved to a country of concern. The point, however, is that it would seem to be a very rare event. Finding the one scientist among thousands that would be willing and able to work on bioweapons requires time and resources. This slows bioweapons efforts down and increases the chance of uncovering them before they are "successful."

The same basic argument is true regarding Gigi's second set of questions. Of course, terrorists can learn through trial and error and generate the necessary knowledge on their own. But again, this is time and resource consuming. We can hope to make the path to a bioweapon longer and more complicated, but it is impossible to ensure 100 percent prevention.

Gigi invited comments on the specific proposals that have been put forward. Before commenting on those specific proposals, I would like to say something general: What we need is transparency, and transparency efforts need to focus on the areas of greatest concern. Only then are we likely to uncover noncompliance. How do you define what activities are of greatest concern? In general, the more dangerous the agents being researched and the closer the work moves toward weaponization, the greater our concern should be.

With this basic thought in mind, the three measures Gigi listed are of limited usefullness, at least if they are not appropriately focused. Jens eloquently described the problems associated with screening gene sequence orders for dangerous sequences. Such screening could be useful in cases where a suspicion already exists, or for very specific sequences, such as parts of the smallpox genome. I agree with Gigi that requiring companies to store their sequence orders would be a minimal burden, and the data could be left untouched unless an investigation becomes necessary, i.e. a suspicion requires following up on a sequence order. I could also imagine using data-mining software to anonymously check sequence orders for suspicious aggregations of dangerous sequences, though I don't know how much this would add to our knowledge of what is being done with sequence data and gene sequences. Do the other discussants have thoughts about this?

Regarding the licensing of scientists, I am not sure that I understand the proposal. In Germany, and I believe all over Europe, scientists working in high or maximum containment facilities receive extensive safety training. What would licensing add, unless licensing were obligatory on a global scale?

It is a relief to me as a bench scientist, but a positive surprise to me as a biodefense professional, that all of the discussants in this roundtable argued in favor of publicly available pathogen sequences. This is an indication of the closing gap in thinking between life scientists and policy experts. In the past, this gap has made it difficult for either group to understand and accept the viewpoints of the other. Pathogen sequences need to be available to every scientist, argue the scientists, because otherwise research would come to a screeching halt or at least be impeded dramatically. Pathogen sequences can be available to every scientist, argue the policy experts, because the information is impossible to control or to withdraw, and transparency in research trumps secrecy. Both groups are largely in agreement on the availability of pathogen sequences being of no great help to terrorists. It is the weaponization of agents, rather than their acquisition through in vitro synthesis, that is most difficult for terrorists to achieve, although how difficult it really is will remain a focus of future debate.

This roundtable has not adequately addressed the risks that publicly available sequences may pose in the future once scientists create artificial organisms, resurrect extinct agents, or make more virulent or more transmissible known pathogens, either accidentally or deliberately. Iris Hunger correctly argues that we need transparency in these areas of research first and foremost and that the more dangerous a research agent (or project), the greater the concern should be. While I am absolutely in favor of the public deposition of any sequence derived from any finished research project, I also favor a priori oversight and consequently possible disapproval of planned research in areas where it is foreseeable that "dangerous" sequences may be created in the absence of a clear public benefit. The emphasis here lies on foreseeable.

I disagree with Gigi Kwik Gronvall's view that the CISSM oversight system would not be a net gain for biosecurity. She argued that it "would slow down science a great deal." However, I demonstrated in a working paper that the system would only affect a very small percentage of researchers and institutions. Only high- and maximum-containment facilities, which already are subject to other oversight requirements, and researchers that plan a very limited set of deemed-to-be-risky experiments (making an agent more transmissible or multi-drug resistant, etc.) would fall under the system.

The issue of what makes an agent dangerous, which set of experiments could be considered risky, and what should be overseen is, of course, contentious, as Gronvall points out. But this did not stop the Fink Committee of the U.S. National Academy of Sciences from outlining what it considered to be dual-use "experiments of concern." Moreover, researchers can and should be able to modify the list of covered research activities to keep pace with changes in science and technology, as has been done for over 30 years for recombinant DNA research. I also don't think that the CISSM system would necessarily require scientists "to devote many hours to the review process [of planned experiments]," as the system is tiered and requires oversight on either the local (i.e. within an institution through, for instance, the already existing institutional biosafety committees), national, or international level depending on the planned project. Some oversight could also be integrated into the grant-review process, thereby further involving bench scientists in the review system.

Finally, that there might not be 100 percent participation in a dual-use review system is not a legitimate argument for dismissing it out of hand. I suggest that how all of this could or would not work be the focus of a follow-up roundtable discussion.

Because we all agreed early on--for one reason or another--that most genetic information should remain freely accessible, this discussion turned quickly to the more general question of what information to oversee, limit, or even prevent and how to do it. This question has worried experts for a while. And though the answer is neither straightforward nor simple, it's a necessary question.

In response to the question about what information should be overseen, limited, or prevented, most scientists would say that higher risk information deserves more intrusive oversight. Scientists have compiled several lists of "risky" research. The often-quoted Fink report lists seven classes of experiments of concern, including experiments that increase transmissibility of an agent and enable the evasion of diagnostic and detection methods. The CISSM oversight proposal mentioned earlier categorizes activities into extreme, moderate, or potential concern; activities of extreme concern would include research with eradicated or BSL-4 agents. In 2002, Raymond Zilinskas and Jonathan Tucker proposed a list of six types of "sensitive" activities, including facilitating dissemination as a fine particle aerosol and improving stability of pathogens.

All of these lists are similar. Most mention synthesizing viruses as a problematic activity. And all seem to take into account which agents will work effectively as bioweapons. Despite these similarities, scientists and experts are not close to agreeing about what constitutes risky activities. After all, the pathogen is only one part of a bioweapon. The dissemination mechanism is the second part, and it is often overlooked. Some of the participants in this roundtable, particularly Jens Kuhn, have referenced "weaponization," but we didn't properly discuss the subject. Discussions about risky weaponization activities address factors such as the open-air release of agents, methods for aerosol immunization, or the use of viruses as transport vehicles. Dissemination methods are important because, in most cases, they spell the difference between a bioweapon attack with limited effects and a mass casualty attack.

Most existing and proposed oversight procedures focus on research before it begins--for good reason. However, some research results are completely unexpected (see the legendary mousepox research in Australia) so projects need to be reviewed after they're done as well. So far, however, post-research oversight is left to individual scientists and is voluntary. Will this work? It’s unlikely. A soon-to-be published survey by the Research Group for Biological Arms Control at the University of Hamburg shows that most major English-, Russian-, and Chinese-language life science journals have not implemented security review procedures, despite calls for them to do so.

All existing and proposed oversight approaches only work when they are implemented widely, ideally on a global scale. Whether pulmonary plague breaks out in Washington, Berlin, or Nairobi it will reach all corners of the globe sooner rather than later. Implementing good oversight in one country will have a limited effect. But how do scientists go about making oversight international? Many states place biosecurity very low on their national agendas; they have other problems to address, for example, hunger, AIDS, or corruption. Biosecurity oversight typically follows biotech industry development, though most sensitive research projects are still done in the Western world, even though biotechnologies are rapidly expanding across the globe. So, for the moment, Western states bear the biggest part of the responsibility to protect the public. They still can and should influence the conduct of biological research worldwide to prohibit the hostile use of biotechnology.