Working group roadmap: Safety engineering

Farren Isaacs, Yale University

I was part of the discussion on safety engineering.

• Farren Isaacs
• Peter Carr
• Brad Schmeir
• Michael Chou
• Daisuke Kuga
• Weiwen Zhang
• Cunjiang Song

What is the role of safety in the context of GP-write? We're drawing a parlalel to a lot of efforts that have been involved in safeguarding information that we care about on the internet whereby over the past several decades this incredible technology has emerged that allows us to send information throughout the world in a matter of seconds. What's important here is that there are a lot of efforts that have been necessitated by various threats to safeguard this information on the internet.

I view the role of safety in GP-write is an important effort to safeguard the full potential of these biotechnologies. I also think that this topic has really important relevance and links to other workshop groups in particular I would draw important ties of safety features as linking to the technical efforts and some of the ELSI components, sort of referencing important quotes from the commission report on synthetic biology from a few years ago from bioethics.gov, for the ultimate goal of enhancing humanity and the environment.

When do we have to care about safety? By and large, people have been benefiting from advancements in biotechnology for 40 years. Mostly these are conducted in closed systems. But now these systems might eventually get into the environment either intentionally or otherwise. In this context, we have to take some of the dialogs and early steps that have been developign in this area over the past 30-40 years and really advance it.

Some specific examples are gene drives, kill switches, unnatural amino acids, conjugation resistance, alternative DNA backbones, biocontainment, genetic isolation.

Biocontainment is where cells cannot just grow in the wild or nature, without some extra component that is safely controlled. Also, we need to prevent the exchange of genetic material between organisms and viruses or between organisms. Recombinant DNA technology originally brought this up, and now they are becoming important engineering objectives that we have to actually address.

We have to engineer these types of safeguards into the systems we're working on. There's also some important ecosystem perspectives that could enhance distribution or foster more partnerships in industry if certain types of organisms and their proprietary efforts can basically be secured.

I would say over the past few years, particularly there has been advancements in engineering biocontainment, like SOA escape frequencies until 2015 was about 10⁸ to 10¹⁵.

Over lunch, one of the interesting points was messaging around safety. Is safety features the right terminology? Is it all vs nothing? Sometimes it's all or nothing, but really we want to enhance safety features or endow more safeguards. The other element is considering relative levels of risk and that might be the more important element to focus on.

Related to safety, how do we define, ... use and application of those, will really die if ... associated risks, a lot of clarification to be produced around this. Drill down more on the topic of risk, really we should come up with metrics with respect to what levels of risk are okay for certain use. Immediately employ that for various applications in closed and open ssystems. How do we quantify the risk, what's the burden of proof? How do we expand that to the regulatory level, how do the agencies that are going to regulate these products, how are they going to enable these products to enter the public markets?

Ultra safe cell lines was published last year, and it's still a good set of ideas.

Q&A

Q: How do you prevent something from being hacked? How do you prevent germline? Can you prevent people from putting genes back in?

A: It's a good question. I think there are ways to do that. We've considered ways to sort of discreetly introduce various safeguards and DRM. Unique ways to decode them into the actual genomes such that it would be non-trivial for someone to go hack the cell to remove those features.