DR SZYMANSKI AND DR SCHYFTER
After coming up with a project idea and designing a way to test out our method, we decided it was time to start a discussion about whether or not our project had practical applications. Our first meeting was with Dr Erika Szymanski and Dr Pablo Schyfter of the University of Edinburgh’s Science, Technology and Innovation Studies Department.
Their input really put our project into focus and laid down a track for the rest of our human practices. They told us that we were trying to target too many people with our DNA storage system and that we needed to be practical about who would actually use our product. As reading and writing data in DNA is slower than storing and retrieving something from a USB drive or even downloading a file from the internet, we needed to focus more on long-term, low access data storage. After this meeting we got in contact with local librarians and data experts.
Another takeaway point from the meeting was that up until this point we had not thought about the physical storage of our DNA. Were we planning on storing the DNA in some sort of time capsule that was only to be opened at certain times? Or were we going to focus on more practical and accessible storage of DNA in vials? We addressed this when we tested the viability of DNA in different storage conditions.
DR ANDY TURNER AND DR ADAM CARTER FROM ARCHER/EPCC
Following our meeting with Drs Szymanski and Schyfter, we set up a time to talk to Dr Turner and Dr Carter from ARCHER and EPCC. ARCHER is a supercomputer at the University that runs simulations and calculations that are too resource intensive to be performed on a regular computer. EPCC hosts ARCHER along with other computing and data facilities. Drs Turner and Carter have extensive experience with generating and storing large amounts of data.
Though very positive of our idea, they confessed that DNA data storage would not work best for them. The simulations and research they facilitate require data to be generated and stored instantly and in high volumes, whereas DNA has a read and write time that is too slow to accommodate supercomputing. They suggested that we meet with librarians or archivists that store important, historical data for long periods of time. It is important to securely store this kind of data since it is often unique and cannot be regenerated or simulated.
They agreed that using Basic English as a proof of concept was very illustrative of our method, however, they pointed out that we should view the information coded in BabbleBricks to be flexible and that we should stress its arbitrary nature to others. Along with saying that this was one of our strengths, they encouraged us to come up with some sort of cost analysis comparing our assembly method to other storage techniques. They said that defining the ‘workflow’ of our project would help us decide whether we envision a BabbleBlock factory that stores all data in DNA, or whether we could scale down the system to a single machine.
EDINA AND MAIN LIBRARY
We were lucky enough to sit down with 5 librarians and data librarians from the University of Edinburgh Main Library. This big group meeting was great since some of the librarians actually had a background in genetics or biochemistry which meant we were able to get valuable and varying feedback.
The main points they raised were that data hardware and software are getting outdated. Often times data access is restricted by the software used to read the hardware. Though we are confident that our hardware (DNA) will last a long time, this really made us think about the future of DNA sequencing. With the current advancements in sequencing technology (ie. MinION) we also feel confident that the software needed to sequence our DNA will only improve.
One concern they raised was the limitations of how many BabbleBricks we could produce. At the time our 5bp information coding region allowed for 1,024 unique BabbleBricks. They expressed concern that this was not enough combinations to store different types of data. This feedback fed into the creation of BabbleBrick 1.7, which is a new version of the BabbleBrick system, storing jpg images in DNA.
Finally, they again questioned how we planned to physically store the DNA. After explaining to them the different techniques of keeping DNA (dried, in a solution etc.) they offered us the opportunity to test the fidelity of our DNA in their existing storage conditions. This was an amazing opportunity to prove the applicability of our project; if DNA can survive in existing control conditions (and in less controlled conditions), then there is no need to specify special conditions that could consume more energy and money. To read more about our control room and viability experiments, click here.
DR DAVID ASPINALL AND DR MYRTO ARAPINIS
From the outset of our project we wanted to be able to include some form of encryption in our DNA data storage as this is something we had yet to encounter in the literature. At the start of the summer we met with Dr Aspinall and Dr Arapinis to pitch to them our initial ideas and see what feedback they had. Dr Aspinall is the leader of the Security and Privacy research group in our School of Informatics and also leader of the Cyber Security & Privacy Research Network at the University. Dr Arapinis is a lecturer in the School of Informatics with a research focus on verification of cryptographic protocols, such as verification of security properties and detection of attacks.
Upon explaining our initial ideas for encryption, Dr Aspinall suggested that we look into applying a stream cipher method to our DNA data.
Within the next few days we also met with Dr Arapinis, who agreed with Dr Aspinall that a good encryption system would include a stream cipher. However, at this meeting we pitched to her the Vignere cipher method that we had developed and she told us that if we kept this system our encryption could be hacked very easily. She recommended we look into Stream Cipher: SALSA. To see the full development of our stream cipher, click here.
DR KAMI VANIEA AND DR AGGELOS KIAYIAS
Dr Vaniea is a lecturer in the School of Informatics with a strong research interest in the human factors of security and privacy. Dr Kiayias is an Associate Professor of Computer Science and Engineering, working specifically in cryptography and computer security.
Dr Vaniea agreed that a stream cipher seemed to be the best option for us, but she had a few hesitations about our plans for key exchange. She encouraged us to focus on CIA when developing an encryption system: that is, Confidentiality, Integrity and Availability. She also questioned whether encryption was really necessary in our system.
Dr Kiayias had similar feedback for us; in order to apply a cryptographic function to encrypt your information, you first need some sort of threat model that you can use as a starting point.
This response and scepticism inspired us to come up with scenarios in which it would be vital to have a form of encryption on our BabbleBlocks. Read more about this here.
Richard Bennett is an Executive Director at Goldman Sachs dealing with technology risk and information security. We were lucky enough to get in touch with him over email and get his advice on our encryption. He cautioned us to think very carefully about our system and come up with some clear reasons for needing encryption (see our scenarios here.).
He also made a point of encouraging us to think about our keys and the best way to exchange them. As he put it, “When it comes to cryptography the key is the keys!”; if your keys are compromised then there’s no security at all! This led to the development of our RSA Public-Private Encryption method. You can read more about it here.
DR JOYCE TAIT
Dr Tait is a bioethicist based in our University’s Science, Technology and Innovation Studies department. She has or is a member of the US National Academy of Sciences, Scottish Science Advisory Council Working Group on Synthetic Biology and the UK Department of Health Emerging Science and Bioethic Advisory Committee. She has extensive experience with policy analysis and stakeholder attitudes in synthetic biology.
In our meeting we asked her a lot of questions about public perceptions of synthetic biology and how she anticipates people might react to our project. Her take-home point was that we can never assume or try to predict how people will respond. She also pointed out that from her experience people are never as worried or ignorant as you think they might be and the way in which people respond to your technology depends on how it is presented to them. For example, she introduced us to the concept of ‘gold plating’; 'gold plating' is the addition of extra measures or the over-implementation of a policy in order to make the policy seem better or safer. She said that’ gold plating’ is a term often used in a legal context, but the same can apply to researchers. As an example, if we put a lot of focus on how safe our BabbleBricks are and how they can never be dangerous, we may give people reason to think that there is some sort of risk we are trying to compensate for. She pointed out that as there are regulations in place to prevent ‘gold plating’ of EU law in the UK, we should be mindful to avoid it as well. To see how this played out in the safety of our project, click here.
DR TRISTRAM RILEY-SMITH
Dr Riley-Smith is the Director of Research at the University of Cambridge Department of Politics and International Studies and the External Champion to the Research Council Uk Partnership for Conflict, Crime and Security Research (PaCCS). Dr Riley-Smith has studied anthrolopology but currently plays a bridging role between the commercial world and government agencies (such as PaCCS). As such he was able to provide a lot of guidance on how to view or come up with policies related to our project.
In particular, he pointed us in the direction of FRRIICT(Framework for Responsible Research and Innovation in ICT) which we used to make sure our research was safe and responsible. To see our reflections on this, click here.
Edward You is a special agent at the FBI working in the Weapon of Mass Destruction Directorate, Biological Countermeasures Unit. We had the opportunity to skype with him and gauge his impressions of the project. He was pleased with what we told him and was especially happy that we included some measure of encryption into our design. Though our main application is archival data, he pointed out that our technology has the capacity to advance and become more practical in everyday life. In this way, he said that encryption is paramount to this development.
After sharing with him the threat scenarios we had come up with for our encryption system, he raised another topic for consideration; data integrity. To see how we have covered data integrity, read more about our error-correcting mechanisms .
LEE HIBBERD, NATIONAL LIBRARY OF SCOTLAND
After our initial meetings with the Main Library we decided to try and get in contact with the National Library Scotland. We were lucky enough to meet Lee Hibberd, the Digital Preservation Officer at the library. He was very enthusiastic about our project, how much we had achieved and agreed with us that DNA could be a fantastic medium for long term storage.
We learned a lot about archival data at this meeting; for example, Mr Hibberd explained to us that magnetic tape copies of data have to be corrected and rewritten every 6 years- costing the library thousands of pounds in extra electricity, man power and materials. He was glad to hear about our our error-correcting mechanisms and said that the Library also uses check-sums to detect errors in their tapes.
We also learned that the cost of storing text, even long term, is relatively cheap compared to other forms of storage. Mr Hibberd explained that this is because text is easy to condense when storing. He recommended that an improvement to our project would be if we could store data other than text.
You can watch our interview with Lee here.
Photo Goldman Sachs: https://en.wikipedia.org/wiki/Goldman_Sachs