#DappCon22 : Day 2 - openlab - A Peer2Peer Exchange For Life Science Lab Services, Niklas Rindtorff

Transcript

okay guys we're back now we have Nicholas rinstock it's a physician scientist and a founder of labdao an online community of light scientists collaborating to increase the accessibility to life science tools he's a research and a scholar training Heidelberg and receive a master's degree in biomedical informatics from Harvard Medical youth School today he will talk about lab Dao an Open Lab a peer-to-peer exchange for life science laboratory services please join me welcoming him on stage Nicholas [Applause] all right hi everyone I'm Nicholas and I'm going

to be talking about laptop where we're building a research organization online so the first question that you should ask yourself when you're trying to build a research organization is what defines an effective research environment and luckily that's a question that we're not the first ones to ask there have been a lot of people asking that question in the past they've been looking at cases like Bell Labs or DARPA and through to try to figure out what makes a research environment very effective and actually people have studied that and then they propose some new ways to build research organizations and this is just a quick outtake of that where for example Juan Bene is suggesting these internet native crypto native powered arbos and Benjamin Reinhardt just thinking about private

arpas Michael Nielsen is writing about Reinventing Discovery using online power risk organizations and um at a marble Stone in San Rodriguez has been writing about focused research organizations so they're we're currently seeing a camera explosion of new proposed ways to organize scientific activity I think there's one case study that has been somewhat understudied and that can be very informative for how you could build a restriction that exists completely online and those are the phage courses called Spring Harbor right after World War II So Cold Spring Harbor is this little town it's about 50 minutes from New York City and very few people know that this is where some of the fundamental research happened that basically led to

the founding stones of molecular biology today as we know it and what happened there well over the Summers scientists from all over the US and Europe would come together and they could spend the summer tinkering while they gave courses about the biology of phages and what's very special about that so phages are these little viruses that only target bacteria and not human cells and what was special about it was that these scientists they came from all different places and all different training backgrounds so for example you get Max Delbert who was actually a physicist by training but he was hanging out with all these biologists and then not only were there very diverse teams there were also at a place where there was some laboratory space to actually go and tinker and so there was infrastructure available and then last

but not least there was a bit of funding so they could come together they could propose a new research idea and then they could go back to their home institutions and like ride the big grant to get serious follow-on funding with that idea and I think that idea is very powerful for what we're trying to do the key lessons here are that an effective research environment needs to have access to Scientific infrastructure like you need a lab space to do the really interesting stuff you need to be able to Tinker at the edge of what's Theory you need to be able to form very Dynamic teams where someone with micro expertise can come in and that doesn't necessarily mean that everybody on that team has to be like has to have grad school experience that can be very uh broadly skilled and then the last one at least there needs to be at least some funding

for prototypes and infrastructure to emerge now you could ask yourself why does all of this matter why do we need yet another research organization and why especially do we need a research organization that's completely online and this is a bit of a personal story so what you can see on this graph here is the flow inventors across countries net net where you on the y-axis you can see a inflow of inventors into a given country in the in Gray and went to whiteboard means that actually this country is churning inventors and an inventor in this study is defined as someone who has issued or filed a patent so here you can see there are a couple countries there are winners in this ecosystem the United States Switzerland Singapore and they're attracting

inventors but then there is a large group of countries that are net exporters of inventors so if you're creative if you are well trained if you have a lot of new ideas there are only very few places in the world where you can be home where you have the the place to Tinker the teams the dynamic teams to form and the funding available to like follow your ideas and test them against reality now I personally was born here moved over here came back moved there came back moved here came back and eventually asked the question how many more times do I have to do this and also how many more people are in this pocket over here of this graph that do not have the chance to like go to the right side of the graph and actually try to pursue

their scientific ideas so let's build a home for inventors online let's put the online community where people can come with their wild ideas people can get some specific training that they require and actually form form research proposals and go that first step that you could do when you were at Cold Spring Harbor which is come together in a team do the very first experiment see whether your idea has legs and then go and apply for follow-on funding and this is all part of a broader mission that this DSi ecosystem is on that you might have heard of before word usai stands for decentralized science what we really want to give everybody the opportunity to raise funding for their scientific endeavor we want to enable everyone to have

access to laboratory services or any type of specialized scientific service and then where we also want to develop New Economic mechanisms that token engineering allows us where data sharing and sharing of materials is both incentivized for the inventor and also is available and maintains accessibility for everybody else in the community and maybe before I dive into what we're doing at lapdow and how the Open Lab exchange fits into this whole DSi ecosystem I provide a bit of a landscape so when we think about signs and this is any type of academic science that you know you know maybe know from grad school or your college experience you can you can segment it into three different boxes where you have a funding ecosystem an ecosystem where you can

actually do the scientific experiment and then you have a place that or a set of organizations that you used to distribute your Insight and over the last couple months really we have seen it came with an explosion of organizations in the web 3 ecosystem that um filled these different niches so for example we have science fund git coin and Vita Dao where open source developers and also scientists can raise funding to do their experiments and in case of Vita Dao they in in return for funding give veto Dao some some IP ownership over the outcome and then we have these distribution mechanisms such as research Hub molecule um and and the DSi Foundation that are working on new tools to spread knowledge oftentimes based on ipfs and direct content addressing

what is somewhat missing in the middle is a place where teams and tools are provided are provided and actually the invention can happen and that's what we're building and one of the key issues however when you try to do this whole thing online is how do you provide scientific tools right they're really clunky and hard to use and complex in fact if we break it down and we just use biology is an example here and we compare it with code code is a very low cost of execution like you can run on your laptop most of the time and it has high reproducibility if I run it again and again it will always give you the same result you know condition you have this seed set to be the same thing in biology it's the direct opposite you have extremely high cost of execution

and very low reproducibility and ideally we want to get to a point where you can have people remote control the con uh the process of biological experimentation so basically how do we get to a point where doing biomedical research is feeling a lot more like writing code than than actually sending the lab pipetting something as some kind of custom-made experiment now this is a pretty big task to tackle and one way you can break it down first is to say well let's focus on applications that are high cost of execution and high reproducibility before we go for the high cost of execution lower producibility type of workflows I'm going to go into that in a sec

and what we want to do is we want to create an exchange lab exchange that you can access with the Open Lab client that is a dow government P2P exchange for data generating laboratory services and just generally scientific services so that if you have someone in the Dow that knows the laboratory workflow or scientific compute application really well and you have someone else in the Dow that wants to consume that service they have a way to lower their cost of trust to engage in a fruitful coordination now we're not going to be able to just attach a lab right away to this exchange but we can start with stuff that's really hard to set up if you don't know how to do it and still highly refusable and only computational in nature and

this is just an example of an application that we've done built together where the original scientist that developed was this method which is called equibind you can use it to basically on your computer model the way that a small molecule which is basically most of the drugs that you know from the pharmacy interacts with protein which is what most of your body is made of next to water and you can model the interaction between these two this is actually a pretty complicated set of software to get running and especially takes a lot of resources to run at scale so it's very unaccessible for a lot of scientists um but with the exchange you can get access to it with a laptop and the author was Hannah Stark and a Community member that's maintaining it as Calvin Wallace

um so we're basically dividing also the labor of there's a scientists at advances but then there's also a group of Open Source maintainers that keeps it going and this is just one at one vignette we have other services as well for language modeling of antibodies RNA sequencing protein folding but also actually some wet lab services that you would know from your like rotations such as protein expression now we don't want to be in the core team the ones that only onboard these applications we actually want to motivate everybody in the community to onboard applications um and we're very I'm very happy that there are four projects particular in this current bitcoin round that are raising funding to build their applications and to provide them to the

community and if you're interested in rainforest monitoring synthetic biology large language models for human uh mental health research and knowledge graphs and give this give this a look you can find them at laptop underscore gr15 in the bitly at the bottom now this is davcon and I know we're all passionate about dapps but the most intuitive way for a scientist especially a bioinformatician to interact with these laboratory services actually not through adapt but through a command line interface where we're basically serving a CLI client that's based on typescript for people to interact with the exchange and when you

look at this here the user can install the client and has a local ethereum wallet where they can manage their account they can list all the apps that are available they can interact with the exchange for example by posting a request for an application run or by actually doing the application on their own hardware and they can manage their file input and output via ipfs and filecoin for the pinning and for this we're closely working together with the backlab team and protocol Labs which is currently working on a compute over data infrastructure to run actually compute very effectively over file stored using the fileco network now how do you keep such an exchange secure when there is no cryptographic way to prove that the result that you got out

is actually the intended result right there's no like prime number or factorization that you can do to very quickly check whether the result is is correct that you've paid for so you need to create some more social technology to make sure that the stability of the exchange is intact and one way that we go about that is with three different methods the first one is an escrow in arbitration service so if a Community member pays another Community member to do a laboratory service for them the money isn't directly sent to them and in case of a conflict there's an arbitration service that can mediate we also have a governance tokens that you would have to stake in order to participate in the market economy and if you're a bad actor your stake and your participation in the community can be reduced and then last but not least the

community is also actively voting on what are the laboratory services that it admits into the exchange ecosystem to make sure that a lower bar of quality is maintained and this is just an example for how some of the metadata can look like when you have a run where there's a name for the laboratory services and then it's always a hex type of information about me as a user the parameters that provide the input data I give you and then you as the recipient the metadata that you have from your run and then the output that you give me back what we hope to arrive with all of this is that we can see a global standardization of laboratory services and the the simple Game Theory here is when you have a high cost of conflict resolution when everybody is doing tailor-made Services over time we will

gradually arrive at a point where these laboratory services are going to be highly standardized for reproducible and also composable and a very good analogy from history is the shipping container where before we had the shipping container in the form that we know it today people were like exchanging a lot of information and overhead information about you know the dimensions of the load that they wanted to exchange until people just said look this is way too much conflict let's just create the standard size for us to transact and I think we're very close to that in the like some of biomedicine today as well so that eventually we can arrive at a point where instead of just having custom one-to-one relationships on the exchange where scientists have these one-to-one interactions to get work done and collaborate we can have many to many

interactions we can have actually some Market liquidity where there's a spot price for getting DNA synthesized or DNA sequenced and then over over time we might arrive at a future where actually the whole scientific project can be described as an executable Knowledge Graph where I can write Json and I Define this experiment than the next experiment and the next data processing step and basically my information can just travel through these different markets at the end of the day science has always been about open source it's been about doing an experiment and then sharing your knowledge with the rest of the world in the most accessible form however a lot of science is actually

built on on tools that where knowing the source code alone is just not enough like looking at a paper and knowing how they did the experiment is not enough what's really great would be if they didn't only tell you how to run the experiment but they would do it for you with your parameters and that's what we want to get to we want to move from open source to open tools where it's not about downloading stable diffusion running it on your own laptop it's about just pinging the stable diffusion API so it can be run on someone else's machine because we believe that open tools accelerate progress if everybody doesn't have that setup cost you can actually improve the rate of technological innovation and then also you can offer like interesting and regenerative economics for the open source scientists that are

maintaining these tools because if I pay for someone to run the computer for me or do a laboratory service for me the person that has mastered how to do it because they maybe even invented it they can keep earning from from that Revenue like for a way longer period than usually also such an exchange opens up a Marketplace for physical Laboratories that are currently competing for open direct requests on the cro Marketplace that the biopharma industry knows and then last but not least with lap that we're trying to build a community of scientists and inventors that can accelerate each other and actually develop new Ventures new IP that can be then even aggregated in nfts which is something that we'll probably hear more about in just a couple minutes and with that I want to thank you for your time if you want to learn more and

stay in the loop on our upcoming client release head to our Discord at lapdow and I see you there and happy to take any questions if the time allows thank you thank you any questions thank you cool thank you