 Next, Jonathan will build off of Colin's presentation, and we'll talk about developing technical solutions to facilitate genomic data access. Thanks, Elena, and thank you, Colin. As Elena said, I'd like to dive into what we've been working on at the Broad to think about how we might handle data access. Once participants have consented for their data to be shared, what can we do with the downstream processes to really facilitate data sharing? As many of you know, the amount of data access requests are likely to increase significantly. The more genomic data becomes available, the more that tools are easily available to analyze it, and the more that individuals are trained to do so, it's very likely that the demand for that data is going to increase significantly and likely exponentially. With that, that's going to put a significant demand on the signing officials and the data access committees that are required to review data access requests before researchers are granted access to that data. While some DAX, and particularly those at NIH, do a great job of turning around data access requests in a timely manner, as more and more research teams throughout the world find themselves in the positions of becoming data access committees themselves, this is an increasing burden and likely to become a bottleneck that will slow research down while we're working in so many other ways to expedite it. Just to dive in, I want to just look at the data access request process as it stands here on the top and compare it to what we hope to do. First, going from the left of the top diagram, we know that data in its future use is governed by the data use permissions that are put in consent forms. That data use is then aligned with that data wherever it's stored in a repository so that a DAC can help manage data access decisions on that data. As they come in from what we see over here on the right, researchers who submit a data access request that they write out, usually in narrative form, have it approved by their signing official and then it's submitted to the DAC who has to make that decision. Now, what's complicated about that is that, first of all, signing officials need to review every single one of these data access requests and then the DAC is really dealing with comparing consent forms, data use terms, which may be a bit complicated, and then also with a narrative description of research that wasn't written to be easily comparable with those data use terms, making that decision a bit difficult. What we'd like to do is really talk about ways we can bring in, particularly machine readable or computable structures to help with the data use terms and consent forms and with data access requests, how we can help get the signing official really continuing to give them the authority that they need, but in a more efficient way in enabling DACs to make better decisions faster. Just to start out, on consent forms, what we're doing is we've been working with the Global Alliance for Genomics and Health where I serve on their data use team to develop the data use ontology, and that data use ontology is just a structured machine readable vocabulary to describe different types of common permitted data uses that you might find in the consent form. Putting these terms and definitions in an ontology then lets them be machine readable, which means they could be easily stored in a database right alongside the data that they represent and then also could be easily searched or even computed upon. For anyone interested in those, I would just call out that the Global Alliance has a nice guidance on how to use these data use ontology terms in your consent form on their website. The other thing I mentioned is helping these signing officials maintain the level of influence that they'd like in authorizing their researchers, but making the process a bit more efficient for them and for their researchers who are seeking to access data. Now, as I said, signing officials need to review every single data access request that goes to each unique deck and then usually in that process are involved potentially in the negotiation of and at least the signing of unique data access agreements with each of those data access committees in their host institution. What we are proposing and what we've been working on, especially with folks at NHGRI and throughout NIH, is this idea of a library card agreement where signing officials can pre-authorize researchers to then submit data access requests directly to the deck, thereby taking themselves out of that process in each case, but still giving them the authority to issue permissions to certain researchers and perhaps not to others. A further thought, which we've been discussing a bit in the Global Alliance for Genomics and Health Circles, is this idea of a master library card agreement or a global one that maybe many institutions would agree to, but would just call that as something if you're interested in to pay attention to via some GA4GH channels or we could talk about in the questions. As I also said, another thing we really want to work on is helping the data access committee with their decision. One thing we're doing, particularly at the Broads, we've developed a software called DUOS, the Data Use Oversight System. And what DUOS intends to do is you can see that same process I showed earlier, is to take the data use ontology and to code the data use terms in the consent forms and the data access requests, both with data use ontology terms, such that they can be easily in kind of an apples to apples way compared by the data access committee. And that's what the DUOS software really allows us to do. And then on top of that, what we're also able to do, since those data use terms and data access requests are both in the same language and it's a computer readable language, we have a DUOS matching algorithm which then provides a suggested decision on the data access request to the data access committee, which they're welcome to use or not to use and we're continuing to iterate that as we pilot DUOS, both at the Broad and then also with NHGRI and again a number of colleagues throughout NIH. And so in that pilot, just to call out what we're really looking into, we want to ask ourselves, first of all, are we accurately translating what is in consent forms right now into coded data use terms? And then are we also accurately able to do the same when a researcher writes a narrative research use statement? Are we able to actually structure that with those data use ontology codes accurately? And then ultimately, once we have those computable versions, does our matching algorithm in the data access committee comprised of human data access committee members end up making the same decision? And it's this process that we're going through iteratively, again, internally at the Broad with our own data access committee and also with a growing number of NIH DAX. And just to kind of give an overview of where this has been and where it's been going, is that really we started DUOS in 2018 at the Broad with a pilot data access committee. Then last year in 2019, we had two NIH DAX, NHGRI and NHLBI. There's some internal testing and piloting, some previous requests. And then this year, I've had some great opportunities to work with driver projects at GA4GH, other DAX at NIH, and a huge increase in volume at the Broad as well. And really excited to say we're going to be moving forward with an official NIH pilot of DUOS for up to six DAX coming up in 2021. So happy to share more and talk more and you're welcome to visit duos.broadinstitute.org if you'd like to play around with the software. And that's it for me.