 Okay, I think that we are starting the session with a report that reports back from the breakout rooms. And I think the first moderator with the report is Adam Adam let me know if you need me to show some of the notes that I took or what are you do this. Let's see how good my memory is and if I miss anything please chime in or bring to make sure that. You know, so we had an awesome discussion and I think he's so much for all the discussions and the input. And let me start by just going through a set of essentially strengths we can sense opportunities and threats but then also try to connect them across that. Some of the strengths really in a reverberated what you saw in the in presentations in the past and and we had a chance to review which is really that. The Annville is a premier example of the use of the cloud on behalf of researchers in ways that allows them to scale activity accelerate the discovery process. Recognition that as and will grows the discovery landscape is going to really expand for those consortia that participate in the use of and bill because they'll be able to interconnect across their own data with data in the end bill. In that strength is also challenges really positioned on the weaknesses which are that in order to interoperate. You know, while the genomics front, you know, I think has made tremendous headway people pointed to really the phenotypic and data model landscape is still being widely diverse in ways that pre built interoperability for search or data analytics is still going to be a challenge but a place where the advent of the cloud will continue to push innovation in that setting. Other strengths are that the availability of those cloud based resources really democratize access. You know, to all types of consortia independent of their local or local resources or differences in the DCC themselves and the capacity to support analytics. And that within that setting it also provides a robust training opportunities for those consortia and the networks that you know the PIs and their students as a tackle domain specific and stakeholder specific interest have already made environment for training in ways that trainees can carry forward and expand independently. And that landscape of intersection with consortia also brought up, you know, really a lot of use case that I had not, you know, fully thought about which is that whether we're willing to always say out loud or not then I appreciate Alex, so transparency, you know, science is a competitive on behalf of the discovery landscape but it's also competitive on behalf of the resources that the NIH is provisioning on behalf of building out the data ecosystem. Competition on behalf of discovery. I think this custom is really pointed to, you know, the strengths of the animal really that, you know, being promoting competitive practices around who can better use the cloud. In some respects, along with the NHS mission to accelerate discovery, because those who can compete to essentially analyze data faster, better, more broadly, essentially advanced discovery, you know, more broadly in ways that could potentially be non local lead constrained. However, the emergence of the anvil and many other platforms also raises the risks that the diversity of DCC based resources, you know, might be a jeopardy, meaning that, you know, that's how many DCCs are going to use just the anvil is the anvil really the DCC or some of the questions mitigating those risks are really opportunities that have emerged under the NC PI efforts where, you know, the anvil is heavily positioned as an interoperable landscape of use, where essentially interoperability with the anvil with other emerging platforms or data resources cloud or otherwise potentially still maintains the potential for diversity in the data ecosystem landscape on behalf of DCCs. The reality is that the data landscape is still too complex in order to simply rely on technology to support advancing the data stewardship needs of consortia and that you ultimately still need DCCs to store this, but that relationship potentially is beginning to be designed and optimized. A key challenge and threat that is not unique to the anvil that we spend some time talking about is cost. Cost is a recurring narrative again not unique to the anvil cost as a DCC cost as investigators want to use the anvil transparency around those cost bases and related to cost are really the challenges of sustainability so even as it relates to the engagement of the cloud use itself. Even if you onboard the cloud you have to make decisions as a DCC or as a user as to the sustainability of those costs in ways that will not require you to then shift or lose the gains positioned in the cloud and it really is the primary way to mitigate that threat is through really NIH engagement and developing models for the use of the cloud in ways that really advances and presses on the value of the cloud and the anvil provide against the locally provisioned typically subsidized use of HPCs within institutions. Together I think these frame a common set of narratives that bridge across the anvil specific landscape in which the Hopkins and Broad team have done an amazing job of really prioritizing the strengths and opportunities but recognizing that these are still early on and that they will be exponentially driven as more data comes in as standards begin to be further established across our social data sets to drive the need to implement them elsewhere. Well at the same time, in an anvil non specific way, there are challenges around the use of the cloud as it relates to the current landscape. I'll end with really one of the cross cutting themes that I think is a really innovative space for the ends of anvil beyond tools and interoperability and that is how can these types of resources like the anvil impinge on the translational impact for today's patients, how can the clinical process itself be advanced through these resources. If the premise is that NIH research drives new discovery on behalf of health, can we bring that into the clinical domain in a much more proximal setting and what are the challenges around that here. Some challenges but one that can be converted to opportunity are one mechanism that essentially drive trust between healthcare ecosystems and the anvil in ways that are transparent and engage the clinical use on behalf of value impart to patient historically hospital systems and the other mechanism that arises eventually. If there's benefit to patients will find ways to leverage those resources on behalf of those patients because in many ways they themselves are competing on behalf of both the disease landscape itself and the care of patients and that's a narrative that will continue to evolve. We're still going to be challenged by current modern ecosystem that again bring to bear models data structures and the healthcare ecosystems overall challenges as relates to their own infrastructure around genomic clinical data imaging data where they their coordination of those are challenging on their own within that institution. And this is where the anvil has already made some some strides by at least developing tools that are clinically useful for the clinical environment and making those available to the community for standardized use implementation. Valentina that's my stream of consciousness memory of the discussion discussion let me know if I missed anything. I think you touched all the key points. Thank you. Thank you, Adam. This was excellent. To remove to Mary. Sure. All right, I'm going to share my screen so that I can share the slides with you. Hopefully you can see those are I can thank you so much for making the slides while the rest of us spoke. So we had a great discussion in our group and I think you'll notice that some of the themes that Adam just talked about from the other breakout group came through also in the context of tools. So first for strengths, where does anvil excel. If we heard from several in the group that the documentation and the tools available in the workflow setup is done very well. Many in our group said that they have members of their team that have been able to get on there and learn pretty quickly within a matter of days how to use the tools in the in the system, whereas just going to a different cloud based system on your own can take the data access and the accessibility that's there now is is easy to use and and folks are happy with that it's a strength. I would say it's good to see it here. We discussed lack of fairness, which is machine, you know, programmatic access is not in place this is the user interface based access. No fairness. Thank you for that. The variant interpretation tools and workflows and the ease to develop new tools in the workspace was seen as a strength. The plans that the end of the team has in place to allow third party groups to build on the platform is also seen as a strength and the way that they have put security first in regards to third party tools and workflow plans was seen as a strength. It was mentioned that, you know, there's a lot of concern in the cloud and the way that they have things set up makes it much easier for folks to bring their tools and get things set up and just go into a native cloud system. While they've done a great job we identified a number of weaknesses. One is developing tools for analysis on open access data set so there are some data sets not there and having the tools to use those would be would make things better. There's a lack of tools for single cell analysis. Specifically, while a lot of the documentation is great there needs to be improvements specifically related to doc store. It was noted that there are a lot of tools and so it's hard to find tools and workflows that have already been developed and so coming up with an improved strategy to both curate and search through the tools and workflows that are already there would be useful. Tools for interpretation of snips and better annotations as well as tools for mediation analysis and Mandela randomization I think co localization analysis was also mentioned. And those are currently not there and that was perceived as a weakness tools that allow for analysis of clinical data that's built using clinical data models such as omop. And thinking about things like PKB algorithms so these are the rule based algorithms for electronic health records. If the data tables were put there for omop clinic, omop based or modeled clinical data, being able to deploy the algorithms would be would be great and currently there's no capability for anything like that, as well as Sam tools that was also mentioned. The mechanism to anonymously log in is not there and improve mechanisms for feedback was seen as a weakness on two opportunities where can handle grown improve adding additional data standards and data models so that we can improve the interoperability model for an individual patient so as we look to the future of and though and more clinical data are there, being able to take an individual kind of set of characteristics about a patient or a machine learning model and matching them with other patients in the envelope would be a great opportunity. Being able to link necessary tools to expand and diversify the envelope user community is also another opportunity and this is something where we think this will come up again in the afternoon session around outreach and training. But as the envelope works to bring that the tool and the capability out to a broader user base so these are, you know, not our typical users of something like the envelope. We will learn a lot about what tools we don't yet have that would be useful to that that broader user community. I'm creating a safe space for groups that are hesitant to host their diverse data sets and public repositories. Data sets were mentioned like the MVP. Some countries would be uncomfortable hosting their data. There are some other kind of protected groups that if there were a way to have the data kind of, you know, in this system but safe such that individuals can't access the individual level records but still be able to run analyses and aggregate across the data sets. That's an opportunity. Having tools that accommodate the admixture and diversity of human genetics data sets as well as the new reference genomes that are going to be available and making sure that they're backward compatible with the older reference genomes is another opportunity as we see the diversity of human genetic data sets increase in the future having tools available to accommodate those would be important. So tools tool based use cases to encourage Anvil being accessible to non cloud experienced groups. So, especially, again, this will probably come up in the outreach session but as researchers from institutions that don't have cloud experience and perhaps don't have capability to even host these large data sets. The benefit of having Anvil, but then having use cases and tool based use cases for those individuals to learn is going to be important. Expanding the tools that are tailored to the clinical genomics community for clinical decision support, as well as tools to facilitate going from crams to variants to the interpretations would be seen as an opportunity. So tools to support training on the anvil for both basic science researchers as well as clinical genomics researchers. I think the user base is likely to expand in the future and having the tools there to support the training would be important. And then finally the threats what jeopardizes the future of Anvil cloud costs which we heard from Adam. This is something that that we're going to need to be able to deal with difficulty in facilitating the culture shift in Anvil cloud. It's, it's a new world. This is the future, but getting people on board is, is a threat. There are challenges in making the tools and resources in a manner that meet users where they are. There are sophisticated users and I think they're going to be able to jump onto the platform and use it and then they're going to be users who need tools to make the tools that are accessible to them. And that that's going to be challenging for the team, because the team is sophisticated in their, in their capability difficulty in making Anvil interoperable with other platforms. One that came up is the research analysis platform for UK Biobank which is built in another cloud system. So as more people get comfortable with that system, the tools there are different from the tools being developed in Anvil the actual kind of algorithms and methods are different, but how they look and how they are deployed is different. Hurtles required to access Anvil just to test the platform. So if there is a big learning curve, you know, are there ways to make it easier, even to just know that it's a good use of time to use the platform. And finally, as we do think more about clinical data and clinical genomics reports and doing those annotations and interpretations in Anvil, thinking about liability and how to deal with that is the final threat. So I think that was everything that we came up with. Ken, feel free to chime in if I've missed anything from our group. No, you did an excellent job. Thank you. Sure. Thank you. I think we have even two minutes left session if anybody has any question for Marilyn and Adam Titus. I'm going to do the this is more of a comment than a question thing but I put it in the chat it's just, you know, everybody wants Anvil to do everything. That's not possible. So to me the existential threat for Anvil is how do you decide which things not to do and how do you, there's a word, communicate that within the landscape where everybody wants you to do everything. And I just want to acknowledge that that I think this is something Valentina and Ken have been struggling with. I just really want to bring that right out there and put it right up front as if you can pull off this hat trick this this type type of walking trick. It will be awesome. It and Anvil is already serving a lot of amazing needs. So I think it becomes more a question of communication and strategizing than than anything else so. This is a very good point. Yeah, and something that you highlight Valentina are always constantly trying to figure out the best ways to mitigate. Yes, I just want to follow up on this comment. I think one way to potentially do this to actually empower additional external users to come in and really make Anvil their home. And that might, you know, so so shifting the responsibility primarily from the Anvil team to do others in the Excel community like having a much wider base of developers I think is an effective strategy and maybe that's the line to draw as to what what is core responsibility and what is like could be offloaded to close collaborators. Let me just make one last comment and both the comments just came up. And one of the things that I think at least for me is exciting is that, you know, we're still living in sort of the lands of metaphors that relates to the cloud meaning that there's an instinct to try to relate what is happening in the Anvil to what happened before, where in reality, we're defining new ways of doing research that did not exist before. And we haven't yet fully transitioned into really, I think at least my sense is as a scientific community, what is really possible in the context of a decentralized cloud environment that supports rapid scalable compute access search. You know, so there I guess my caution here that I don't think we have to define everything that the Anvil needs to be because some of it will actually, there's a risk of creating boundaries around what's possible and some of it will emerge, as long as we can get people to use it in the first place. And that setting in, I think new and hopefully wonderful ways. That's the positive spin on Titus' cautionary tale. Marilyn, do you want to add anything to it? No, I agree with that. I think that was the great yin to his yang or yin to his yang. Okay. It's 2.16 according to my clock, so why don't we adjourn. We have a 15 minute break. We're going to be back at 2.30 and that's when we're going to start the second session with the two other two breakout rooms. So enjoy your break.