 All right, let's go ahead and get started. We want to have plenty of time for discussions. So good morning, good afternoon, everyone. My name is Guy Palmer. I'm the committee chair for the Wastewater Surveillance Committee on Community Wastewater-Based Infectious Disease Surveillance for the National Academies and Stephanie Johnson, the study director who we've interacted with, is here as well as several members of the committee who all have introduced themselves in just a minute. Next slide, please. So just very briefly, and I think probably all of you know this, is this is a two-phase study. The first phase was released in January 2023. And it really built that really focused on was there value in what we learn from the rapid startup of wastewater surveillance due to the COVID-19 pandemic and emergency response, and is what we've learned there and the benefits worthwhile beyond COVID-19. So that kind of laid out just basically an outline and structure of what a national wastewater surveillance system would look like. So in phase two, we're going to hang a little more detail on that phase one. And the meeting we're having today is particularly important because it really addresses one of the key questions and challenges that the committee has to work through. And that's really looking at analytical methods for wastewater surveillance, what's currently there, what can be used in the future to improve that, and also looking at not only the known pathogens, obviously a lot of the information we've learned has come from COVID, but how now we can apply that to both more rare or infrequent pathogens or even unknown pathogens emerging into the human population and therefore picked up in wastewater. So it's really exciting to have your input. One of the goals we have is to essentially future proof this report so that it's not dated in two years and then no longer has benefits. So we're really looking very much to the future. So we've laid out the statement of task. We're really focusing today on analytical methods. As I've mentioned, that the full statement of task is obviously present on the website. Next slide. So our committee really has a range of expertise from really detailed understanding of how municipal wastewater systems work through the analytical methods, through how that data is then managed and reported into the public health system. And I am fortunate to have multiple of the individuals here today. I'm gonna go in order of which I see them just have them introduce themselves very briefly. Ami, you're the first that shows up. Hi everyone, my name is Ami Pad. I'm an associate professor of medicine and of genetics at Stanford University. I mostly focus on the microbiome. Thanks so much for being here. Christine Johnson. Hi everybody, thanks so much. I'm Christine Johnson, a veterinary epidemiologist at the University of California, Davis One Health Institute. You Raul. Hi everyone. I'm Raul Gonzalez. I work for Hampton Rose Sanitation District in Virginia and I'm a scientist there. Thank you Scott. Hi, I'm, problems with camera and video. I'm Skameshky, I'm a professor and associate chair in environmental occupational sciences at University of Washington. Rekha. Hi, I'm Rekha Singh. I'm serving as waste water surveillance program manager for Virginia Department of Health. All right, thank you. I'm scanning through to see if I missed anybody. It's an imprecise system. So if I've missed you, any of the committee members, please introduce yourself at this time. Okay, hearing none, if they join later, I'll make sure that they get introduced. Next slide. So I think without much further ado, we will progress with this. The first part is advances in research needs and analytical methods for wastewater surveillance. And after this, we'll have time for questions. And I kind of beg your forgiveness in that I really want the committee members to have first shot at asking questions. But if we have more time, I'm happy to open it up to a larger audience, which also is helpful for us to hear your questions. So I think we're gonna switch this order very slightly and Marlene Wolf from Emory University is going to be the first presenter. Great, thank you. I'm gonna pull my slides up and share them right now. Okay, can you all see? Perfect. Great, okay. Thank you so much for having us today and I'm looking forward to the conversation. I'm gonna talk a little bit about both pre-analytical and analytical methods that results in the data that we are looking for, for these wastewater surveillance programs. Allie Bame from Stanford is also here representing us in our group. So I'm gonna talk a little bit about the perspective of what we have learned in our work in wastewater scan. So wastewater scan is a wastewater monitoring program that's based at Stanford in partnership with Emory. And so Allie and I are the program directors for this initiative. We have a network of 185 sites across 36 states. You can see the map here of where we're located. And our goal has really been to use consistent methods and to improve methods for monitoring infectious diseases through municipal wastewater solids. The goal to expand wastewater science and to inform public health response and to do so using a network of geographically and in many other ways, diverse sites from across the country. We work closely with Verily as our laboratory partner who as academics don't have the capacity for all these sites in our labs. And then also with the NLC and then many public works and state and local public health partners to also make the most use of this data. So that's who we are. And as we think about sort of what we need from our analytical approaches to meet those goals that we have for what we want to develop and show that wastewater monitoring is hopefully capable of doing. I wanted to kind of outlight some of those things that we need from our analytical approaches. So I split these into things that are more on the side of the pre-analytical methods and things that are more on the side of the analytical methods. So some of the features that we really prioritize were sensitivity, so making sure we're using methods that consistently provide results for the disease target. This is also pre-analytical and analytical but a lot of our work has looked at pre-analytical approaches to improve sensitivity. Comparability, so that especially when we're looking at quantitative results, we have consistency and comparability across sites and across time. Flexibility, so a process that can be utilized for a range of pathogens because that's one of the beautiful things about wastewater monitoring is that we can monitor for many things in a single sample if our methods are suited to those things that we're trying to monitor. Reliability, so we wanna have methods that are sensitive and specific for the disease target. We feel that it's very reliable that we're detecting what we think we're detecting. And then also utility, we wanna make sure that these methods are designed to produce data and that we can verify that they're producing data that's related to the disease outcome of interest and then therefore possible public health use. So these are some of our priorities and I'll try to kind of talk about how those come through in the work that we've done to develop more methods for wastewater monitoring. Quickly, this is our overall process right now which is again based on using the solids from wastewater. We have a pre-analytical step where we dewater the solids and then we take that pellet and we resuspend it in a buffer solution with grinding balls and beat it up so there's chemical and physical lysis happening. Then we do a nucleic acid extraction from that buffer and we do an inhibitor removal step. And then we use Droplet Digital PCR and we multiplex for a number of different targets which I'll show some examples in a few minutes. So that process is consistent for all of the samples that we're running for the program. And the reason that we've chosen the solids is because at the very beginning in March of 2020 we started doing work looking at where we could get the best results from wastewater for SARS-CoV-2. And many of the targets that we're interested in including SARS-CoV-2 naturally concentrate in the solids. There's previous work that suggested that this would be the case and what we have seen very consistently is that there's a thousand times higher nucleic acid concentration in solids compared to liquids. This is a recent paper from Ali and her student Laura demonstrating some of that partitioning in wastewater. And of course the question is, well solids, how do you get the solids? There are solids in all wastewater and so this is just to show that while it is nice to get a sample from maybe a primary clarifier where a lot of solids have settled out there are solids also in liquid, influent wastewater or wastewater that you pick up from anywhere and we can use those solids for this same process. Again, we've consistently seen that these targets are more concentrated naturally in the solids versus the liquids. So this is just, I'm gonna go through this slide quickly but this is just to show you that there's many, many examples as we've added new targets. We've looked at this difference in measuring in solids and liquids and consistently we've seen orders of magnitude higher concentrations on a per mass basis in solids which are the top in all of these compared to liquids. And that's consistent for viruses of many different types and morphologies. So the solids have really contributed to our priorities in terms of important considerations for comparability because we're focused on having this national network. So some of the things that we have prioritized are the use of a single pre-analytical method which for us is the solids that supports comparability across all the sites. You saw in that previous slide big differences in the estimates we're getting we're using the same assays for all of those, right? It's the same digital PCR process but because our pre-analytical methods are different the results that we're getting are different. For quantification of biological analytes even the same method will produce often variable results from different laboratories. So working with a laboratory team that has capacity to handle all the samples for this program has also helped us to have good comparability among our samples. And then selecting the solids allows for a high throughput simple approach flexible use for monitoring pathogens that are also not viruses like bacteria or fungi that may even meet the definition of a solid in wastewater and therefore be found in the solids. And then also expressing the results in dry weight of solids and also adjusting for PMMOV improves comparability across sites and time. You can see that a bit in these plots which are sites from all over the country and you can see that they sort of have an even more consistent relationship between the wastewater measurement and cases as you normalize for PMMOV. So now a little bit about analytical methods. So we found it really important to have a process for new targets and developing those and including those in the program. The basic process that we go through is we have a new pathogen or variant that's selected and that we get that information about what to prioritize in conversation with public health and others. So we want it to be a priority and we also want shedding into wastewater to be likely. Then we'll either find an existing assay or design primers and probes and check for specificity in silico, procure the materials to do the test which is an important and sometimes lengthy step then test the assay specificity and sensitivity against other viruses and the positive control and then we'll apply that completed assay retrospectively to a time series of samples that are gonna be useful for helping us understand how to interpret the data. We'll compare that wastewater data to relevant clinical data and then we'll pilot prospective application of the testing, observe performance, get feedback, socialize it with all of the stakeholders who we know are gonna be using the data and then make a decision about whether we will actually include it in our program. I'm gonna skim through this but this shows our timeline of program targets as we've added a bunch of new targets, new diseases to our program over the year. So you can see we've added bit by bit as we have gone through this process of developing and establishing the kind of supporting information to allow that. And this is where we currently stand for our targets. We have just added six new targets to the program actually. And so we have a pretty significant panel of respiratory viruses, six different respiratory viruses including influenza A and B, enteric viruses and then other pathogens including Mpox hepatitis A and Candida auris. And we also measure these controls in all of our samples and all of this data is publicly available on our dashboard. I just wanna show how the process went for a few of these targets that were and were not added. So for influenza, which was one of our first additions to the program in January of 2022, we selected CDC primers and probes and tested the sensitivity and specificity. We analyzed wastewater from pre outbreak and outbreak periods during a time when there was prospective monitoring on a couple of university campuses. So you can see here data from Ann Arbor and then we also looked at an outbreak on the Stanford campus. So we compared to this high quality clinical data and we saw that the measurements in wastewater were associated with incidence rate and public health partners were very interested in this data and excited to see its potential use. And so given all of that supporting data, we included it in the program. On the flip side, we also looked at enterovirus measurements. So we selected a pan enterovirus primer probe set tested sensitivity and specificity applied this to wastewater samples over a two year period, but there really is a lack of clinical data for comparison. This is a diverse group of viruses that are captured. There's many different clinical manifestations that they can result in and there's consistent detection at fairly high levels year round. And this just didn't sort of produce data that had a lot of interest from public health or that we could really help to interpret or make actionable. So we did not select this for inclusion in the prospective program. And then EBD-68. So we looked again at some primers and probes tested sensitivity and specificity analyzed wastewater over a two year period and compared it to California case data. And what you can see here is wastewater measurements from two different sites. And then EBD-68 cases in California. And we saw that there was a strong association between the wastewater in the state, sorry, the wastewater measurements and cases in the state even though no cases had been identified actually in the sewer shed area. So there's very strong interest in this from public health because there are difficulties associated with traditional surveillance for EBD-68. And so therefore this could be particularly useful for keeping an eye out. So we just added this to our program this month. Just a quick note, when we're adding these new tools into practice, we really focus on early and often communication to hear preferences and feedback from partners to decide what we're gonna investigate. And then also before launching anything, we've worked on preparation of talking points and fact sheets and releasing our research results and protocols rapidly so that they're there for support and interpreting this data. And then we also support and collaborate with public health to examine and interpret the data after launch. And this is an example of some of the data as you can see it on our website. We have added a new feature where you can look at categories for data. So you can actually see based on levels and trends what the category is that the wastewater is showing for some of these different diseases. And you can see right here the sort of beginning of an uptick in influenza in the Bay Area in the past month. So our key goals are to integrate with public surveillance by doing new research to show again this relationship between these measurements in wastewater and some key elements of surveillance that is commonly used in public health, having these website tools to show categories to aid interpretability and to collaborate to incorporate data over multiple seasons to build use cases. Because again, these analytical approaches have to result in data that we can demonstrate is useful and interpretable. So quick takeaways, choice of both pre-analytical and analytical methods is necessary for comparability and consistency. And even differences in laboratories can make a difference for this type of analysis. Use of wastewater solids supports sensitive measurements, high throughput and comparability in a structured process of choosing and developing new targets and the evidence to support them is important to have that evidence base to include these targets in prospective monitoring from wastewater. And this is our team. Thank you so much. And I look forward to more discussion and questions. Super, thank you, Marlene. That was excellent. I'm sure there's gonna be multiple questions on that. I know I have a few, but I think we're gonna hold those until after the next speaker, which I think, Mariana, are you ready now? Yes, thank you. Super, all right. You can take over. Good morning, everyone. This is Mariana Matus. I'm co-founder and CEO of Biobot Analytics. We are a wastewater intelligence company fully dedicated to wastewater epidemiology since our founding back in 2017. My co-founder Nusha and I met while we were both at MIT. And by the time that we founded Biobot, we had been doing research on wastewater epidemiology for about three years under an interdisciplinary project run at MIT between six different labs. With COVID-19, before COVID-19 started, we were focused on the use of wastewater epidemiology to improve the space of substance use and to tackle the overdose crisis. But in early 2020, as COVID-19 started, we saw the opportunity to leverage the platform that we have been building to then start looking for SARS-CoV-2 in wastewater. We were the first report of detection of SARS-CoV-2 in the US, second one in the world by a couple of days. And we reported those very early results via preprints, which then later on became as well scientific publications with our methods. Since then, our priority has been how do we scale wastewater epidemiology to create a nationwide network of data? That is the vision that has always been our vision from the beginning, is to be able to look at wastewater, not just on an individual community basis, but really be able to leverage it to answer questions that feel very difficult right now. Like how is a particular disease spreading? When is it gonna arrive to a particular location? How do we make these communities more resilient? Again, at the national and eventually global level. You can see here a little bit of a snapshot on all of the experience that we have accumulated over the past few years, testing almost 40% of the US population and having representation in every state and several US territories as well. Most of the testing that we have been doing has been for SARS-CoV-2, a targeted analysis of wastewater to understand the level of circulating virus in the wastewater. And we report that via normalized concentration that we call effective concentration in copies of virus per liter of wastewater. The main correction that we do to the data and normalization that we do to the data before reporting is normalizing based on the PMMV virus. There have been other methods of normalization reported, but PMMV has to date proved to be quite effective at normalizing the data and correcting for differences in dilution. I will talk later on the platform that we're using in the lab. But what I can share here is that comparing the data coming out of wastewater and comparing that against clinical case data shows that in the throat of the pandemic when clinical data was robust, wastewater was a leading indicator of clinical cases. Showing a few days and up to a few weeks in advance of what we saw in the clinical data. As the clinical data became less readily available, wastewater has now become a more reliable source of disease prevalence than clinical data. Nowadays, most people are using antigen testing at home. If anything, maybe many people are not testing anymore. So the most reliable source of data is actually the wastewater data. And this is just to make the point that it's really important to complement the wastewater data with other sources of surveillance, for example, clinical case testing to really interpret what's happening. But the relationship between the two is not constant. It really depends, wastewater will continue to be to have its own capture and sensitivity, but the clinical data, actually the relationship with clinical data will depend largely on the quality of the clinical data, how available it is and how much people are using it. So we cannot make a generalization or a general assumption or conclusion as to what the relationship between the two of them are. And this is likely to hold true for other disease areas as well. So we just need to be very careful as to how we interpret the relationship between the two. Early in the pandemic, this was, we started doing sequencing of wastewater, targeted sequencing to identify and be able to call out variants of concern from wastewater samples. The most standard practice in the industry was to leverage rapid development of PCR assays in order to look at different variants as they were emerging. But our approach was to actually expand into another detection platform targeted sequencing in order to be resilient to the changes in variants and be able to call out all of the new ones as they emerge, just purely by modifying our computational or bioinformatic analysis. So you can just see here a time series of the sequencing data of the state of Massachusetts as collected and reported to the CDC, so all of the sequencing data that Biobot has been producing has been posted to the NCBI SRA for public access. So all of this data, you can see here a time series and you can see how just some variants sweep very fast, maybe just in a few weeks. So needing to develop a new assay for a new variant it can be challenging just from an operational perspective to be able to gather quick enough to be able to seed in wastewater and therefore sequencing technologies that are resilient to that sequence change will be more feature-proof to be able to stay ahead of new variants and this strategy will also be relevant outside SARS-CoV-2 into other pathogens, for example, even influenza where we see a large diversity of strains and dynamics between the strains year over year. So this applying sequencing to wastewater can be a very powerful tool to be able to look at that. Just to mention that, you know, with wastewater I think we generally have been talking about conclusions that we can extract from wastewater treatment plants as the sampling location, but there's also an entire kind of area of exploration which is how does the wastewater data help us how is this national network of wastewater treatment plants can get complemented by adding other types of sentinel sites and what new insights we can unlock by designing that nationwide network to be ever more sensitive. So just as one example of that we have been running a pilot at an airport here on the East Coast and what our early results indicate is that airport wastewater data was a living indicator for the rise of the EG-5 variant. So what this is teaching us is as we design a nationwide program to understand the spread of disease in a country it's also important to think about not just the assays that we are applying but also the specific sentinel sites that we need to include in the network and airports, for example, as well as other types of mass transportation sites could be very interesting to consider. Along the way of expansion into new assays sequencing holds a lot of promise for us to be able to expand into a wide diversity of viral and bacterial pathogens. So you can see here some bioinformatic analysis where we made a list of different pathogens both bacteria and viruses of public health interest and we just look at previously published sequencing data in wastewater. Sequencing data has been produced in different parts of the world and what we found is that indeed wastewater was able to have detection for a very wide variety of these pathogens. So continuing the development in parallel of not just our PCR assays but also sequencing base holds promise for us to arrive to the right combination of technologies to get the intelligence that we need. Now, just very quickly following the prompt we want to touch on how we develop and validate assays at Biobot. I want to talk about the analysis of an inter-lab analysis that we do to look at the laboratories of the entire US wastewater industry and ideas as to how we can achieve more industry-aligned data because it's in the best interest of our country to have as most comparability across labs as possible to leverage all of the data together. So starting on how we develop and validate assays at Biobot, these are our rough processes which have to do with designing, reagent acquisition which during times of pandemic time that actually can be quite that can add quite some time to the development if you don't have the right partnerships with the vendors. Initial testing, assay performance characterization then we want to do some white-scale testing with as many variety of wastewater types that we have access to. For example, wastewater coming from different population sizes, different geographic areas fresh wastewater stored wastewater to what extent can we use this assay on different types of wastewater. We are focused on influent wastewater and as we do our processing basically our processing captures both a combination of liquid and solid wastewater to go into the processes and finally tech transfer. So coming out of research and development how do we transfer that into a full production lab that can run the assay on an ongoing basis at scale. I think very similar to what was mentioned in the previous conversation in the previous call we look at multiplexing efficiency right to be able to make the platform as flexible as possible to do different pathogens at the same time, accuracy, specificity limit of detection precision reportable range lot to lot variability and during the white-scale testing it's more about really understanding that the assay will perform in real world wastewater samples of different qualities. I just mentioned that we have spent time comparing different platforms and different options and the pros and cons of different PCR technologies we had by about started with a QPCR platform back in 2020 and have been testing multiple technologies since then and I think that the LDR of this table is just that there isn't necessarily like the one perfect platform to implement wastewater there are going to be trade-offs between cost, the ease and speed of assay development data quality defined as sensitivity and variation and the turnaround time so it's just important to understand that there's those trade-offs between different technologies and we wonder at least our thinking internally is we wonder if maybe a nationwide wastewater lab should have access to at least a couple of different PCR technologies depending on the targets and the objectives of the program or if it's enough to adopt a single platform you know we don't know yet and finally just to mention here that our lab at Biobot the production lab we are building it with quality management at the core our head of lab ops comes from the clinical testing space and she has been building our lab to be able to be ready for future regulation of the space so far wastewater testing is not regulated so there isn't any external agency that will be able to receive documentation and say yes you are testing for X or Y it's not regulated that opens the opportunity to do a lot of assay development very quickly and be able to test things but I think we also need to start working towards an industry where there will be some level of regulation and more control of the quality so on that front we also mentioned that we have been curious about what can we learn about choice of platform and others quality of the data if we look at data produced nationwide and to answer that question we actually have a very nice resource which is a very nice natural experiment which is that CDC News collects data from 84 different labs in the US and they have been collecting this data for now close to two years and this data set is available like if you request it you can actually get access to this data for a study so that's what we did we requested access to the CDC News data set we want to thank CDC News for allowing us to do the analysis on that data set and also for their permission to present those results here today as you can see here the time period that we selected to analyze covers June to December of 2022 so it was a time period where there were high levels of SARS-CoV-2 in the population in the US and based on that data we developed a couple of quality metrics related to variability and to sensitivity basically variability defined as to how much fault change do we see week over week and sensitivity defined as again this was a period where there was a high level of SARS-CoV-2 so we expect most of the samples to be positive if we see a high level of non-detects that would be an indication that the laboratory method isn't as sensitive as we can see here a plot on an example of a location or a lab that is reporting data on a location where there's high variability of data week over week low variability of the data week over week it was interesting to see that there was a very wide range of data quality across different labs IDs in use the lab IDs are blinded so really the only variability ranged up to 3x week over week there was a high level of variability and sensitivity similarly ranged across two orders of magnitude so this was just interesting to see there was this much level of variability and sensitivity in the quality of data being reported again not surprising given that every lab is 80 plus can choose to do the work as they see fit Mariana I'm going to interrupt just briefly could you try to wrap this up in the next two minutes because we want to have time for questions yes absolutely all right thank you so I'll just mention that the effect of platform didn't seem to influence as much this metrics of samples processed by the lab seem to matter more and that you know one option of course should be to try to create more guidance for the laboratory methods that are happening but another way an alternative way to complement that standardization of processes would be to start to align data computationally and on that front we have been working on algorithms to do automatic quality control and alignment of data coming from different labs because what you can see here in this plot is that data coming from different labs actually tends to match it's just the baseline level in which they operate is different given the differences in methods so it's possible to actually bring them closer together to make them more comparable computationally I'm very excited about the potential of using some of this to continue to enable a decentralized approach to testing but still achieve industry-aligned data that we can leverage as a country so thank you super thank you very much let me open it up for questions for either Marlene Ali I think is also here let me start with the committee if you raise your hand first off thank you both for your presentations we're both excellent Marlene I want to commend using D68 I think that's great that's amazing data that was one I was pushing for from the beginning I do of course Ali can already guess what my question is going to be around the PMMOV and the normalization in your methods are you finding you're doing both dry weight and PMMOV and does that correct for smaller populations because I know PMMOV seems to work above what a few hundred thousand but in lower populations a little more variable yeah we see good comparability of the raw measurements in the copies program dry weight actually PMMOV is very useful actually even without PMMOV normalization but we also have seen benefits especially when we look at time series and sort of minor adjustments then recovery for example in the process over time PMMOV has really supported that there is a little bit more variability in small populations that are constantly useful I know this is a big topic of conversation right now among all of us is what is the value of normalization is it normalization how do we talk about making adjustments to the data by using other pieces of information whether it's a measurement of an endogenous marker or other approaches to making those adjustments I think that those can be helpful and consistent and comparable actually even without some of those extra adjustments so from a quick follow up do you find with PMMOV that you find a similar ability to normalize for the non-enveloped and other pathogens that you looked at yeah that is what we have observed I think it will be interesting as we expand we just added candida oris which is a fungal pathogen we have new things to explore as we continue to look at targets that are more different from some of the controls that we've been using so I think it's an ongoing conversation and process to evaluate it but yeah so far we've still found it to be useful Ali I don't know if you have anything to add on that no but I think we need to change the language from normalization to data adjustment and we need to make sure that we're explaining and we do it in manuscripts and in talks yeah and you'll notice I put adjusted for PMMOV on my slides not normalized so yeah I think that's important right thank you questions if not I'm going to sneak one in which I'm it's probably going to come up later as well which is how do you look at not only detection but the reproducibility of rare pathogens that you know you're not quantifying as much as you want your answer to be plus minus it's here or it's not here so probably an example would be if we went back to the original SARS-CoV that CoV-1 now now called CoV-1 you know that spread had very limited spread you know Toronto and some other places or MERS-CoV or we can think of other pathogens you really trying to say is it present or not what kind of approaches would you do to kind of have reproducibility in that yeah I can answer but sure so in our case I think one example of that was Mpox it was way less abundant in wastewater samples than SARS-CoV-2 at the time so what we had to do in order to get to reproducibility and to trusting our negatives to be negative is to increase so we created what we call like a low prevalence workflow in the lab so basically we increase the volume of wastewater that gets processed and that gets concentrated to look for Mpox and that basically allowed us to have an assay where we trust our negatives to be negative and has enough sensitivity we tried other ways as well but what worked is basically to then think about a high prevalence workflow and a low prevalence workflow and for low prevalence workflow basically that means just using more volume of wastewater maybe I'll add to that a little bit first of all I think that there's a real divergence between the reproducibility aspect with clinical data versus environmental data and I think when you're dealing with rare targets as many of you on the committee have Scott, Raul your experience working in environmental surveillance for many decades you know that it's very easy to get a positive result on one sample and a negative result on the next sample even if it's from the same location so I think in wastewater scan we actually have three archetypes of pathogens as Mariana suggested one of them is these are these rarely detected pathogens and we're learning that in order to understand what the wastewater data is indicating with respect to contributing infections you have to look at frequency of detection and as that frequency gets higher and higher it means that you're more and more likely to have like some real potential transmission going on and I think it's important to have these rarely detected pathogens in our data so that we can learn to understand how to interpret the data and as I understand CDC has taken all the Mpox data in news and done some analysis as has CDPH to really understand what these intermittent detections mean so sorry it's a bit circular but I think we should be careful not to ask too much about too much of our environmental surveillance to provide reproducibility because we shouldn't expect it to always be reproducible due to the fact that we have a rare target it doesn't mean that our environmental surveillance isn't working it's just the nature of the complex matrix and the rare target and I'll just go ahead Marlene one more sentence I think that's why we focus so much on the reliability and the sensitivity to make sure that we can detect something that's there at low levels and that we have done our testing to sort of very reliably be able to say yes that is what we think it is we are detecting that it is specific to this and then interpreting it along lines of everything Ali just described you know I'm actually kind of more worried about false positives and then you retest and you know the variability and then how that translates into a public health response or lack of so yeah on that note I also wanted to comment on one of our experiences is that yeah MPox has been interpreted so far as detect non-detect given the the low prevalence of the pathogen and we did experience you know we have been testing for MPox for now over a year and we did experience one instance where you know the lab thing happened of a primer set was contaminated and we have three false positives for MPox given the the low prevalence aspect of it it triggered an immediate public health action in that local community which then you know once we fix it once we address it we discover it a couple of days later like they have already reached out so probably I think also for low prevalence pathogen there needs to be an added level of just process quality you know and and also to some extent a little bit more sober to on the public health side maybe it would have merit you know a location that had been negative all along and suddenly had a positive would have met it a quick call with us to double check before the communication goes out so I think like for those red pathogens specifically creating more SOP because you know despite best quality just sometimes these things still will happen yeah guy I wanted to add you know in our work with CDPH in California they actually are more worried about false negatives than false positives so I think that's something to consider as well like are you like where do we fall on that you know all right Raul I do see your hand but I think we're going to move on but I will call on you for the opportunity we don't want to cut him short No problem. Kyle Bibby from University of Notre Dame who's going to talk a little bit about establishing baseline information great thank you so much for the opportunity to talk today so I am going to be far from comprehensive and based on talking with a few folks try and offer some more thought-provoking opinions on this topic that might be helpful for you all in your report so my personal journey in this sort of space began during my PhD work where I was looking at viruses in sewage sludge which at the time we were interested in human exposure subsequently not for tracking diseases in the community and of course we saw things that we might have expected like adenoviruses, rotavirus is all sorts of adenoviruses but we also saw lots of things that we had never even thought of and we weren't sure really what to do with this is from metagenomic sequencing shotgun metagenomics so we could have gotten almost anything we saw coronaviruses at that time seasonal coronaviruses and we saw other respiratory diseases like rhinoviruses and so that really opened my mind to thinking about the diversity of pathogens that could be in wastewater and we've been working on that topic ever since. So as you all know there's a fairly high temporal diversity of any signal we might want to look at in wastewater and this gets to the point we're talking about data adjustment, data normalization what we see there and I think it's really important to take into consideration when we're developing new targets so I've heard statements to the effect that sampling once per week or a set number of times per week is adequate and I think that that opinion has been developed in response to monitoring for SARS-CoV-2 but I think that it really might be very different for other pathogens and we're going to have to be very careful with how we adjust for that and we even see very high variability within day and so this might get at some of this most recent discussion where we have very low level so this was a study led by Aaron Bivens at LSU who I think talked to some part of this panel recently and this was again in May 2020 so you see here of course there's very low prevalence of SARS-CoV-2 in the community and depending when we would sample you'd either get detects or non detects and so again I think when you sample how often you sample should be sort of informed by the prevalence of that disease and not informed by prior experience with other pathogens I think there's also a potential that you could miss based on your sampling scheme so this is sampling over a three week time period just here locally and I could have actually picked a couple of different examples of this this is we saw a spike of adenovirus 61 which is a mild respiratory infection adenovirus group A but if we had sampled at the front or back end of that week we wouldn't have seen it at all and so I think depending on our goals that might inform sort of our sampling scheme in addition to prevalence and abundance of our target pathogen I didn't want to talk a lot about concentration methods we published a paper recently but I think the concentration methods really bias your results in what you see and they can even drive what you see as baseline so here we did some sort of the genomic sequencing and I just picked papilloma viruses because it shows this point nicely if we do direct extraction and this is like a pool direct extraction to try and account for sample volume we see far less diversity on the left than on the right where we use this nano trap concentration technology and I'm not advocating for nano nano trap specifically I just think it's something that we need to be aware of when we're talking about new targets as much as we might want to be done with the concentration discussion it's not really one size fits all different pathogens with different physiologies are going to respond differently to different concentration methods and that especially matters when they're very low prevalence so if they're high prevalence and you have some percentage difference you're still going to see it and probably consistently but if it's a new target or if we're looking for these rare targets to establish a baseline it's really really important another really important discussion point that I was hoping to make is that wastewater based surveillance or WBE whatever we want to call it is really not one size fits all and I think there's a couple of main factors that drive that and so I heard earlier a couple of talks were talking about wanting to select a target that is shed which is of course good but the actual concentration that is shed really matters so these are two different papers where we did some model based evaluation of monkeypox and Zika and they line up here and there was some data on shedding available from those and these papers we actually account for urine shedding and saliva shedding and anything we can get data on and you can see that monkeypox has a higher concentration of shedding into the wastewater stream than Zika virus and so we've seen we've done this for many other pathogens and it's really sort of drives the suitability that you might have for some of these targets and I think that that should be a bigger part of the discussion I think frankly I don't know if lucked out is the right word but we we got somewhat lucky in terms of wastewater monitoring with COVID that the shedding direction or shape or amount is what it is and you know if it had been a little bit different this whole discussion might be a little bit different. In addition to the shedding of infected individuals of course case rate plays a really important role right so those together are going to dictate the amount of genomes that are available in the wastewater stream to detect another really important point that hasn't gotten a lot of discussion is that the wastewater flow rate or use of wastewater or use of water that enters the wastewater stream is going to be a main driver because it dilutes those genomes that are entering the wastewater stream this is going to matter a lot if you begin to look at other countries the US uses more water than some other country almost anywhere else in the world and so if you look at other countries all of a sudden the calculations might be a little bit different because you're going to have maybe even 10% of the total water use and so that signal is less diluted of course process limit of detection is going to drive some of this and that's something that we can work on to try and improve detection of rare targets but another aspect is the number of PCR replicates and so we're talking about at these very low prevalence pathogens where you might have a single detect of course you can drive the probability up of having a single detect if you're running 8 or 12 or more PCR replicates depending on your resources and goals so all of those are considerations in this and this is one way we like to think about it if you imagine these bars again are shedding by an infected individual into this sewer so if you have increased shedding less wastewater flow increased infection rate that's going to sort of drive you over this blue line process limit of detection if you increase your PCR replicates that can take you closer to that blue line and if we improve our methods and decrease that process limit of detection again it improves the probability that we're going to get that detection the last point I wanted to make is that I think I've heard several discussions about using wastewater surveillance to identify a pathogen X or to identify truly novel targets and in my opinion that is going to be extraordinarily difficult and by this I mean something identify a new pathogen through wastewater for example and the case that I'll highlight why I think it's so going to be so challenging is the development of a fecal indicator based on the bacteria phase crash phase so those that are familiar with gut microbiome type work may have seen this paper come out about 10 years ago where these authors used a new bioinformatics method to reexamine some gut microbiome data and discovered a very abundant bacteria phase that they named crash phase and they saw that it was very abundant and nearly ubiquitous and so we saw that when did our own metagenomic screening and saw that it was way more abundant than pretty much any other human fecal marker that we had seen yet and developed a new assay and saw that it was indeed so the two assays are there on the left and this comparing was sort of some of the best performing fecal indicator assays in wastewater at the time we saw it was more abundant than those so we literally it was sitting under a nose it was been there the whole time and we had no idea that it was there and so if something that is that abundant is not going to be seen without prior knowledge of its existence I think it's going to be very difficult to use wastewater based surveillance for discovery applications of totally novel targets not variants or things where we have something that we can use to identify it but truly novel targets I think is going to be a really tall ask so with that I don't know if I take questions now but I'd be glad to answer any questions that we could move on all right thanks Kyle that's that's great really appreciate that and I promised Raul he would get a question in there oh I'd take back my hand I'll save it for later I'm still simmering now not after Kyle's talk you can't take your hand back it doesn't work all right any other questions at the moment we can get back to kind of that broader question of detecting things that are either really rare unknown or as you know completely unknown so given that could I just ask a quick question sorry to jump in yeah thank you Kyle that was really interesting I'm wondering sort of the last statements that you are making I wonder if we could distinguish between variants total unknown unknowns and like near neighbors of known pathogens so pathogens that are of concern that we're monitoring but new sort of I guess it would be novel viruses that are related to those but that they might have some sort of conserved consensus that could affect their their genomes that could could facilitate that detection yeah you're definitely right there's more of a gradient than I painted it as there you know I guess my first thought if you're talking about sort of like a near neighbor type virus is wastewater is so diverse and has so many inputs and sources that it would be really difficult to definitively declare that a human pathogen as opposed to say virus or something like that without, again, prior knowledge. It might depend on how closely related it is to something we know or it could maybe inform us going and looking elsewhere for that. But to me, if we were doing some assemblies and, yeah, that looks really close to the polymerase from whatever, you know, it would be hard to take that leap, I think, without clinical data, that's my opinion. Don't see any other questions at the moment, but if you could all the speakers so far, if your schedule allows, you could stay with us. We're going to have time at the end for a little broad-ranging, more broad-ranging discussion. So with that, we're going to have two presentations here on analysis of novel organisms. So, Kyle, you kind of, you put that up quite nicely in implications for wastewater surveillance. And Mark, are you going to go first? Sure. Okay, Mark, and then Kimberly Dodd after Mark. Go ahead, Mark. Okay. Oops, there's my data from this week. I meant this to be more of a discussion than a really formal presentation, so feel free to jump in. Just raise your hand and I'll stop talking. There was two points I was going to make. I want to make one third one just because of the discussion earlier. PMMOV is great, especially because it can be, you can measure it with the same assay as what you're already doing, which is sequencing. But it drove us nuts, particularly with the prisons, because it would just be through the roof one week and not the other. I think it was taco night or something. We've done a lot of work at it and we find the caffeine and caffeine metabolites, if you have the means, are incredibly reliable measurements of human contribution. But what I want to talk about is, all right, so there's different populations that you want to study. There's the patient and then there's the city and there's all different gradations in between. And not every application is the same for each. In the same way as when you're looking for new viruses, are you just looking at which SARS-CoV-2 specifically, or you just want to know everything what is present? And different techniques we use for different approaches. For the intermediate stuff, just like what we're doing mostly with wastewater of looking at new variants, I want to make the point that linkage is mostly overlooked, meaning what mutations go together. There's been some that looked at, but most of the programs just look at polymorphisms, what mutations are present. I just want to point out, if you just have 105 nucleotides of SARS-CoV-2 and you can maintain linkage, these are the different things that are polymorphic just within the omicron lineages. And with about 95% accuracy, there's a few places where two lineages have an identical sequences here. But for the most part, you can almost exactly nail every virus just with those 105 nucleotides if you maintain linkage. So the strategy we use in Missouri is not that small. We use 500 base paired chunks where we do PCR and then we multiplex them and deep sequence them. And that's how we get our output. And we don't ask the health department what they want to know about. We just look at what's there. And we say, what's this? Oh, it's EG5. What's this? What's this? We don't know. But we just report it to them and then they decide which ones they want to report. But there's none of this like trying to figure out what we want to look for and then looking. We just ask what SARS-CoV-2 is present and then we tell them. It's been very effective. We got it up and running just before the first alpha wave. Advantage is it's very fast and cost effective. The cost of reagents and sequencing is less than $50 a sample. It does not need to be concentrated. And the results are never ambiguous. It's just this is the sequence. Occasionally, two lineages will have an identical sequence and then we can just pull out another part of the genome and figure it out that way. And the important part is we don't miss the unknowns. You can decide for yourself whether you think this is important or not. But there are all these places where we started to find lineages that just did not match. And for the longest time we thought they were coming from an animal reservoir. We call these cryptic lineages. We found 50 of these to date. Sometimes we find them ourselves. Sometimes I dig through Mariana's data. Thank you, Biobot, for putting everything on SRA. We found a new one yesterday from Washington, D.C. It looks like it's about a year old. So we thought for the longest time they were coming for animals. You've probably heard this in Wisconsin. There was a really wild one of these where we actually tracked it down all the way down to a single set of toilets. So it was obviously coming from a person. At that point we had the whole genome. It was a virus that was probably in that person for at least two and a half years. It was from a late 2020 lineage is where it was derived from. It's about as divergent as Omicron. But the important thing is we also found out that the amount of virus this person was shedding was exponentially higher than anything we'd ever seen. It's so much that one person you can detect in a sewer shed with hundreds of thousands or in the Washington, D.C. case that we found yesterday, we could detect it in a sewer shed with 1.6 million. It's pretty incredible. So this idea of linkage, you don't have to, it's not just with SARS-CoV-2. It's also we also have done with the introviruses. There are some very highly conserved domains where you can design primers, you can amplify and just deep sequence it, and you can get a rundown of every introviruses presence. The coxaxes, the D68s, the A71, and polio if it's there. We accidentally found a polio outbreak in our work with this. Fortunately it appeared to be in a very contained community, and it fizzled pretty quickly. But this was unrelated to New York, but it was in the U.S. It wasn't really exactly what we were looking for, but it was the kind of thing that will show up with this, and you can detect it. Again, just focusing on linkage between everything. And then the other point I wanted to make is that wastewater is not always the answer, particularly when you start getting into the metagenomic sequencing. When you're looking at, particularly when you're looking at people like places like airport schools or airport or airlines, you're really cutting yourself off from most of the data by looking at wastewater. Because this is not places where everyone uses the bathroom. It's a sampling, but it's sort of way less than what you could be getting, which is where I'll put my plug in here that when we've done work, we've done some work with David O'Connor at University of Wisconsin looking at air. It all started off because they were doing a continuous air sampling from a bar, and they wanted to know whether they could use my technique to figure out what lineages of SARS-CoV-2 were present, and so they sent me the RNA from these air samples. And yeah, we could very clearly tell from the movement, the transition from Delta to Omicron, just from these bar samples. They've also had a case where he had a small household that had, they all came down with COVID, and so they did simultaneous rat and air testing every day, and it worked really well. But one of the nice things about when you're starting with air is that we took the same samples from that bar, and we just did some shotgun sequencing, and there there's much greater enrichment than what you find in wastewater. So we've done shotgun sequencing in wastewater as well, and you'll find a lot of things, but it's a lot harder to find some of the things on this list in wastewater than it was from air samples, particularly like the skin viruses and the respiratory viruses. You didn't have to sort through nearly as much, and honestly, I think if you could do air sampling efficiently on a whole city, I would go that way. But the beautiful thing about wastewater is you get everyone incorporated. It's all mixing and you can bring out your sequencing heavy guns and really look at what's there. So I left that mostly for discussion. Am I actually under time? Yeah, so if happy to discuss any of this if anyone's interested. Thank you, Mark. Scott's got his hand up. Hey, Mark, nice to see you face to face for change. So question for you. The idea of identifying an unknown unknown, an agnostic sample with this approach, where do you stand on that issue? We can't do it now because most of what we find, we don't know what it is. It really offends my sensibility the first time I did that to realize how much is in us and we don't even know what it is. But I think it's not as that far away. I think once we know what we're looking, what's there, we will be able to know when something new is moved in. And so are there other challenges besides just not knowing what else, knowing everything that's normally there? Are there other challenges? Well, the challenge is that most of 99.9% of what you get from wastewater is bacteriophages. So there's a very large denominator that that's the challenge. I think it is, I don't think it's insurmountable though. You're probably not going to get a, I don't think you're going to find the whole genome of something new very easily by just plain shotgun metagenomics. But you might be, you'll get enough pieces like, well, what the hell is that? I think it's possible. I think we're close. Thank you. I think someone else had their hand up. Maybe not. Hey, Mark, good to see you. I always enjoy your talks and almost a similar question is, Joan asked already. So I was thinking to ask another question, maybe it's a little bit more, you know, about ethical consideration and privacy. So you just mentioned you reached out to a single bathroom. So how did you got, you know, like permissions to that level? I know like you must have done some sub sewer shed sampling and stuff. So can you just share some details about that? And how you reported that data? Well, let me first say that we thought we were looking for an animal. We didn't know we were following a person. And I have to say that I didn't make any of these decisions. This was the Wisconsin Department of Health and the CDC, which are on board communicating with local leaders and the company at every step along the way. And once we got to, once we realized it was coming from a person, sort of everything changed years. And it took several months while they worked out the messaging, they went to the business, and they wanted to communicate with the workers. And that's actually why we know it was coming from a single set of bathrooms is because the company came back and said, we don't want you to talk to our company unless you narrow out and rule out that it might have been coming from someone from the public. So we got to figure out which bathroom it's coming from. And so it was with the company that we sort of narrowed it down to this set of bathrooms. They then communicated it with the group. They were all offered free testing. They were all negative by nasal. And then everything stopped for five months where we went out and got an IRB that we could find a way to get informed consent to actually take stool samples. This is an issue that we all need to work with as a group. There are considerations. I don't think it's really been decided where privacy, where's the line when you're violating someone's privacy. I don't have the answer on my own, but that's what happened in this case. Thank you. That helps. Thank you. And let's move on to Kimberly Dodd. There she is. Hi, Kim. Hello. Make sure. All right. You should be able to see my screen. That looks good. Hi, everyone. Thanks so much for the invitation to be with you today. I'm a bit of an outlier in this group. I'm director of the Michigan State University Veterinary Diagnostic Laboratory. And so I am not an expert in wastewater surveillance. I'm not even a non expert. What I am is a veterinarian and a virologist who spends a lot of time thinking about ways to maximize detection of high consequence animal and zoonotic diseases in order to protect both animal and public health. Prior to starting at MSU, I served as director of USDA's Foreign Animal Disease Diagnostic Laboratory, or FADL, which is the National and International Reference Laboratory for High Consequence Animal Diseases. And so the impacts of a potential incursion of an exotic high consequence disease into the US is certainly not lost on me. It would have devastating impacts, not just on human and animal health, but also on the economy. And so it's really a priority for us to better understand how we can expand our surveillance or response activities for new pathogens. The core of veterinary diagnostic labs, or VDLs globally, is really to ensure rapid detection and effective response to an outbreak situation. And here, when I'm talking about VDLs or veterinary diagnostic labs, I'm generally referring to members of the National Animal Health Laboratory Network, or the GNOME, which is a network of 60 academic, state, and federal diagnostic laboratories that serve as the first line of defense in the case of an incursion of a high consequence animal or zoonotic disease. And so what am I doing here? Here, I want to share a little bit about the capabilities of veterinary diagnostic labs, how I think our work can help inform wastewater surveillance efforts and end with a vision for the future of how we can help support national priorities related to disease surveillance and monitoring. You know, veterinary diagnostic laboratories kind of had an opportunity to shine during the COVID pandemic. Like public health laboratories, you know, we watched the number of cases rise and saw public health labs really struggle with meeting that huge amount of testing that was required. For veterinary diagnostic labs, we're all about herd health. And so for us, this is something that we're constantly training for. And so we're always preparing to be able to rapidly scale up testing in a large-scale outbreak scenario. And while historically, our mission has been to provide surveillance for incursion of foreign animal diseases or exotic animal diseases like African swine fever, foot and mouth disease virus, or zoonotic diseases like high-path AI and swine influenza, the reality is, is coming out of this outbreak, we recognize that we all need to be better prepared to detect emerging infectious diseases before they have a chance to truly emerge. So what we do is veterinary diagnostic laboratories, while we don't use wastewater as a sample type, we do leverage really creative population-level sample types for which we face exactly the same challenges that you all have been talking about today, ensuring that they're sensitive enough that you can repeat them and that you really have confidence in the results. And this is particularly important in animal health, particularly when we're talking about high-consequence diseases where a false negative could result in movement of animals across the country or even out of our country to somewhere else. Some of the sample types that we think about using at times are bulk tank milk samples. So milk samples collected from hundreds of cattle, which are then utilized for testing for infectious diseases or antibodies. We also do a lot of environmental testing and this is particularly relevant at the end of a large outbreak of disease where we want to ensure that after the premises has been depopulated, that it's clean before we bring new animals in as well. And finally, when dealing with live animals, particularly for example in swine populations, we do things like hang ropes in the pen, allowing up to hundreds of pigs to chew on the rope and it's about as gnarly as it sounds where you then twist that rope to extract the saliva and frankly it's a great population-level sample for potential infectious diseases. But again with all of these, we face the same challenges you guys are talking about today. The ability to get really cast a super wide net to get population level testing while maintaining that sensitivity and repeatability and really be able to have confidence in those negative results. And so that's something we continue to work on. And again, as I've mentioned to date, historically we've really relied heavily on pathogen-specific tests to be able to detect individual agents. But again, as we look to the future, we want to leverage new technologies and to really be able to try to detect emerging diseases a bit sooner. And so our lab, like laboratories across the country, are working to leverage new laboratory capabilities with a particular emphasis on NGS and harnessing the power of AI machine learning to support real-time disease agnostic testing to ensure we're not only detecting known agents but maximizing our ability to detect novel agents or new pathogens as early as possible. So why is this the job of veterinary diagnostic laboratories? Well, you guys all know the numbers, right? 75% of new emerging diseases come from an animal reservoir. What better place to find those than an veterinary diagnostic laboratory where we have tons of samples coming in from sick animals? In our laboratory alone, we run a million tests a year on samples from hundreds of thousands of animals from every state in the country and around the world. There's huge potential here. So, you know, while we know that zoonotic diseases are caused by a huge range of agents, viruses, bacteria, prions, parasites, etc., from a huge range of potential animal hosts and could be transmitted in a number of different ways, whether it's airborne or an arbovirus, the one thing remains the same, that if our animals are a reservoir, then again, what better place to look than in the animal samples themselves? And so, this all sounds great, but the reality is, is that today our capabilities are still kind of limited. We run, we know that our routine diagnostics don't always find an answer. Our disease-specific tests are highly sensitive and specific and validated. We're really confident in those results, but they're not always all encompassing. So, again, it's not uncommon for us to receive a sample from a sick animal where there's a suspicion of an infectious disease. We'll run several tests to rule in or rule out potential causes, and the end often come out negative. And that doesn't feel great for anybody. And so, we're excited to recently partner with USDA on a new program to be able to chase down diagnosis in animals that are clearly ill to help identify a cause of disease. This is called the Unusual Morbidity or Mortality Event Program, or UME. The goal of this, again, is to be able to ensure that we're finding answers when we have an unusual example of disease in an animal. Any species is eligible. Now, the reason we see these negative test results are usually for one of two reasons. Either the veterinarian, the farmer, the producer, whomever submits the samples has limited funding to put towards testing. Or, and perhaps sometimes more frequently, the laboratory runs out of potential agents to look for. And so, this program will do two things. One is that it'll provide funding for additional testing, either at the current laboratory or it can be sent to any of the other laboratories in the network, and to really ensure that we're getting, we're able to take as comprehensive as a look as possible in being able to rule in or rule out potential causes of disease. Now, that's really just the first half of it. Because there, we're still relying on looking on disease-specific tests. Again, we receive a sample. We run multiple of our highly validated tests to rule in or rule out potential causes of disease. The challenge here is the same you guys have mentioned earlier, is that we need to know exactly what we're looking for. And sometimes, we just don't know what that is. Through COVID, which I think really kicked off was already going to be a global evolution in diagnostics, we're seeing rapid changes in the way we approach diagnostic testing, particularly for infectious diseases. There's more and more of an emphasis on multiplex or panel-based testing to be able to run a single test on an animal sample, but then to be able to analyze it from multiple different pathogens. This is great. It's more efficient. It's more cost-effective, which can be really important in the animal health sector. But on the other hand, we still need to know exactly what we're looking for. So it's an exciting time that we're all talking about NGS as such a routine way of performing diagnostics and being able to harness it more directly for diagnostic purposes. We all know that just like a wastewater sample, any sample from a cow, we can completely sequence very quickly and relatively cost-effectively. The challenge then is always going to be in the analytics. And that's multifactorial. Being able to go through and identify any potential pathogens in there. And the second critical part of that that a few of you have mentioned is this idea of how do we interpret that data? How do we know what matters? But again, coupled with our new laboratory capabilities are also new analytical capabilities. And this is something that I think a lot of us are starting to lean more and more on is leveraging the opportunities through AI and machine learning to be able to train systems, to be able to detect any potential pathogen in the sample, as opposed to having to go through it sequenced strand by strand to identify known agents. So, you know, what if we were able to use the same technologies that our cell phone has that's able to utilize spatial recognition to open up our phones? It doesn't matter if I have my glasses on or off, my hair up or down, I'm smiling or frowning. Because of this analytics, it's able to identify who I am. So the same thing we can apply here to sequence-based data as well. How do we leverage these machine learning systems to detect not only the coronaviruses we know today, but the distantly related ones that could potentially emerge? And this, again, as the veterinarian or the veterinary diagnostic lab here, I really want to stress the breadth of utility of an analytics platform like this. We can utilize it to detect known pathogens to help identify the next disease aspects. We can also leverage it for genetic disease testing for cancer diagnostics and prognostics, and really be able to enhance our capabilities to understand the health status of individual animals or animals at a population level. And finally, sort of my last play to bring it back to where we're talking about today about public health is that these sorts of platforms are really one health. Because just like the output is disease agnostic, so too can be the input. And so how we can leverage these capabilities and work together to really build these unique sort of pie in the sky platforms that we can utilize, again, whether it's a sick cow or whether it's for wastewater surveillance, to really be able to enhance our abilities to understand disease prevalence and to be able to monitor it over time. So with that, I will stop sharing my screen and see if there are any questions. Thank you very much, Kim. I do have a quick question. So if you take, as an example, your unusual morbidity mortality events and you detect something either just not typical, a variant or some degree of novelty. How does that information then and at what time scale get communicated to, for example, the CDC so that then that would raise to their level of this is a potential target for wastewater surveillance, even though there may be no known human cases at that point. How does that information, can you can you kind of just outline how that information flow either works now or how it should work? Yeah, so it's a really great question. And I think, honestly, it's something I didn't appreciate till I came to this role because we really serve as a state lab for animal health. And we are in constant communication with our state health officials, both in the public health sector, as well as the animal health side as well. And so if we were to detect something unusual in one of these samples, we would immediately work with our state veterinarian, our state public health veterinarian, and the other public health officials to have this conversation and bring in CDC quickly. So to answer your question, the initial conversation would be within hours effectively. Again, it's a really tightly knit group, we're a partner in trying to understand what's happening on the infectious disease landscape here in Michigan. And then again, to be able to allow our public health counterparts to take that up to CDC as needed. And I think, you know, an important part of this, too, is really being able to sort of utilize our animals and sentinels for potential agents, and then be able to build those built the specific disease specific tests were needed to be able to allow for more efficient testing. And then to determine whether or not it's something, you know, that that's a big public health concern as well. And whether or not there needs to be national programming related to it. Yeah, yeah, thanks, Kim. I think it is kind of an intermediate step between the unknown unknown. And you now have a known but at the animal level with an unknown impact at the human level. But but that unknown unknown is better than an unknown unknown. So that's absolutely true. But I think just the other thing too is that as it's become easier and easier to be able to run some of these diagnostic tests is leveraging public health. I mean, we have seen during COVID we were sampling animals who had owners who are positive. So being able to leverage those relationships and look directly to animals or humans directly in contact with the other. Super. Chris. Kim, thanks so much for coming out to to share your ideas really around it. I guess a dream system for what the future should look like. My question has more to do with maybe your role previously with USDA, but are you aware of any sort of official systems or or even pilot systems to look at wastewater effluent from intensive livestock production facilities? Is anyone doing that? And is if not, do you see any potential role for something like that for like foreign animal disease detection or anything on the animal side? Yeah, it's a really good question. And to the first part of that, no, I'm not aware of any nationwide program efforts to look at wastewater surveillance. But I think it's a very good question as it relates to some of the large scale production facilities. The second part of your question is was the value. And I think one of the biggest challenges we face when it relates, especially as it relates to high consequence or foreign animal diseases is that the second that we have a positive trade comes to a halt. And there's no sort of backstepping away from that economic impact. That's why testing for these diseases is very heavily regulated by USDA. And I can tell you in my previous role at Faddle, I lived in fear that someone was going to have a min-ion in their basement and detect ASF before we knew about it, because the impact that would have if it made it to social media. So I say that as a reference point, which is that as we look at looking at some of these really high consequence diseases from a national or from a regulatory perspective, there's a real need to be to very closely define the circumstances under which testing can be performed and how it's performed, the sample types that are used. And that's not so much to be able to minimize our opportunity for detection, because certainly the more we test, the more likely we are to find something, but rather to ensure that we have confidence when we have a negative test result. And so I think that's where the problem lies with wastewater surveillance. It's why we don't use bulk milk tank for foot and mouth disease yet is because there's that concern of until we have greater confidence in that limit of detection, it's not something that is going to be easy to embrace from a regulatory perspective. Yeah, super helpful. I think that's for us to learn on that perspective. So because if you have a test, you have to also immediately concurrently have a plan on what you're going to do with the test result. That's something that I think we're navigating into space here that's really helpful to have your ideas on that. Thank you. You bet. All right, thank you. And we're going to move on to Peter Dorstein from UCSD, talking about alternative detection methods with a focus on metabolomics, which is one of the questions that really the committee has looked at and needs to address as we move beyond kind of the focus that we've had with COVID and viruses into looking at different approaches. Peter, it's to you. All right, perfect. Just like Dr. Dodds, I don't do wastewater monitoring, but what I do do is I try to understand microbial communities and the chemistry associated with microbial communities. And so that's sort of the foundation of where I come from. But before I go on, I do want to disclose I'm a scientific co-founder and advisor for these companies. So because most of you are not familiar with untarget metabolomics, which will be the topic of my talk, I want to carry you a little bit through what the data actually looks like. So first, when you have molecules, you need to ionize them. And then we can detect them, can detect those ionized molecules as MS1 signatures. We can then isolate those MS1 signatures and hit them over the head with a sledge hammer. And now a sledge hammer is helium gas. But those molecules fall apart and they create a signature. And those signature represent the molecules that are present in the sample. And this is actually what it looks like. This is a relatively low scanning one, but you can see how many scans per second we're doing. And then this needs to be translated, much like metagenomic or metatranscutomic analysis. There's really a thousand and one ways to create data tables from this that you can then use to do further analysis. And so if we, the promise of untarget metabolomics is that it will be really easy to do medication readout and their metabolism, microbiome derived molecules and exposure type molecules, as well as post derived molecules. And so what that means is that whatever goes into a person you should be able to detect in the wastewater stream as well as the microbial processing associated with those different molecules that are present in wastewater. But let's talk a little bit about reality of untarget metabolomics. So let's say you collected all this data and that's represented by that dark circle that I just showed you. In 2016 we could on average annotate 1.8% of the data and that has grown to 14% of the data could be annotated nowadays. But the type of annotations that you might get look like this sometimes. This is the IUPAC name for zithromycin. But if you did not know that it would be really hard to interpret that that this is zithromycin. And so how do we get insight into the dark matter? And so there's different types of algorithms that can be developed. One of those is molecular networking that's very similar to sequence alignment where in sequence alignment allows you to look for mutations and analogs of sequences. And so we built the same thing for spectra. So these these signatures that I just showed you. And then if you can match it against a reference library of known molecules you get some annotations of these particular signatures. But because you have these delta masses that you can begin to interpret and build relationships among the molecule and even discover new molecules. Sometimes it's associated with metabolism. Sometimes it's microbial processing that different microbes process these molecules in different ways. Those are the kind of things that you can capture with this. So let me show you briefly. No, this is not a fecal or this is not a wastewater sample, but it's it's maybe most closely associated with a wastewater sample. And these are ocean derived samples just before rain and just after rain. So this is 2017 and then 2018 three days later. And so when we look at this and you look at the overall metabolomic profile completely changes the chemical makeup of these different samples, even though they were collected in the very same location. And so when you subject this to molecular networking, we ended up getting 56,000 nodes. So these are 56,000 unique SMS signatures that represent the molecules that are present in this particular sample. And the way to think about this is really that this is a knowledge base of all the detector molecules that are present. And we're using a low efficiency instruments, modern instruments, we're probably you're probably going to be looking at at least 10 to 20 fold increased number there. But we're going to just focus on some of these nodes. So here we have a small cluster of related spectra that represent the spirit spirit spirit mole lights. Two of those are annotated based on reference library matches. And then two of them remain unannotated. But these are molecules that are used in the agrochemical industry. And indeed, before rain, we didn't see any of this in the ocean. But as soon as she's at the rains happened, you know, I'll start all of a sudden started to see by the San Diego River, as well as the the pentascretus river that has a runoff right here. And so so those are the type of things that you can begin to monitor. You can also see changes in molecules associated with organisms live in the ocean. So here we have an algae molecule, frugus antin. And that essentially is diluted out after the rains, whereas other molecules that are more human derived and particularly associated with beaches that are inhabited by people that before the range, you don't see that molecule in or not very large quantities in the oceans. And right after the rain, I think it's just runoff on the beaches, you start to see this particular molecule that's generally actually considered a marker for kidney disease and diabetes in humans. Molecular networking can capture metabolism. So here we're looking at an antifilic agents association called Irigar goal. And if we monitor the the molecule where this cyclopropyl group has been removed due to essentially microbial metabolism, what you will see is that this metabolized version sits in the harbor, which makes sense. It's an antifiling agent that's added to ships. And then it's diluted out and enters the ocean after the rains emerge. But again, molecular networking really nicely captures some of this metabolism. But we still have 86% that remains unannotated. And one way to begin to help this to annotate that unknown information is really connecting the world's community. And this is one of the things that we've started to do. There's a couple of other resources that are beginning to do this as well. And basically, it's where people can upload their data. You people then add their knowledge associated with it. And right now it's about 1.2 billion spectra that we've accumulated with the worldwide community. What we do behind the scenes, because mass spec has so many different data formats, we conferred it to a uniform text readable format. And basically what we do is continuous reanalysis of the data. One way we enhance the information is that we have started some of these community curation projects and build search engines so that we can interpret the unknown information in in some unique ways. So I'm going to show you an example of how we started the community curation in the microbiome space. Basically, what I did is I sent notes out on social media, Twitter and other places as well as some of our lab members and said, if you have LC-MS-MS data, so on target and metabolomics data of microbes grown in culture, reach out to us, we're building a tool so that we can make it searchable and informative for the community. So we had about 100 scientists worldwide that contributed to this and that resulted in about 60,000 LC-MS-MS data sets files curated associated with microbial monocultures. Pathogens unfortunately for this group are still under represented. So if there are people that have a lot of pathogens that have been cultured, we would like to add this, but essentially you can search an MS-MS spectra now, a single one or a batch of them and find what organisms are capable of producing this molecule or a close analog of it. And in our case, we're trying to see how, where these kind of molecules are then distributed throughout the body or what diseases they're associated with it and sample types that they're observed in. But you can easily envision, we can add wastewater here and do the same type of analysis or wastewater over time. And then the second example is the medication curation that we're doing. We're basically creating a drug library where we attach ontology to it. So the original drug source where it was isolated. So being non-indogenous or food or non-food. And then endogenous and food where these molecules were originally isolated from. So epinephrine, for example, is an endogenous molecule. Vitamin C is a food-derived molecule whereas penicillin is a non-endogenous or non-food derived. Then we have the indication attached mechanism of action, Alzheimer's drugs. And so if you're interested in monitoring whether people are just getting sick, you may be able to look at infectious disease molecules or treatments, drugs that are starting to appear in the wastewater. You will start to see that global population, well, depending whether it's a virus or other pathogen, you will be able to monitor this. And this is how it looks like for our brain extracts, but you can easily envision each wastewater sample could be on the x-axis as well. And really monitor the type of diseases that are happening in that community. We've done similar things for food where we reposition the food data, in our case, for a diet readout in people, but you can start to assess food patterns in relationship to maybe pathogen exposure in wastewater. Here's what that looks like if we map this onto molecular networks. We can quickly identify what are medications, what may be coming from microbes, even if the mass spec data doesn't have a structure associated with it. And that's really the strength. Plant foods, these are all the ones that we've curated so far, but I can envision that if you curate pathogen data, if you curate infection signatures, so maybe human data sets or animal data sets with infection signatures, you might be able to monitor this as well. But we're also trying to curate some of these structures. For this, we have the mask URL search engine. But basically, this allows us to look for patterns. We look for patterns such as ace of carnitine. If there are specific classes of molecules you're interested in that you want to look for, you can begin to build these kind of queries. And in this case, we found about 1,800 case of carnitine in the public domain. 800 of those have never been observed in biology or described in biology. If we do this with bile acids, we found 20,000 different bile acids. About 19,000 of these have never been reported in the literature. So there's a lot of room for improvement of what we're going to be able to annotate in the future. What to expect in sort of the next 10 years? I think in the next 10 years, we'll be able to annotate 50% of the data. My guess is, and I wouldn't have said that in 2016, but my guess is that we will be able to annotate 100,000 molecules or more in a single project. And that real-time monitoring or near real-time, so within a few minutes, could be done with this type of an approach and reporting based on illness, pathogens, or medication juice. How do we get there? I'm happy to discuss. And of course, I have a great team and I'm thankful for them. And I will take any questions you might have. Super. Thank you. If you stop sharing, then maybe. Yep. Okay. Super. All right. Questions from the committee. And they can be for Peter, or we can open it up for anyone, or we can open it up actually to the broader community as well. Ami. Thanks to all of the speakers from the second session for great talks. Peter, this question is for you. Keeping in mind Mark's comment earlier on about how mass spec measurements of caffeine are actually some of the most consistent for data adjustment or normalization. How straightforward do you think that would be to measure in mass for these types of normalization or data adjustment efforts? I mean, I imagine for wastewater caffeine would be somewhat variable. Maybe I'm wrong on that. But if that's truly a constant, because I don't know wastewater, because if I look at individual people there, it's variable. But maybe in wastewater, it's consistent. Then I think you could routinely compare it to caffeine. There may be lots of other molecules that we can identify that will be consistent that we can also. What did you call it? Not normalization, but data adjustment. I really love that term, so that you can look at their relative levels for sure. Yeah, and just to follow up on that, do you have any suggestions just off the top of your head of other types of molecules that might be used for that type of data adjustment that would be easy to measure and consistent with osteography? Yeah, so in urine, creatinine is commonly used. So that's a really standard way where people normalize in clinical studies once they do urine monitoring. So pretty much any physician when they mark, you know, by lots of panels and such, they will normalize it against creatinine. So that would be a good one to also include. Great, thank you. Any other questions? Let's see. Yeah, Mark. Question from the scan folks. After seeing all your work with solids, we tried it and it worked great. But the problem was half of our samples don't have enough solids to do the extraction. What are you doing? There's no solids. Are you, I guess, if there are no solids, like by definition, you can't work with solids. So I guess it depends on where you're getting your sample from. But we samples that have been pretty thin like are maybe from schools or buildings. And so we normally can still get solids from those samples. I don't know. I mean, I can talk to you offline about it if you want. I was just wondering if you had like a whether you have an adjustment because a lot of our sewer sheds are very thin, particularly ones that have a lot of industrial input. Okay, I didn't know if you had a standard workflow for it. If there was. We haven't encountered that problem. So we could talk offline if you'd like, you can reach out. The questions for any of the speakers being none gives me the opportunity to thank each of you. And I think there's some just as we did in our last session kind of following up, you may get some additional questions by email or we may set up an individual meeting with you as we work through our report to draw on your expertise. But it's certainly extremely valuable in kind of framing how we go forward. So I just appreciate all of you taking your time to help us out. Can I jump in? Because we do have a couple of minutes. I'm really curious, like there may be the potential for wastewater. But as Mark Johnson noted, like air may be better in some places. I just want to would love your input on where wastewater seems particularly valuable for these kind of novel organisms and where we may be trying too hard because there are other methods. And I think Kim was talking earlier about the value of just looking at diseased animal samples rather where you know there's a problem rather than looking in a mixed sample where you don't know there's a problem. But I wonder if you have broad input on that question about really the opportunities for wastewater analysis for novel organisms that we're not using and where we may be trying too hard. Yeah, sure. I could throw some initial thoughts out there because we have other projects in other areas of environmental surveillance. And I think unfortunately the answer is it depends. And part of that looks at like what we're looking at with different shedding and where it might be most appropriate to look at and with your goal of what you're trying to get out of it. I would never say wastewater should be a sole data source for disease monitoring or is one size fits all. So I think it has to be fit into the bigger clinical picture and that's going to vary for each pathogen. The consequence, the shedding, the prevalence, where it's most appropriate to look. The great thing about wastewater is it's the great equalizer. It's not a sampling. It's everyone. And there's just nothing else that can accomplish that. And where you start to stumble is when you start going down lower and lower to where you're not really sampling everyone. So in places where people aren't living, then you're no longer sampling everyone. You're only sampling the ones that are using the facilities. I definitely don't think that air or fomites is really going to be better than wastewater. There's just it's such a dilute or surface waters in most places unless they're directly receiving raw feces like feces, maybe in low and middle income countries. Because what like the waste stream is just very concentrated. It has at least in the United States like the highest concentration of pathogens. Once you get into air or on fomites, for example, it just becomes like more dilute. And we have done a systematic review of the literature of viruses on fomites, for example. And most of them are non-detect. There's very little data on human viruses on fomites. So I guess I respectfully disagree with Mark that air is going to be the way to go, but that's my opinion. One opinion as we think about wastewater and the complement to other types of surveillance, specifically clinical surveillance, is that for diseases where we have an asymptomatic shedding period or for diseases where there's asymptomatic cases, a fair fraction of asymptomatic cases, wastewater can really be very additive to what we do in the clinic. If you have a disease that, you know, where you show symptoms right away and where you really don't have an asymptomatic population, maybe it can be more difficult to think about what are you gaining from the wastewater. So just kind of another framework as to how to think about wastewater and clinical surveillance. I think for as for the air, probably just like the logistics and the operations of sampling air at scale has already been done before in the large program. I forget the name by DHS and it has not been very successful. So I guess we do have that that evidence, but maybe there are other implementation, more local implementations where I'm sure the data could be very helpful. I think for what could be really hard. Yeah, so it looks really well for us. Yeah, I think I think weight and wastewater is a nice equalizer, as Mark said, so that you can continuously monitor the population. I actually removed the slides from air monitoring that we had. What we found was that it's really dependent on the air direction. So of course, if you're doing indoor air, that's that that doesn't matter. But the kind of signatures that we see was really if you got inland versus offshore versus the southern direction, you get very different patterns of signatures that you're able to see depending on what is being utilized upstream on those winds. And so air will be much more variable. Now, having said that, there are some indication that that pathogen exposure may be sometimes wind dependent. So one of the observations that one of my clinical colleagues had was that whenever we get winds from the from the ocean near the southern border, there's a huge increase in Kawasaki disease. We don't understand that relationship yet, but it's certainly something to look into. And it may be something that people can think about in terms of monitoring. I just mentioned too that wastewater has turned out to have this amazing cut of all of these different types of diseases. It's very amazing how many different pathogens are shed in a way that makes its way down the drain. We hadn't really used these technologies much beyond GI viruses prior to the pandemic. And there are so many different ways. I mean, Mpox, for example, the shedding of Mpox into wastewater and being able to get a useful signal from wastewater on a virus like that I think is very powerful. And I think a challenge with air sampling is that you may not get sort of the same diversity of different diseases that can be effectively monitored through the air as you do through wastewater. But whether you do or not, you may have a different set of targets that are useful in air versus wastewater. And so that may be something to consider as well. But regardless, I think that looking to clinical and public health surveillance data that shows us that something is not just present but is causing disease in a way that is of public health significance is critical for choosing these targets. And when we think about discovery of new things, because that piece of knowing what impact is this having in a community and what's the public health significance of that pathogen spreading, that's a piece of information that right now I think is difficult for us to assess from these environmental samples. That's true. I didn't see any introviruses in our air samples, thankfully. I mean, they aerosols happen. I was surprised that the viruses in most of our air were skin viruses. Like, what's the pox virus that causes athlete's foot? We saw a lot of that. But nothing, yeah, it was mostly respiratory and skin. It's true. It doesn't capture everything. Yeah, nothing captures everything, right? But interesting. Well, that's going to contain Joseph. That's it. All right. I saw Marlene. I'm going to give her the final comment there. She just wrote in the chat gross. So, given that, again, thanks for everybody. I really appreciate it. We'll be reaching out, I'm sure, to some of you to tap into your expertise.