Applied Genomics: 12 years of marine eDNA secrets

Hello, everyone. Hi, Jonathan.

Morning.

How you doing?

Good, good, good. Good to see you. I think you're going to enjoy this NTC now.

Yes, looking forward to it.

Yeah, I think I saw Gunther coming in as well. Oh, he's okay. No, there we go. I gotta admit them. There we go. Hello, Christian. Nice to meet you. Hi, Gunther. Hi, Steve. Two veterans of the NTC now that I love to see every week. Yeah, good to see everyone. Well, you know me, I'm going to be super efficient and just kind of launch right into the general NTC intro. So, Sebastian, now's the moment that I'm kicking you off your screen share going to mine. Can you all see my mural screen? All right, fantastic. Well, welcome everyone. Welcome to NTC now our weekly community meeting of the Nature Tech Collective. It's an event that's open to our members, but also the larger public. Actually, I'm going to start reminding everyone to kind of like name yourself first name, last name and organization if you can, because actually that's best practice for what this is a networking event at.

The end of the day.

And we are the Nature Tech Collective, a collective of nature tech organizations dedicated to advancing a more effective coexistence between human and more than human, closing the nature data gap to eventually close the nature finance gap. And we do a lot of things weekly community events, monthly educational webinars, in person conferences, gatherings whenever we can. But also we release sector intelligence reports, lots of stuff. And this is one of the community events that we have. So anyway, at your fingertips, members, you have the slack to our member forum where you can connect with each other. You have our luma or calendar of events, this is one of them. And you have our LinkedIn follow us and tag us when relevant so we can amplify you in terms of updates to the members. This community report, impact report is being written as we speak. And so I'm gathering kind of like your feedback and testimonials so that I can sure make sure to put it in there and create a plan for 2025 that makes sense for all of us. So if you can fill the report. That's the moral of that slide. In terms of events coming up, we have a few more NTC nows, but actually I've already created the one for the December 4, which is all about the mapping that we've created around cattle tech. So that's on the horizon in terms of opportunities. Two things that have popped up on the slack today. One's about the sustainable beef industry. We're creating this Map, and that's why there's the event. So if you feel like you're in this industry and you should be on the map, let me know on Slack or by email so that we can make sure to map you. And then someone here is creating a challenge with their students for nature, for Fintech in general. Nature fintech. I'm going to be linking this mural so that you can kind of click and read more precisely, because I won't go into too much detail about that. In terms of resources at your fingertips, you have two things that came out very recently as well, the Nature Tech Guide for businesses and the Nature Tech Taxonomy Framework, on which we would actually love your feedback. If you haven't given it and you'd like to, we're very keen to know what you think of what we've built with your help. A lot of you were there at some of the meetings that we had together and this is kind of the result of that. So there we go. And now I'm going to pass it on to Sebastian for. I'll let you explain yourself.

Thanks, Paige. Thank you, everyone. Thank you for turning up this evening or this evening for me. I'm not sure. It might be good morning for some of you. Can everyone see my presentation screen?

No.

No. How about I hit the share button?

There we go.

There we go. Nice. Yeah, thanks. I'm the founder and general manager of Applied Genomics. Applied Genomics has been doing environmental DNA analysis and developing new techniques for EDNA biodiversity monitoring since 2014. And in this presentation I'm going to be telling you about a collaboration that we had, recent collaboration we had with Plymouth Marine Labs in which we had an opportunity to analyze a number of samples which they'd been collecting over a 12 year time period. And the objectives of the project were to compare EDNA methods to conventional survey methods. We wanted to know about EDNA as suitable for fish phenology and diversity and how well it did for analyzing trends in communities and population genetics. We had a highly successful project, I would say, in which we were able to demonstrate that EBNA reliably monitors marine fish over time. It's highly accurate for biodiversity monitoring on site and it was found to outperform conventional methods in both quality and speed. And we were able to demonstrate that it had a significant return on investment. Now, the study site, in case you're not familiar with it, was the Western Channel Observatory, which is off the. Off Plymouth on the south coast of England at their L4 buoy. And if you're not familiar with it, this is one of the parts of the ocean which is probably one of the most well studied spots in the ocean. It's been, it's the L4 Western Channel Observatory has been going for decades and decades. And so the amount of data that's been collected here is really quite staggering. Plymouth Marine Laboratory has been working at that area for quite a very long time. And about 12 years ago they had the insight to or the foresight to start collecting samples, filtered water samples, purifying those, the EDNA found in those water samples and essentially storing them, not necessarily knowing what it was they were going to do with them. And so they approached us for analyzing these specifically looking at fish. And we thought that was a fantastic idea because they had about 12 years worth of samples collected on a pretty much weekly basis. And we had the opportunity to also compare their sampling method with our sampling method. We've developed a, a large volume automated sampler. It's called the in depth EDNA sampler. You can see it there on the left. And that's a device where it can be tossed in the ocean and it automatically filters over a long period of time. So just to put that into perspective, most DNA samples that are conducted these days are collected as point samples. So essentially somebody throws a bucket in the water or a niskin bottle collects a few liters of water, filters that. And what we found is the signal that is received by that EDNA sample is entirely dependent on the direction that the tide was traveling at the time the sample was collected. I like to explain it as in if you were to go into your backyard and you were to smell barbecue, it doesn't necessarily mean you're having a barbecue. What it means is you have to understand what direction the wind was going in order to be able to know which neighbor to whose door to knock on. If you want to invite yourself over for barbecue. In order to be able to overcome the natural variation that's in the marine environment, you need a sampler that's going to sit there and sample over a long period of time, ideally one to two tidal cycles, to allow that water to flow back and forth over the sampler and thereby get a very large area that becomes more representative of that geographical spot in on the ocean. And while that might seem a very large area to say, well this is what's happened here, zoom out, it's the ocean. It's really, really big. It's actually not, not a lot of space. In the end, this is what we got out of the DNA. As far as DNA recovery, there were 460 samples that were analyzed. In total, we analyzed three DNA markers targeting marine vertebrates, fish and elasmobranchs. And on average, we had about 20 times more Edna recovered from the large volume samples as we did from the small volume samples. And this is it. This is the. Probably the world's longest fish EDNA biodiversity time series, as far as I know. It had an awful lot of assays that went into it. And I'm very proud to say that we had a 100% sample success rate, as in we got data out of every single sample, which is amazing. Amazing testament to the guys in the lab. And from those samples, we were able to characterize 213 taxa to the rank of species, which is very cool. Now, I'd like to think that Jackson Pollock would be really proud of this plot. And like all great works of art, if you sit with it for long enough, there are some patterns that actually start to come out at you. And the patterns here that we're actually having a look at are seasonal patterns where you can see slight differences between winter and summer, autumn and spring. Drilling that down, drilling down into that a little bit, you can see which are the more. Which are the species which are kind of driving those underlying patterns. And drilling down even further, you can see the really trendy species that are in there which are really showing a strong preference for certain, for seasonality throughout the year. Back to Jackson Pollock. And probably the more interesting stuff is when we split this off into years. And we can see that there's a clear difference between the early years and the later years in the study. If we consider the blue lines on that plot to be richness. Higher richness on the left, lower richness on the right. Drilling down into that, let's just think about the data in terms of the number of species that were recovered in each sampling year. You can see that in the orange is the edna, in blue is the catch and count, conventional scientific trawl data. Unfortunately, the scientific trolling ran out of funding in 2018, so it's somewhat incomplete. But the EDNA shows a similar trend overall to the catch and count data. And you can see how both data sets start out in the high 60s, even high 70s for Edna and trail down over time to become worryingly low towards the end of the data. If we look at the underlying trend, though, in the EDNA data, we can see that the EDNA sort of predicts the decline perhaps one to two years ahead of the catch and count data, which I think was pretty cool. And for posterity, we had to look at the landings data, which is, which is the fish species declared landed in each year by small boat fisheries, local small boat fisheries, and this is a flat line obviously, unsurprisingly because local fishermen are incentivized to maximize their, their, their catch quotas. What we don't see in the landings data and the landings data are really the data in on which a lot of fisheries policy decisions are made. What we don't see is the additional effort that the fishers need to go to in order to be able to meet their quotas. So then to dive into the data a little bit further and look at maybe some functional diversity, we've got, we've got the trophic levels here looking at planktivorous fish. In this case we don't really see a trend in planktivorous fish over the time series. But if we look at the generalist fish we do see a statistically significant decline in generalist feeders over the time period. And that's also the same of the benthivorous feeders and also true for the piscivorous feeders. Overall the winners and losers, 25% of the data of the species that we identified showed a positive Trend. But only one and a half percent of those data were statistically significant. And 68% of the fish that we identified were found to be in decline. 25% of those were found to be statistically significant. So pretty, pretty disappointing result for the, for the fish biodiversity overall. However, we're not here to talk about fish policy. We're here to talk about EDNA methods and all the rest of that to be able to uncover these methods. So if we look at the comparison between EDNA and the scientific trolling, EDNA overall recovered more species than the catch and count scientific trawling. And remembering that these are the small volume samples that were collected by PML over that 12 year period. If you scale that down to the samples that we were able to compare directly between the PMLEDNA samples in the tan and the indepth large volume samples. Obviously the in depth significantly collects more data and that's simply a reflection on the differences that you get for being able to collect over the full tidal cycle. And we've seen this repeatedly over multiple validation studies. High volume samples do a much better job at detecting more species, more rare species as you might expect really. I mean it should be Statistics 101. You exert more effort and you will find more things. Anybody who is a parent of teenagers will know that return on investment wise we see that large volume samples have a 7 to 1 return on investment based on the amount of data recovered compared to cost. And Edna methods had a 5 to 1 return on investment compared to the conventional methods of scientific trolling, with the additional benefit that EDNA methods are non invasive and so no fish are harmed in the collection of the EDNA data. All of this leads to a. What are we going to do with it with this information? Well, what we're trying to do now is set up a system. The problem that we have here in the uk, and I suspect it might be a similar problem worldwide, is that we don't have a systematic method for monitoring changes in bio, in marine biodiversity over time. There are a lot of projects that are very siloed, they're all very, very much looking for very specific things. And so there's. It's very difficult then to compare these data sets and get any meaningful data out of them. So what we're proposing is to have a network of monitoring stations around the country in collaboration with marine and maritime industry stakeholders, fisheries, fisheries police, natural capital regulators, academic institutes, all the rest of that, who regularly go out to their local area and are able to just basically pop a sampler in the water, collect the sample once a month and then have it analyzed. And what we would have then are a network of nodes that are regularly monitored all around the country and then we would be able to see what are the trends that we, that we're seeing both in a phenology sense as in year on year or you know, what are the larger trends perhaps due to climate change or other sorts of perturbations that are coming on stream. So audience participation time. If you think it's a really good idea, I'd ask you to please register your interest in this so that what we're doing is we're canvassing people who are interested so that we can build up a network of evidence that this A should be funded for here in the uk, but also B, perhaps you are interested in setting this up in your local area. And if that's the case, then let's see if we can get some collaborators together, get some funding together and set it up for local area monitoring on a large scale at a systematic level so then we can actually start uncovering the real trends that are going on in the marine environment. Now, hopefully that wasn't too quick and hopefully it all made sense. But just in case it didn't, I'm happy to answer whatever questions you might have.

Thanks so much, Sebastian. Will you also pass me. I don't know that QR code or whatever other form that you have so that I can put that out in the slack in the newsletter and things like that for.

Yep, yep. Well, it'll point to a page on our website. So I can, I can, I can point you to that, no problem.

Great.

I think Steve had his hand up.

No, I put my digital hand up now.

Nice one.

Sebastian. That. Very interesting. Just a quick question and maybe I don't want to geek it out too much, but how, how do you guys validate that? How do you know that the, the species that you're detecting are actually from native fish in the area and not changes in current or, or you know, DNA material being drifted in from the tropics or remote areas or brought in by a fisherman dumping garbage off the side of his boat?

Yep. No, it's a, that's a very good point and I'm, I'm, you know, thanks for the question. I'm happy to nerd out as much as possible. So. Yes, so the question is how do we know that EDNA is from the local area and not blown in from somewhere? The studies that we've seen conducted specifically in the area where we were collecting these samples showed that EDNA or DNA in the environment has a half Life of about 21.2 hours. So I mean roughly if you think about it, it's almost two full tidal cycles is the half life of EDNA in the environment. So a couple of things there is that, is that first of all the, you know, the, the, the cyclical trends and the, and the longshore current in the, that you have in any environment will provide a certain amount of mixing. But you do have the depuration of edna, the concentration of EDNA that will be available and recovered by any sampling method that's out there. So essentially the benefit that we have with EDNA as a method is that EDNA is fairly short lived when it's out in the water. So the cells that get sloughed off by these organisms, it doesn't tend to stick around for very long. And so we can be pretty certain that the trend is localized. As far as say a fisherman coming in, let's say you've got a massive trolley that's just come in from, well, offshore and then it decides to dump a bunch of, of bycatch. That's not necessarily something that we can control for. So it is a risk. And what we do as a company is when we perform our analyses, we do an analysis of credibility. And what that credibility analysis does is it looks for what are the sequences that we detected within each species? And are the species that we detect expected in that area based on historical data sets? And if they're not expected, and we have relatively few sequences for the species that we're detecting, it elevates the risk that we have a false detection. And so we are open and transparent whenever we deliver our data to be able to say that this detection may not be credible. And so we flag it as a low credibility, low credibility detection, but it doesn't necessarily mean it's a false detection. We need to report it. It needs to be out there and fully transparent so that people can make their own judgments about it. But we also need to provide that scale so that people can say, before I choose to manage for this weird species in this location, in case it's an invasive species, given that it's a low credibility detection, I need to first invest in confirming that that species actually does occur in that location. And that can be more EDNA sampling, that can be more conventional sampling, whatever it is. But first find out that you've actually got the problem before you spend money on the problem.

And just as a, just another side note on that one and the reason I bring it up because I thought it was really interesting. I read a study, I think it was from Nature Metrics and from the universe researcher from the University of Toronto, which were doing same thing with air. They were, they were, you know, doing EDNA analysis, capture, capturing air and they're trying to set the airflows and stuff because they're using filters to actually capture the DNA. And they found a lot of just strange DNA that made no sense. And you know, somebody going, having their lunch near one of their, you know, eating a roast beef and you start finding a bunch of different, they got a bunch of weird DNA and exactly what they were saying, they were statistically trying to say it's getting it out there, putting it on the radar and just letting everybody know that, you know, it's, it's a rare, it's, it's these outliers that, that we really don't know what they are.

Yeah, and that's, I mean, I can't comment on that study, but I mean that's, that's exactly why we, we, we came up with our credibility detection analysis is we don't want EDNA to be a black box. We want it to be fully transparent and be able to parameterize how much confidence you should have, the client should have in the results that we're presenting. We don't want each species to have equal weight in a client's mind because some of them, as you'll see in other companies studies, some of them may not be legitimate detection.

And then we've got Jonathan who's raising his hand.

Great.

Thanks Paige. Thanks, Sebastian. This is super interesting. So I have two questions. I'm not sure if that's allowed, but they are kind of connected. So the first one is kind of a version of what Steve was saying just now. When you're looking at these sort of long term trends in a, you know, disappearance of taxa in a given region, presumably one possibility is that they've just relocated.

Right.

So you know, as opposed to actually died out. So, and presumably the only way that you can really be sure about that is to have other observing networks somewhere else so that you know that they haven't popped up there when they disappeared there. And so this gets to your sort of idea of this observing network, which is my second question, which is why they're connected statistically. How, what distribution of these sort of observing sites do you need to have around the world to be sure that you can pick up these sorts of movements of taxa from one to another?

Yeah, great question. So in some of the validation studies that we did with the local government regulator here in the UK Natural England, we were able to determine the distance, the minimum distance that we need to have because we're sampling over a very large time period and we're, we're sampling larger volumes. So each, each sample will collect about 50 liters or more. What we notice is if we had two samplers that were less than say 100 km apart, we could see that the two data sets were starting to have some spatial autocorrelation to them. And that's partly because the water moves so much, but also because the species that we're detecting in the water also move that much. Right. And they're not necessarily moving in the same direction as the currents and tides. So because everything is so very, very dynamic, we need, we need to have a very large spatial separation between the samplers which overall, when you're building a network that has to go, you know, the, the entire length of a coastline ends up being that it's a very efficient way to be able to, to survey a very, very large expanse of area because you don't need these samples right next to each other. In fact, it's much better if they're not right next to each other. We had, in that validation study, we had two samplers that were separated by about 5 or 6 kilometers. The data that we got from both of those samplers might as well have been the same, the same data set. They might as well have been sat right next to each other. It was effectively the same data. There was no statistical difference between the two and that's five years apart. So there are plenty of studies that are out there where they're trying to take a sample in one location and then they go, you know, 200, 250 meters away or a kilometer away and they think, well, this is going to be an independent sample. It's not, not in the marine environment it's not. It might be if you're looking at sediments and it certainly will be if you're looking at soils. But you know, when you're talking about marine waters, there's just too much mixing, too much movement, too many, too many tidal variables and current variables that are going on.

Okay, yeah, that makes perfect sense. Can I, can I ask my follow up question? I don't want to hold up.

It's okay with me. You'd have to, you might have to fight paid for it.

Okay, Christian, apologies. So this, based on what you were saying there, what I was wondering was whether there's any possibility of leveraging existing observing networks and piggybacking off that in some way. Like for example, the goa, you know, the Ocean Acidification Observing Network does, you know, is that a possibility? Is that a possible route to quicker rollout?

Yeah. So I mean the key advantage that we have with the EDNA is the device that we have is very easy to deploy, very easy to recover. And so if you've got any site where an operator, a user, it can be a company, it can be a research institution, it doesn't really matter if they're making regular trips out to or past a certain area in their regular day to day job, then all they need to do is throw this over the side, let it do its thing for a day, pick it up the next day, and then that's the sample done. So we've made it as easy and as bulletproof as possible in order to be able to get the most reliable sample collection method that we could come up with and as simple as possible for anybody to use it. So yeah, I mean, that's the idea is it wouldn't be us doing it, it would be the people who are experts in the area where they are already doing the work. Like we've got fisheries officers that regularly go out from their local station and those stations are spread all around the country. They can easily just be dropping this off one day on their way out and pick it up the next day. Not a problem.

Okay, thank you.

And Christian, go ahead.

Thanks Paige. Fantastic survey for so long. Very nice to see a longitudinal time series like this. And interesting to see the decline in species richness over the time. Even though at the end of the study period you increase the sample. Well, the volume sampled that had a like a really clear effect on the amount of DNA you recovered and yet the number of species was still low. So quite reassuring.

Sorry, I should mention that the very last part of the study was done in 2023 and we sliced that data off of the time series. So the time series only goes up to 2022. So that we used the same sample type, same sample volume, same treatment, essentially along that entire time series. Sorry if that was unclear. That. Yeah, so we had, we had samples from 2012 through to 2023. The Edna comparison in small volume to large sample was undertaken in samples that were taken in 2023. But the entire time series and the comparison with the. Obviously the catch and count, all of that was done from samples collected 2012 to 2022 to make sure that we're comparing apples with apples.

I see, I see. And would you have seen an increase in richness by the end if you would have considered the large volume samples?

Well, I mean, we did. Obviously with our sampler we were recovering more DNA and more species simply because we were doing a better job of collecting the sample. But yeah, we, we shouldn't let that change the actual trend data because we don't want.

Of course, of course, yeah. But then looking at the, at the figure, you presented some rarefaction figure and it was interesting to see that as you increase the size, obviously the number of species increased, but it never reach a real asymptote, especially for the large volume sample. And one would expect a large volume. You are eventually encompassing or capturing every species present, but it's not the case at all, in fact.

That's right.

Whether you would like to increase even more the sample or whether the different sites you were sampling were heterogeneous in terms of ecology or homogeneous. What is going on there? Why you wouldn't see an asymptote and do you think is that.

Yeah, I thought that was interesting as well. And I'm not sure why it didn't reach an asymptote, but we certainly did have a very large increase in the sample number with each iteration of the in depth sample collection. So we haven't had the opportunity to run the in depth sampler on a large time series. And ideally we would See, we would hope to see and ask them to come out of that in fairly short order.

And the large volume was standardized. Large volume, or it was up to plugging the filter or something like that.

Yeah, so the standardization comes. The device works by drawing water through the filter. So the filter actually sits at the end of the device. And so the first thing to enter from the seawater is, is the filter. And the pump sits on the. On the downstream end of the filter. So it's drawing water through. And the pump works at a fixed pressure. And so the pressure is essentially standardized. The effort is standardized, if you like. And what we've done is we've modeled what is the throughput of water over a long time period over that two tidal cycles in order to be able to know how much water gets filtered over that period of time. And then the onboard computer scales the effort over the full two tidal cycles. So that one liter that is collected at the beginning of the sample is also one liter that's collected over the end of the sample. So you've got an even distribution of water that's collected over the full tidal.

Cycle, provided that the clogging rate is constant or you have a pressure sensor.

No, what we do is we increase the effort over that tidal cycle to compensate for that. So, yeah, we've done a fair bit of testing and created models which tell us how rapidly the filter will clog. And then the computer compensates for that by increasing the effort towards the end to make sure that we collect an even amount of volume of water over the two full tidal cycles over the full programmed time period.

Well, thanks, Christian.

What organization are you from?

I am from New Atlantis Labs and Universidad Austral Chile.

Okay, super cool. Those questions went way over my head. But if you get. If you don't mind me asking a question, since we have a little bit of time. So. Okay, so this was all quite technical kind of content, which is amazing. And I love that we can dig into those things during these NTC nows. But as we do try and write little blog pieces about it for a larger audience, I'm wondering, Sebastian, if you can kind of go over for a more general audience. So when would someone decide to use EDNA for their monitoring? And then how do they kind of choose the company that they go with? I remember when you and I had our initial chat, you mentioned some differences between the different EDNA companies and how they do their sampling. I'd love to just get that on the record here.

Yeah, well, I suppose EDNA as a method is relatively novel for Measuring biodiversity ecology has been done in largely the same way for the past 60 to 100 years. And so whilst we're seeing that EDNA does an overall better job at recovering species and characterizing biodiversity at large scales, there is, there is a hesitancy on the part of regulators in order to be able to change, because the regulators themselves are responsible for permitting certain projects which are going to have impacts. Now, increasingly we're seeing that the evidence is because EDNA does a better job at characterizing biodiversity, particularly in very difficult or complex environments, that there's a big enthusiasm on the part of regulators and of course, on the part of businesses to be able to represent their impacts on biodiversity using EDNA data, simply because it will be quicker than traditional methods or less costly than traditional methods. I think I've shown that there's a good return on investment that you get when using environmental DNA techniques. And so certainly, I mean, as a business, our ethos has been that organic growth is sustainable growth. And we are seeing significant growth in demand for our services now than we did, say, 10 years ago when we first started out. And no one really seemed to know what EDNA was or much less what it could do. As far as differences between applied genomics and other companies, we mainly differentiate ourselves in a few different ways. First of all, in sample collection, we recognize that sample collection is the greatest risk to any project that you undertake. The greatest expense to any project is going to be getting on site and collecting a sample. So the methods for which you collect a sample need to be 100% reliable because you, you have an opportunity cost to collecting those samples which you, which you simply, most projects just cannot afford to lose. So if you can't get that right, then there's no point in actually going out and doing a study. So that's number one. The way that we collect our samples or the tools that we have for collecting samples that we provide to our clients has to be 100% reliable. Secondly, as I demonstrated with this project, the sample needs to be at a scale that is appropriate to the environment that's being sampled. And in a marine environment, for example, where you've got a lot of variability, the sample that is collected has to be able to account for the natural variation that is in the environment. We're talking about very large and complex environments. No matter where you go, it's very, very large, very complex environments. Larger samples will do a better job all the time of being able to collect rare species and have less variation between sample replicates. And then on the other side of the Well, I mean, the whole analysis process is pretty much well known between all companies and more or less standardized. We are quite happy to be very transparent about the methods and the tools that we use in order to be able to provide reassurance to our clients and to regulators that we're doing the right thing. And then, as I mentioned and discussed about the data credibility analysis, we do several quality control analyses throughout the process in which we take the results of that. Well, okay, let me back up. The key advantage EDNA has over traditional taxonomic methods is that we're able to quantify the uncertainties essentially at every stage. When a taxonomist is looking through a microscope and, and identifies an organism, they'll say, all right, well, this is one of, this is one of these pieces. This is one of those species. They're not able to tell you how certain they are. They're not able to quantify how certain they are about each of those identifications. Whereas with edna, we can say we're using a standardized data set and comparing the sequence to the standardized data set, we're able to say with a quantifiable amount of confidence that this sequence is from this species. But we don't just stop there. We use that data then to quantify the credibility of the detection of each species that we, that we get, which then gives our clients additional reassurance that we've done a reasonable job to be able to tell them which biodiversity is in the environment.

Okay, so what you're saying is a good choice for any DNA company is that they're doing large sampling and that they have a credibility kind of rating system.

Yeah. A robust quality assurance system that they can be very open and transparent about. Yeah.

And last tiny question for me, EDNA is useful for any type of environment in bio, or are there specifics?

I would be hard pressed to think of an environment that EDNA couldn't be used for. There are certain applications where it might be more difficult, but certainly in soils, freshwater, sediments and, and seawater, all of these things are, are very applicable. Somebody mentioned air earlier. I think, I think that's something that's interesting to look at. But again, the same issues that you have in understanding the provenance of the EDNA in a marine water sample, you will have that even greater in an air sample. So being able to understand, let's say, in air, EDNA taken from the air, where is that EDNA signal coming from and how meaningful is it? That's, that's still an open question as far as I'm concerned.

I think what just as a comment I think the biggest challenge with EDNA is it's very, very good for qualitative, for qualitative data. It's not very good for quantitative data. So yeah, combining that with just the more sample points, the more we can build these data, larger data sets, I think we'll get closer to being able to generate quantitative assessments.

Yeah, I mean, we, we at Applied Genomics have a workaround for that where we can have a pseudo quantification and that's that I'm not talking about the application of sequencing reads. I think, I think the number of reads that come out of a sequencer is very biased and is not an appropriate measure for quantity or biomass or anything like that, simply because there are too many sources of bias in the entire workflow. So I'm not talking about reeds specifically, but we do have a method for quantifying, let's say, the diversity within populations that we can then provide our clients with an indication of, of. Yeah. What is, what is the diversity within and between populations?

And I think it'll only get better as if you start when we start cross. If, for example, if I'm taking, I'm going back to air, I go back to. If I'm taking and I'm thinking specifically of bat projects. So I'm trying to, I'm capturing EDNA data in the air from, trying to identify bat populations and I start crossing that with EDNA from guano or excrements, then I can start trying to come up, I can come up with an algorithm or model to quantify the populations.

Yep. I think, I think that's a, that's a, that's a good place to start. I think understanding, understanding where the, the airflows are at the time that you're sampling the air is probably a really important factor to put into seasonality. Yeah, yeah, that's right. But yeah, I mean, I mean that's not to say that air sampling hasn't got its uses. I think there are a lot of applications that we're looking into just as much as others are looking into. But you know, we want to be quite sure that we're able to provide a data set which has a certain amount of reliability before we, we're happy to release that to, to our clients.

Go for it, John.

Thank you. So, just one. I just want to make sure I understand one thing. I got the impression from that your in depth sampler, it is able to sort of operate autonomously so it can basically float around and collect these samples and then submit the data back wirelessly to some Central hub and so on, and then be retrieved later. Is that the case? Is that where the sensitive.

No. So it's an automated sample collection device. So what it does is it does one job and it does it really, really well, which is to collect the sample into the filter. And we've purposely not built robotics into it because we know that it's going to be used by some person on an oil platform where some tree hugger has come along and wants to know what fish are existing underneath the platform. And he's got to add this to a hundred other things he's got to do. And if it's a complicated system, it's not going to be used properly and the sample is not going to be taken correctly. And as I said, the opportunity cost is far too high. So the system that we come up with, it does one thing. It collects a sample over a period of time, it can withstand a lot of abuse and it will still keep running. We've had it to the bottom of the North Sea. So it's got, you know, it's got a depth capability of more than 250 meters. And we've got a deep dive version actually that goes to 4,000 meters. And it's just really, really robust. It'll do the job, it'll do it every time and it'll do it reliably. That's, that's the number one thing. Once that, once we have robotic systems that will do all of the other things reliably and can, can also withstand that abuse, then maybe we can look at building that in. But the most important thing, the thing that we have to do right now, more important than anything else, is just start collecting samples. There's a massive opportunity cost because we're waiting for some tricorder or some perfect robotic system to be able to do everything for us. We're missing out on the opportunity to just take these samples, right? Every month that goes by, that's more variation that's happening in the environment that we're not able to capture, we're not able to understand what's going on. So simplest possible system to get it done.

Makes sense.

What's the shelf life of the cap of the samples?

Once the samples are recovered, we have a DNA preservative that is then put into the capsule. And once that DNA preservative is added to the capsule, there's no need for a cold chain. And it will actually last for weeks in that preservative with no degradation? Oh yeah, several weeks. As soon as you've Recovered the sample, the operator will just simply. Well, that's the capsule there where you collect the sample, you pour the DNA into the capsule to flood it with the DNA preservative and then that's it, you're done. You don't need a cold chain, it's non toxic, non hazardous, it can fly, it can be taken on any regular courier route, totally fine. So yeah, there's no need to worry about sample integrity once you've got it preserved.

And questioned.

Yeah, by just seeing the capsule, you just showed, it seemed that the filter, the capsule itself is reusable and you take the filter out of it to extract DNA.

No, the capsule is a consumable.

It's a consumable. So it's like a water type of capsule, for example, similar to that.

Yeah. So. And that's also why we built the sampler to have the consumable as close as possible to the source, the water source as possible. So the device itself doesn't need to be fully decontaminated in between sample replicates. It's downstream of the filter. There's nothing upstream of the filter to contaminate the sample. The capsule ends up being the only consumable in the sampling process.

Then aside from collecting the sample in the way you're doing is very nice, but another sensitive part of the analysis is extracting DNA and building the libraries. Are you doing that in home and do you think that is a part that I don't know, deserves some attention standards to?

Yeah, absolutely. So, so we, I mean we do, we do the whole thing in house and so, and we, we, I mean a lot of labs have gone towards automation for their, their DNA DNA purification. So taking it from a sample to, to a purified DNA state, we see too much risk in that. Actually we prefer to do that process manually to maintain quality in that process because it's only until you get the sample from the field to a purified DNA state. At that point you can actually repeat samples. You've got enough DNA recovered to be able to repeat analyses. You can do multiple sequencing studies, you can do multiple PCR based studies. You can have things fail and you can try them again because you've got the DNA purified from the original sample. Up until the point where you've got purified DNA is 100% risk on the project, you haven't reduced the risk to the project at all. The most important part when I talk about sample collection is all the way up to purified DNA. So you have to maintain quality 100% right? Up until that point after, like if you don't. Well, what's the point? If you have, if you have a great sample but then you're not able to purify the DNA to a very high quality, then you've just wasted all that effort.

And then if you're not a specialized lab and you are dealing with tissue samples with high density DNA, you have.

A bit of a problem.

Right.

Because yeah, there's always cross contamination and all the rest of that. Yeah. Like we, we are very, very careful with how we've structured our laboratory and our workflows in order to be able to make sure that we've got, you know, appropriate measures in place to maintain quality. Yeah, absolutely.

Thanks.

Thanks.

Well, that was a serious deep dive.

Big questions.

Yeah, I agree. Got kind of like eight minutes left in, in this kind of allotted time. I don't think it's enough to go into breakout rooms. So either we have a few more questions.

Yes, I'll geek out one more.

There we go.

You guys have, have you ever had any issues with over saturation of samples from, you know, big species or schools of fish or certain that are in a given area?

Yeah, so that's, that's like what I was talking about with the, the problem with reeds as, as a measure of, of or as a, as a, a technique for quantifying say species abundance that, that you're recovering in. You will get, let's say if you take a sediment sample, most sediment samples will not have a lot of organisms in it, but you'll get a lot of EDNA recovered. But every once in a while you'll get a whole worm in there. And so you'll get a huge representation of that DNA in the sample. And if you were to go by reads as a measure of abundance, there would be a ridiculous number of reads associated to that single organism, which is not indicative necessarily of the abundance of the organism in the environment. So yeah, the way that we deal with the data enables us to be able to remove that bias and still quantify the abundance of arguments that we're seeing.

Can I ask one follow up as well? So back to your point there about this issue of sample collection versus sample analysis. So I totally get what you're saying there and how you can, you can sort of ensure the quality of the purified DNA and that you can do multiple analyses and so on. Thinking about having a global, fully automated observing network, presumably ultimately we do want to get to the point where you can, you can send these things out and they can do their own thing in terms of what makes that possible. Is it, can you kind of, can you automate, kind of get a binary automated analysis of whether the purified DNA you have is accurate is one thing. You know how. I guess what I'm trying to figure out is how intelligent one of these systems could be autonomously and how much it could figure out whether it's got good data on its own versus not.

I mean there are a lot of organizations that have much better resources than we do that have been working on the problem of fully automated EDNA based either a sampling system or a sampling and analysis system. And everybody wants the Star Trek tricorder, right? They all want this one thing that's going to give them the entire thing and they never have to get away from their desktop and all they can do is data analysis. But until we get to that point, the key problem that you're always going to have is that you've got a very large environment that has a lot of variation in it. So how do you account for that variation? Either you take lots and lots and lots of very small samples that you can then put through, you know, some sort of microfluidic device to be able to purify that to a state that you can then get it into a very small analytical device, you know, a small sequencer of some kind which is on site and all the whole thing is robust enough that you can actually be able to check the quality at each stage and you know, either have the, have the go ahead for the analysis or eject it and just take another sample. You know, these are all, these are all possibilities that people I'm sure have been working on. We don't see that technology as being realistic at the moment. And so, yeah, we stick with the stuff that we know is going to be the best way to get the best answer that we can. And the best answer we've found is to be able to take a very large sample, to be able to remove as much of the variation in the environment as possible or to, to average that out, if you like. Yeah, or.

And I think also aligning the data quality objectives. Again, there's a direct cost benefit between data acquisition and aligning data quality objectives. Sometimes we want, there's, there's no, there's a diminishing return on overwhelming precision.

Exactly. Right. Yeah, yeah. And being, being overly precise can, can, can lead to, I think it's called a type 3 error and producing the correct answer for the. Right, so yeah.

That's what politicians do all the time.

There we go.

I've been, I've been working on algorithms to, especially in soil because that happens a lot when we're not a ocean water but matrix of soil.

Sure.

There's just a point where the cost of getting this false sense precision goes way beyond the. So I've been working on an algorithm trying to quantify some kind of indicator whether it's just that diminishing or where we're getting enough answers out of it at a reasonable price. So it becomes scalable.

That's right, yeah. So, so the number of sample of sample replicates that you have to take in order to be able to get to the point where, you know, you, you've sampled appropriately and yet there's a, there's a cost to each replicate and so how many, you know, what, what is that balance?

Right, yeah. And I would imagine the ocean is just an immense. I mean that's. If soil is complicated, I just imagine the, the ocean must be. Multiply that times thousands.

That's it. I mean when, when we came, when we came up with the in depth sampler, we were quite, quite proud of the volume of water that we could, we could filter through it. But then when you still think about the scale of the ocean that you're trying to estimate. Well, it's, it's still nothing, is it? So yeah, it's, it's literally taking a spoonful of soil in, in the middle of a farmer's field and then trying to say, well, is this representative of the field? Right.

And, and, and just in the human interaction with that, with that environment is also. It's just trying to understand. It is so immense because it's the true butterfly effect. What somebody's doing in Hong. A ship leaving Hong Kong can affect your sample in the northern Sea of Europe.

Yeah, yeah, that's right. From, from all the ballast water exchanges and all the rest of that.

Hadn't thought of that. But yeah, kind of beautiful, beautiful image.

Well, think about it. If you have a ship, a cruise ship and you have a bunch of people using a septic system and then the sea and then, and you know, they're leaving Costa Rica or Mexico and they got their septic system full of garbage and they're in there swinging by the northern sea and they're dumping it in the oak in the ocean, what's that doing to the. Imagine the amount of DNA that's being dumped into that environment? Totally from a different area.

Yeah, yeah.

Certainly mad. Well everyone, thank you so, so much. That was really engaging. I learned a lot looking forward to distilling it. Edna 101 deep dive into Edna. And yeah, I'm gonna post the recording. And Sebastian, do send the email with the links necessary so I can share out.

I will just now. And thank you very much for this opportunity. It's been really good.

Thank you very much. It's been really, really interesting.

Yeah. Thank you everyone for your questions. It's been great.

Yeah, thank you. Invest in a page.

Have a lovely one, everyone. Good day.

Evening.

Take care.

Bye. Good evening.

Good day. Thanks.