Cole Hons: Greetings, fellow Homo sapiens and welcome to The Symbiotic Podcast. I’m your host, Cole Hons. For this episode, recorded on August 20, 2020, we set two new records for the show – one for the greatest number of guests interviewed and second, for the greatest number of scientific disciplines represented in the same conversation.

The project you’re about to hear explained is truly remarkable, and answers the question, “what do Raman spectroscopy, epidemiology, 2D materials, carbon nanotube filters, and deep-learning algorithms all have in common?” All of these disparate topics – and more – have a role to play in this team’s research.

In a nutshell, the scientists involved are developing new techniques that could one day make possible a handheld device that could detect, in a matter of minutes, not only the SARS CoV-2 virus responsible for the current pandemic, but a wide range of other viruses.

Through an ingenious process of sample concentration and enrichment, this technology could detect these viruses – and also real-time mutations of these viruses - at the nanoscale – without requiring the presence of antibodies.

To tell the story of this work we’re about to hear from the following 5 collaborators.

Maricio Terrones is Verne M. Willaman Professor of Physics and Professor of Chemistry and Materials Science and Engineering at Penn State. His research centers on a wide range of bio-applications for nanoscale materials and devices.

Elodie Ghedin is a Professor of Epidemiology at NYU’s School of Global Public Health. A molecular parasitologist and virologist, her research spans biology, genomics, infectious diseases and viral infections.

Tim Yeh is an assistant research professor of Physics at Penn State, focusing on point-of-care technologies for the rapid detection of infectious diseases.

Shengxi Huang is assistant professor of electrical engineering and biomedical engineering at Penn State. Her research interests include low-dimensional materials, spectroscopy, optoelectronic devices, and chemical/bio sensing techniques.

And Sharon Huang is Associate Professor of Information Sciences and Technology at Penn State, focusing on biomedical image data analysis, machine learning, data mining and computer visualization.

Mauricio Torres kicks off the discussion with a short slide presentation about the work. If you’re listening to the audio of this episode, I encourage you to go check out the video version to see the visuals. You can view that on our website www.thesymbioitcpodcast.com.

And now, our conversation with the scientific team doing this remarkable work.

---

Mauricio Torrones: I think I will explain basically the platform and then the project, the seed project, how this has resulted in a seed project. But we are working on this to detect and identify different viruses, and these include flu viruses as well as other coronaviruses. So the idea here is, what you can see is that this is the structure of a flu virus, which is you have the lipid bi-layer, which is this sort of shell. You have the genetic material, which could be RNA or DNA. And then you have the surface proteins here. And the diameter of these virus, in the case of flu, it's around 100 nanometers. In the case of the coronaviruses, maybe it's 155. So when they mutate, usually you have this envelope protein then sort of changes. And when that changes, your antibodies are no longer effective, but the size does not change.

So the question is, can you develop a technology which is label-free without using antibodies to detect and identify these viruses? And this is something that started several sort of years ago with the Tim [inaudible 00:01:55] here and he developed a technology based on growing aligned carbon nano tubes. So basically we can grow patterns of aligned carbon nano tubes. And by controlling the intertubular distances between the tubes, then you could track by size. You could track different particles, including viruses. So you can build a microfluidic device to induce [biclosinate 00:02:20] with a polymer. And we're talking, this is the first generation of devices, this is one square centimeter. And you have an inlet and an outlet, and this sort of goes through. It's like a three-dimensional filter.

So let me just give you an idea. So you have your samples here, your virus is going, your liquid goes in and then viruses get trapped and you have the smaller particles that it goes through. So the idea is that you end up having a very effective concentrator of viruses that you can then analyze. So just to show you an image of this is the following, this is the H5N2 virus, and you see the carbon nano tubes and then the colored ones are the viruses. And this is very effective. This is the third generation. And we started this in 2014, 2013. So it's been a long time. But here, the point is that when you process a sample, you pass through all the samples for the device, and then you are able to enrich viruses. For instance, if you do the PCR and then you collect only one part of the sample you used to get part of the volume, then you might get a false negative if the concentration of virus is not high. But if you pass through all the sample through the device, then you can concentrate most of the viruses through here.

So the whole idea here, and that that was done in 2016, was published in 2016, was that you can enrich viruses samples. Like, for instance, before enrichment, the sample could have only 1.6% of viruses, but after enrichment, you have around 90% of viruses. So once you have [crosstalk 00:04:02] enrichment, this is very important because then you can develop other technologies to detect. And this is when it comes to Raman spectroscopy, and that's the project that we're working on. Let me just give you an idea with the ... This is the second generation. It's called [virion 00:04:25], so virus capture with rapid Raman spectroscopy detection and identification. And the idea here is the device are a bit longer here. These are for high flow rates. And what you are doing here is basically the similar thing. But at the same time, once you capture the viruses, you will record a Raman spectra, which takes you, let's say one minute or two minutes. And then with that Raman spectrum, you can then compare it with the database and then you can maybe identify different types or subtypes of viruses. This is the key point.

Just to go through quickly, this is the idea. The way it works, so you shine a laser to the sample where you have the viruses here, and then you get a Raman signal, which is based on vibrational modes of the proteins that you have there, or different molecules that you have there, and it has fingerprints, and each virus would have its own fingerprints. It's very fast. And just to show you here, these are different Raman spectra of different viruses, and you can classify them using a machine learning model, and then you could basically get very, very important significant data. So the second part of the work is using these Raman spectroscopy, and this is a recent publication that was developed with Elodie [inaudible 00:05:51] and team, and the team as well, we're working on this. It's for human respiratory viruses, and you can then identify different types of viruses or subtypes here using this technology of using Raman spectroscopy enrichment. So our team consist of virologists, engineers, physicists, chemists, data scientists. And that's why we are all here.

That's the summary of the work. And what we're doing right now with the 2D materials is trying to enhance the signal. Can we enhance this signal here that you see here? Can you enhance the signal in the Raman spectra? And can you get better accuracies? So right now we have the accuracies between 90 and 95% for identify different viruses. And of course this works for flu, influenza A, and other ... we're working on coronaviruses like chicken coronaviruses, but we are getting more in-depth information. So this would be the summary of the whole technology in a few slides. And yeah, I don't know if that sort of gives you an overview of what we're doing.

Cole: Got it. And for those who are listening to the audio podcast, I'm going to make sure to encourage everybody to go watch the video to see these sides as well, because it adds so much in terms of the understanding. So is this group right here pretty much the first group in the world to try and put things together in this particular way? Is that right?

Mauricio: Well, I mean, maybe people can answer. I haven't seen this before, so I don't know of anyone that is working on the same problem the same way we're doing it. We got NSF funding for this growing convergence research project. So I think we are maybe one of the groups that have this different expertises and trying to communicate and solve this problem. I don't know. Maybe other team members can add to that.

Elodie Ghedin: Well, if I can chime in as a ... I bring the virology component to this team. And as a virologist, it's been sort of the Holy Grail. If you could enhance even your enrichment of virus particles from any kind of sample, it would really allow you to do some virus discovery or characterization or identification, and the tools to do that are not very good. There are different ways of enriching, but nothing like this. So there are people developing some microfluidic platforms, but not combined in the same way as Mauricio described, nor any platform that does this kind of capture and enrichment. And what's unique also that Mauricio hasn't mentioned is that when the viruses or the particles are captured through the carbon nano tubes, they're actually intact. So you could even open up the device and then culture your virus directly. So it's as if it stabilizes the virus particle also. So it really opens up a lot of applications besides just rapid diagnosis. And then of course, there's the Raman component, which is the rapid detection component.

Cole: Thank you, Elodie. And as you said, you were the epidemiologist on the team. Could we sort of go through everybody on the team and hear what each person brings in? Tim? Could we start with you?

Yin-Ting (Tim) Yeh: Oh, yeah. My name's Tim Yeh and I pretty much, I initialized working on this project when I was a PhD student at bio engineering. So this is where I met Mauricio, he was joining to Penn State like 10 years ago. So this is a project I'm working with him and I'm working on the micro ... My expertise is making microfluidic channels, microfluidic devices, and bio engineering for bio application. And then after I graduated ... So at the beginning, we tried to using these to separate plasma, blood plasma from serum. And then after I graduated, I joined Mauricio's lab as a post doc. We tried to further integrate because I know noticed, and this is where I learned we can fine tune the intertubular distance. And I further developing this to trapping viruses particles.

And then I met Elodie. We have a exchange program supported by Princeton University. So I got this exchange fellowship. So I visit, she hosted my visit. So I joined her lab to learn some biologists, to see what this can really help addressing the virus problem, which is enrichment. So this is where I developing ... I met Elodie, just I joined her lab with developing this to fine tune how can we address the biology problem? And I met Shengxi because she studied the Raman spectroscopy and her lab at MIT. I've been following her work. And when she's a student at MIT, she was doing the Raman spectroscopy. And after she joined Penn State, I talked to her to see if we can really collaborate in this study, enhancing the Raman spectroscopy. So I know her since she was in MIT.

And we find Sharon, just we are collaborating on these proposals. She's an expert in machine learning. So she's really helping us to do the machine learning. So, yes. So pretty much I'm doing the microfluidic and this is the project I'm working on with ... when I was student, mentored by Mauricio and Elodie, and we are developing this right now with them.

Cole: Thank you. Thanks, Tim. And you are the PI on the seed fund grant with the Huck, correct? Shengxi, would you like to tell us a little bit about your specialties and what you're bringing to the team?

Shengxi Huang: Yeah, sure. So my name is Shengxi Huang. I am here at Penn State from electrical engineering. So as Tim said, my research mostly focused on 2D materials, Raman enhancement technologies, and how to use these technologies for biomolecule and chemical sensing. So here in this project, also as Tim has mentioned, my focus is to develop this Raman enhancement platform using different types of 2D materials, there are [inaudible 00:12:35] structures, and also the combination of 2D materials and [inaudible 00:12:39] nano structures to really enhance the Raman signals for the viruses so that we can detect the virus, and also very specifically, because Raman spectra are fingerprints of molecules. So they have those special advantage of a high specificity and high multiplexity. So as you can see in this slide, these are some of the 2D materials we have been working on, and there are many, many types of 2D materials and they have different enhancement effects for different types of viruses' molecules. So we are in the process of trying to select the right type of 2D materials and plasma in nano structures to enhance the type of viruses that we're working on.

Cole: Thank you. Sharon, you're bringing the data science to bear on this project, which is an interesting angle as well, more information technology side of things. What can you tell us about your contributions to this effort?

Sharon Huang: Right. So my training is in data science, machine learning, and so I bring the expertise of data analysis and knowledge discovery. And in this project, we have Raman spectra data of different virus types, subtypes and strains. And as a data scientist, we'll develop the algorithms, including deep learning algorithms, to analyze these Raman spectra data, and then develop a system that will automatically identify different types of viruses, different subtypes of viruses, and also even strains, different strains of viruses. So that's the main expertise that I bring to this. And that's on the technology side.

And then on the fundamental knowledge side, we can also look at the Raman data and try to find signatures in the Raman spectra and see what kind of Raman spectra actually correspond to what types of biomolecules that are being detected by the Raman spectroscopy. So machine learning really, you can use it to answer a lot of questions. And when I'm on the team, a lot of times I ask a lot of questions and then we try to develop machine learning methods to answer questions. I joined Penn State two years ago, and I gave a talk Millennium Cafe, hosted I think partly by Huck, partly by MRI. Right?

Cole: Yes. Yeah. Sounds right.

Sharon: So, and I got to know several co-PIs here through that talk. So it's really, really helpful.

Cole: That's great. That's what those talks are for. It's the power of coffee and donuts, right? Everybody has coffee and donuts and listens to each other present about their work, and we hope that scientists come together. So here's an example of it working. So we love to see that.

Sharon: It really ... yeah. Yeah. There was a spark and then things just happened. It was great.

Cole: That's terrific. Thank you. And so for non-technical folks who are listening in and watching, and they want to sort of take all this in and figure out, well, what are we really talking about here? My understanding is, and I think there's a slide for this as well, that you're trying to create a little device that one could hold in one's hand, breathe into, and within a matter of minutes, have a diagnostic of whether you're COVID positive, for example, or may have a different kind of virus in your system.

Elodie: That's the plan.

Cole: That's the end game, right? That's the goal.

Elodie: That's the end game. It's the end game. Yeah, yeah, yeah. That's ... Here you.

Cole: Right. That's it.

Elodie: We're not there yet, but that is sort of what we proposed, even from the ... We started all this with influenza and then the seed was to expand this in a different way, but it's totally applicable for any virus. Right? So for SARS-CoV-2 it's the same idea if we could do this, enrich from ... And that's what Tim has been looking into, what kind of platform to enrich from the air, from aerosol, which is not easy with a virus. It's very diluted. And then with the Raman, be able to miniaturize, and Mauricio can speak to that more than I can.

Cole: So is the idea that in a hospital, let's say, people could have one of these devices, right? Would you just need the device itself or would you actually need Raman technology as well in the facility to read it? Or would everything be self-contained in that device?

Mauricio: So you have a Raman spectrometer, which could be a tabletop, or it could be one of, I mean, portable, handheld Raman. But you need these cartridges that I showed that the team is developing, and these cartridges will be the ones that will be responsible of enriching the sample. I mean, viruses from patient samples, right? So you have disposable cartridges. And so the idea, and this is still need to be determined, you can either use swabs collected from saliva or from the patient, or you can have some sort of, you can breathe through, and then you can get the sample. And then once you have the sample collected, then you have this cartridge, you put this in this spectrometer, and then you will get a signal. And the idea is that you can screen very quickly. If you have the very good enrichment platform, you could then identify the types or subtypes of viruses that person has.

And that's the way we foresee this technology, into this very big vision into the ... That's what we would like to have in five to 10 years, right? Or even, yeah, I mean, in five years, moving into that direction. But again, there are many barriers we need to overcome from different areas and different problems. But if this technology could be used for screening very quickly people that could say, okay, it seems that you have this virus. And then using that same cartridge, you can then ... That's what Elodie was mentioning, you can do other processing of the samples and you can do PCR analysis or next generational sequencing, or even you can replicate these viruses once they are in the device, because the viruses are intact. So there are many ways that this technology can be used for different applications.

Cole: Thank you. That's fascinating. We're going to take a quick break and we'll be back in just a minute with part two of our conversation with these fascinating folks. Thank you.

Commercial: The Huck Institutes of the Life Sciences at Penn State offers six intercollege graduate degree programs in bioinformatics and genomics, ecology, integrative and biomedical physiology, molecular, cellular, and integrative biosciences, neuroscience, and plant biology. We also offer an accelerated, professional science master's program in biotechnology. At the Huck we immerse students in a groundbreaking environment, built on interdisciplinary collaboration among some of the world's most innovative research scientists, and our students receive unparalleled access to leading edge technology and world-class core facilities. If you're looking for a deeper, more holistic grad school experience, you owe it to yourself to look at The Huck. Visit us online today at huck.psu.edu.

Cole: Hello and welcome back to the Symbiotic Podcast. I'm Cole Hons here with five research scientists who are collaborating on a new, exciting diagnostic technology. And I'm really curious to know from the group, beyond diagnosing just SARS-CoV-2 virus, this technology holds promise, as you mentioned, to do a lot of things. And one of the things I was reading about is rapidly detecting mutations within a virus, as a virus rapidly mutates over time, which is a very important factor. Could you talk a little bit more about how that would work with this technology?

Elodie: Yeah, so we're doing right now experiments with influenza specifically, where, as Mauricio had mentioned, we're already being able to detect subtypes of influenza because they have different surface proteins and they seem to emit a different signal, a Raman signal. But now what we're testing is if you take the exact same subtype of virus, so a subtype is like H1N1 or H5N1 or H3N2. And if I take the H1N1 that we know is circulating, and it's slightly different than the H1N1 that was circulating a few years ago, can I differentiate that using the spectroscopy? Well, the Raman, and the device?So here you actually have on that figure, it shows that we've done phase one. We know we can differentiate subtypes. And now what we're testing is if I take an H1N1, but I have different strains that come from different years, that means a different virus that's circulated different years, can we see a difference in the signal? And that's what we're working on right now.

So why is that important? Well, as you mentioned, viruses mutate all the time. And when we select a vaccine strain, it's based on the fact that the virus mutates and it mutates on sites on the surface protein that are recognized by our antibodies. And if it changes your antibodies that you have against the previous virus strain, it will not recognize, or not well, the new virus strain. If we could detect very quickly that there is a new antigenic variant that is circulating using our platform, then we could very quickly identify what that new strain is, and it could accelerate the vaccine selection, for example. And so we're even trying to go even more fine point, where besides just these antigenic changes, are there other mutations we can capture, even if they're not important mutations for the surface proteins.

And so at this point, we don't even know what Raman is totally capturing. Is it just changes on the proteins, or is it also capturing changes in the RNA, the genomic RNA? So these are all things we're planning. We're planning to have answers to all this in the next six months. So that part should go relatively quickly. And then with new technological developments that Mauricio will mention, we're ... and even in the detection level in how we interpret the signal that Sharon is working on, we're hoping to be able to apply this to other things like SARS-CoV-2 and see how that evolves, too.

Cole: Wow. And about how long do you think we are away from seeing this become a viable technology that could be out there in the field helping to diagnose things like SARS-CoV-2?

Mauricio: I mean, we're doing all the tests just to demonstrate that this technology is viable and it can differentiate different types or subtypes of viruses. And we have done some preliminary work with the chicken coronaviruses as part of this seed project that the team is leading. And we have seen that the coronaviruses, because you have the spike proteins, you have a quite different Raman signal, which basically will be very easy to differentiate between the flu virus and the coronaviruses. Now we're going to try different types of coronaviruses from different animals and from humans as well. But we are building these with Cheryl and Juan, who are building this sort of a platform for data analysis and data models that you can differentiate between or among different viruses with a very good accuracy.

Elodie: And actually, Mauricio, just to interrupt you, that's a crucial point, because actually even in a short term regarding the current pandemic, and we're entering flu season, it would be pretty fabulous if we did have a device that could very quickly tell you whether the fever you have is due to flu or to SARS-CoV-2, right? So even at this level, that would be a good development. The question is whether the device could be made this quickly and approved and all that. That's another question.

Cole: Could you unpack a little bit what those steps would look like to bring something like that into reality?

Mauricio: I think we need to test and confirm that the technology works for specific viruses, and then we have a good database. And then after that we need to develop what type of spectrometer we're going to use first, and you have to design specific cartridge for virus enrichment. Once we have designed and have that specific cartridge designed perfectly, then you will have to scale up the production of those cartridges. And if you want to produce billions of these cartridges to distribute around the world, before that you need to get FDA approval of this here in the US, but who is going to make such amount of cartridges, or which companies or which manufacturers.

And then also you will need Raman spectrometers, which could be tabletop or could be handheld. So the handheld devices are ... I think that's a further technology, but I think the tabletop Raman spectrometer could be used, the actual ones. Right? But you need to place the cartridge in the right way. But there's some design there, and so we're approaching to a level of, yeah, okay, we have demonstrated the technology works and then we need to bring it to the next stage. And bring it to that next stage, we will need to interact with different stakeholders, which could be, of course, federal agencies and different companies around. I mean, here in the US, right? First.

Cole: Right. Right. So you take care of your interdisciplinary team here and then you bring in even more players and more disciplines to try and make something viable out there in the world. It's a process, it certainly is.

Tim: So this project, actually, we just like ... We tried to enhancing Raman, and the goal is try to trap in viruses and try to enhancing Raman signals by using the 2D metal Josh Robinson and Shengxi are developing. And so the idea is, this material, it can enhancing Raman signal to an order of magnitude like a thousand times. And so once it has that, we should have a very strong data, and very detailed, like sensitive. Hopefully we can really ... Our target is try to detect new mutations, like point mutations. So we need the really high enhancement, like a very high datas for [inaudible 00:27:06] Sharon to do the detection. So the goal here is try to using these 2D material to enhancing Raman data for Sharon to play with.

Cole: Now, so that I understand, are you trying different 2D metals? And 2D being two dimensions. We're talking about nanotechnology, we're talking incredibly small. And are you trying different metals and running lasers through? Is that really what's happening?

Tim: Yeah. Maybe Shengxi can comment, but yeah, it's a two-dimensional material. And yeah, Shengxi, do you do want to comment on this?

Shengxi: So I think this project mainly focus on 2D metal as a special type of 2D material that Josh Robinson has synthesized. And we have discovered that these 2D models can provide a really good Raman enhancement effect. Based on all the molecules we have tested, it's about 50 to a 100 times stronger enhancement than the other three materials we know so far. So we think this is very promising, and that's why in the seed project we try to integrate those 2D metal together with the enrichment device that Tim has developed to really achieve very sensitive Raman detection.

Cole: Got it. So you're getting more that virus that we saw at the beginning of the show here, a lot of the virus being captured. Then that enrichment stage. And then the 2D enhancement of the Raman is just amplifying that signature so that it can be more easily recognized, right?

Shengxi: Yeah. So it's like a two-step. First step is to enrich the virus and then enhance the signals for [inaudible 00:28:47], so the signal will get even stronger.

Cole: Terrific.

Sharon: From a data science point of view, based on our current experiments, we developed a machine learning algorithm based on convolutional neural networks. And that classifier can differentiate coronavirus from influenza or flu virus with 99% accuracy. Again, like what Mauricio and Elodie was saying, because of the spike proteins on the coronavirus, their Raman signal look very different from the influenza viruses. And we can achieve very high accuracy, like 99%. And with 2D metal and other enhancements, we expect that we can use the Raman spectra data to achieve very, very high accuracy in virus identification, differentiating coronavirus from other types of viruses. And then within coronavirus, we expect that once we have the data and we can develop machine learning methods to differentiate avian from human and from other types of coronaviruses and so on.

Cole: Fantastic. So what are next steps for your team? With all the data that you have now, what's right on your plate at this moment? And where are you headed at this moment?

Tim: Oh, so right now, just we try to ... Of course, firstly, we try to learn from each other. So we all come from different backgrounds. And then we try to really characterize in this device by really processing the clinical samples. And clinical samples, you can imagine, they will be from saliva, from breath, some will be from different types, from liquid, air, and from different volumes, from order of magnitudes difference, like from microliter to liters. So in terms of engineering design, we need to really, if we want to address the clinical samples, we need something we can handle a wide range of volumes, and then the wide different type of ... like breath, saliva, and even tissues.

So the first will be, we have to really targeting the clinical samples, really characterizing ... try to make it we can processing clinical samples. And on the downstream, we try to enhancing the Raman signal. So this is where Shengxi is helping with try to using 2D materials, we try to enhancing app to tracking viruses, we try to enhancing Raman signals. And Sharon is helping in developing the neural networks. So we go different projects going in parallel, but the end goal is ... And Elodie is helping us generating different reverse genetic viruses to let us ... mainly testing the sensitivity and of the sample. So it's a different project we are going in parallel.

Mauricio: Yeah, I think it's the next stage is ... Now we can differentiate between, for instance, different subtypes of viruses and different types of viruses. We would like to then push forwarding to now coronaviruses and different, you know, influenza A. That would be great. And then just once we have also identified that we can even identify different mutations, that will be key, right? So we need to demonstrate different things and to prove others that this technology works so that we can then attract stakeholders to start moving things step by step. But I think it is very rich project in terms of we have meetings every week and we are trying to educate each other and we're trying to speak or develop a new language because everyone needs to understand what the other one is doing. And that has been very exciting.

Cole: Right. That comes up a lot on this podcast, of people from different disciplines having to learn each other's terminology and language. And it's almost like learning a foreign language, right?

Elodie: It's true. Yeah, yeah, yeah.

Cole: But that's when the exciting stuff happens and new discoveries happen that wouldn't happen if you didn't get the right people in the room together, right?

Elodie: Exactly. Yeah. Some of the stuff that Tim was mentioning also is ... one aspect is the sensitivity. And not just whether we can detect a mutation, but also when you have a clinical sample with very low viral titers versus very high, does it enrich enough from the very low so that we can get good, real positives, not false negatives, if somebody is positive for flu or for SARS-CoV-2? And that's sort of the big hurdle we'll have to overcome is the fact that a clinical sample is dirty, right? It has a lot of other things in there. And so we've been doing a lot of work with cultured viruses, but a clinical sample will be more difficult. So that is the next step, is really to go mostly clinical, real samples, do dilutions also to see what is our level of detection.

Cole: Got it. And so you'll probably have a partner or multiple partners for the clinical trials, healthcare facilities, et cetera?

Elodie: Right now I'm at the NIH and I have collaborators who have clinical samples at NYU, at Cornell, at NIH. We can get clinical samples. So I think we have access to things we can test here. And Tim will be spending some time here too, so that we can do this together.

Cole: That's great. Real team effort. [inaudible 00:00:34:39].

Sharon: So we have done extensive literature review, and to the best of our knowledge, there are some previous work led by Stanford university that use Raman spectroscopy for bacteria identification. But for virus identification, and also looking at the Raman signature of different viruses, we believe that our team is the first one to do this.

So we are developing a new technology, but also really developing knowledge for fundamental understanding in these different areas of domains, neurology, and Raman spectroscopy, data science and so on. So for instance, for machine learning, we focus on explainable machine learning because we can't just say, okay, we're not satisfied with just getting really high accuracy in identifying the viruses, but we also want to understand why and how. So we want to develop these machine learning techniques that are explainable, that we can explain to our colleagues from disciplines.

Cole: Terrific. And everybody benefits. That's another big theme we've been exploring on the podcast with this pandemic, this global pandemic, is the entire scientific community sharing with one another in a bigger way than we've seen in the past, that everybody's just grabbing knowledge from everywhere else as fast as possible so that we can have sort of a united front and confront what we're dealing with together. So I see you making a real contribution to that. I salute you all for your work. I think it's fabulous. Is there anything else you'd like to share while we're still together?

Elodie: No, it's fun.

Cole: Ah, good. Yeah. If it's not fun, why do it?

Mauricio: Yeah. I think it's important to just mention that when we submitted this first proposal to NSF or when we published the paper in PNIS, the pandemic was not there. Right?

Cole: Right.

Mauricio: So now it became ... We were just concentrating on flu viruses, but now it's becoming ... It's a game changer. Now people in the future would like to do very quick test for viruses for different people when you arrive at different countries. And so this is very exciting, and every time we meet every week, we get very exciting data and that keeps us moving faster and faster. Hopefully we will make great improvements and breakthroughs in the near future.

Tim: Yeah. And I also want to, yeah, I want to thank Huck to give me the funding to support it. And Huck has been supporting me for my research when I was a student there. And I think I'm the product of this multidisciplinary, and Huck is a very multicultural, and all the people. And we have a Millennium Cafe, like we integrate with each other. So it's a very multidisciplinary environment. And I think I'm the product of this activity. So I really thank Huck with this very good environment. I learn from each other and have been benefit, and I have a chance to interact with Elodie when she was in NYU. So I learned from ... And Mauricio has been also been very supportive to mentoring me. So yeah, I really think Huck in helping supporting me and funding my research, and I'm the product of this in multidisciplinary activity.

Cole: Well, thank you. Thank you, Tim. You make us proud. You're doing exactly what we're trying to do collectively. So it's just fantastic. Well, again, thanks everyone. Thanks for listening to the Symbiotic Podcast. Please do share this with others and for all of you out there in the world, to keep on listening and don't stop co-evolving.