It's Monday morning at the Renaissance Boston waterfront. Posters are going up, bright imaging displays are flickering on, and you can hear the buzz of conversations filled with words like AI scientists, multi-omics, and 3D organoids. This is SBI2 2025, the annual meeting of the Society of Biomolecular Imaging and Informatics, where imaging meets information and information meets discovery. I'm Carli Reyes, and in this episode of Science in Real Time, we're taking you inside the energy, the science, and the community of this year's event. For anyone new to SBI 2, here's the backdrop. The society is dedicated to advancing quantitative imaging and analysis and connecting biologists, data scientists, and engineers who want to measure biology visually. The program featured everything from deep learning tutorials and 3D model imaging to informatics and essay design. But this year, one theme stood out above all and that theme was integration. Integration between imaging and informatics, between wet labs and data pipelines, between people who capture images and those who make sense of them. The field is embracing reproducibility, interoperability and cloud computing like never before. And that trend was visible on the show floor too with vendors showcasing integrated platforms and automated pipelines. Of course, it wasn't just about the talks. It was about the people and the ideas that they sparked. At this year, Araceli had an incredible present as SBI2. Victoria Arinella, a summer intern at UF Scripts Institute, presented her poster on improving essay reproducibility using automated image quality metrics. It drew a lot of attention from researchers curious about how better QC tools can boost confidence and speed up analysis, something that we care very deeply about at Araceli.

And here we are in SBI2 in Victoria's poster and we have her right here where she's going to be talking to us a little bit about what this project entailed. and thinking about it. So, this project is one that I did as an intern at the UF Scripts Institute. It was a two-month internship over this past summer as a part of the NSF sponsored surf program. And um I was assigned to be in Dr. Spicer and Scampavia's lab, which is a screening lab that specializes in the high throughput screening of small molecules. Specifically, my project focused on cell painting. Um, which is basically an assay that we can use to stain different cellular components using various dyes and then image using uh fluorescent microscopy. And what was kind of interesting or or novel about this particular project is that I had the opportunity to use the um high-speed and high throughput instrument called the RSLI endeavor which allowed us to increase the high throughput ability of these cell painting assays because often they can be limited in high throughput ability due to the amount of time it takes to image a single plate. But with the RSLi, we were able to image a single plate in under 6 minutes. So, it really was optimal for a summer project because I was able to perform multiple of these cell painting assays and increase my optimization um during each experiment. And so, it was really um really just great for a a short two-month project where otherwise we may not be able to have done too much if the imaging alone were to take hours. Then came Claudia McCown's talk. Claudia is a postoc at UF Scripps Institute and she looked at collaborative imaging and data sharing across research sites. She made a strong case for how open and connected workflows can accelerate translational discovery, breaking down silos between labs and organizations. Across sessions, you could feel the momentum. The multiomics integration talks explored imaging side by side with genomics, transcriptomics, and proteomics. Truly multi-dimensional biology. Meanwhile, technological innovation sessions feature platforms for 3D screening, organoid essays, and automated pipelines that tie hardware and software together seamlessly. And beyond the science, the networking and community energy were unmatched. poster sessions humming with collaboration, bender boots alive with conversation, and an evening reception that felt like the entire imaging world coming together. What makes SBI2 special isn't just the technology. It's the culture of openness, curiosity, and mentorship. It's students presenting their first poster next to CEOs and principal investigators, all united by the same goal, to see biology more clearly. And to cap off the Araceli crew, Kristin, our senior business development manager and new SBI2 president, captured the future perfectly.

This has been a very exciting year at SBI2. Once again, we had a full day of educational and hands-on workshops, which is great for everyone to learn about different things. And then the last two days have been full of really fantastic speakers covering everything from imaging, data science, the interface between the two of them. Um we have a bunch of fantastic vendors here to uh exhibit all of the things that they have to help expand this research and we are excited about next year. We'll be back here at the Renaissance. Um even more exciting speakers to come. Um if you have ideas, suggestions or would like to join the society, definitely do. Um next year is going to be even more exciting and we hope we see you then. Next year's SBI2 promises even more focus on AI integration, automation, and 3D models. And we at Araceli are already excited to be a part of that journey.

And that is a wrap for today's Science in Real Time episode. In our next episode, we'll sit down with Claudia from Scripps exploring how collaborative imaging is transforming translational research and discovery pipelines. If you enjoyed this episode, remember to like and subscribe to Science in Real Time wherever you listen to your favorite podcast. Follow Biosciences for behindthe-scene flips and conference takeaways. And if you were at SBI2, we'd love to hear what inspired you. Share your thoughts and tag us to join the conversation.

Carli: Welcome back to Science in Real-Time, the show where we dive deep into the discoveries, technologies, and the people shaping the future of therapeutic research. I am your host Carli Reyes and in today's episode we're joined by Tarun Vemulkar, CTO and co-founder of Semarion, a company that's redefining how we handle and study cells with their innovative SemaCyte platform. We'll explore how their technologies bridges biology and material science to accelerate drug discovery and how it's opening up new possibilities for scientists working with complex cell models. Let's jump right into the conversation.

Carli: It is amazing to be able to have you here and be able to talk a little bit about Semarion and your journey as well and I thought that it would be very fitting to start a little bit with your background. So can you share with our audience a little bit about yourself?

Tarun: Hi Carli, absolute pleasure to be here, thank you for letting me join. My background is a slightly interesting one, I guess, given that I currently work in the world of cell biology in which I'm a little bit of a black sheep because actually I'm a material scientist slash physicist by trade. So I began my career basically researching and developing the next generation of field effect transistors. They're things that go into microchips that power computers.

Tarun: I then worked in the semiconductor industry at scale seeing how these microchips were manufactured in an absolutely massive way. But then what I really was interested in doing is taking some of the precision engineering and advanced materials from the semiconductor industry and deploying them in more biological applications. So I then went to do a PhD at the University of Cambridge and that was focused really on developing a novel type of magnetic nanoparticle where we actually borrowed things from the magnetic memory industry, so hard drives basically, and that nanoparticle we used as a novel way to treat glioblastoma brain cancer basically.

Tarun: And I guess, long story short, a lot of the learnings about interfacing that advanced material system with the cell biology then turned into the foundation behind Semarion's platform today.

Carli: For those new to Semarion, how would you describe the company's mission?

Tarun: Yeah, so Semarion fundamentally what we're trying to do is really empower drug discovery teams and massively increase the throughput of their cell-based assaying workflows essentially. And that's by delivering them cells, specifically adherence cells, as standardized multi-flexible reagents that can slot into their existing research infrastructure.

Carli: I believe that you have also, you're also the co-founder of Semarion and I couldn't help myself but ask, what was the original spark that led to Semarion? And we'll definitely go more into detail in a little bit, but I couldn't hold myself, I'm wondering what was that initial spark?

Tarun: I think over the course of my PhD, I really always wanted to commercialize the science I was doing and I really wanted to see sort of short-term impactful results from some of the more complex deep technology that I was interested in building. So that was always sort of a ground motivation for me fundamentally, I guess. And then I founded Semarion together with my co-founder, Jerone, and we were both good friends over the course of our PhDs.

Tarun: Jerone is a clinical neuroscientist and I guess at some point we just got to talking about the challenges in the world of clinical neuroscience that I as a physicist, material scientist, maybe didn't quite have much of a handle on. And what he described to me was basically that he was trying to do a lot of these really, really cool, I guess, sRNA-based drug treatments on these complex cell models. It sounded really amazing, but then he ended up saying, well, I do spend a lot of my time just managing and culturing and babysitting my cells and what that tends to lead through is just basically issues in reproducibility and throughput and really a lot of weekends spent going in and feeding my cells.

Tarun: And I guess over the course of us talking about it, we tried to unpack maybe what was the fundamental root cause of this. And we had a bit of an 'aha' moment where we realized, you know, a lot of this stems from the fact that these adherence cells just fundamentally really like to be stuck to a surface, right? Their biological phenotype relies on them being stuck to a surface. And a lot of the inefficiency comes in when you've got to somewhat aggressively take them from one surface, like a growth flask, and then put them into a different vessel, like a well plate in which you're doing an experiment, for example.

Tarun: And that just sort of breeds a bunch of inefficiencies in the workflow. And so we were just wondering, just as a simple concept, well, what if you could leave the cells attached to the surface, but then move the surface itself into suspension? And that sort of became the birth of the kernel of the idea behind Semarion, I guess.

Carli: That sounds wonderful. And as somebody who has experienced the pains of cell culture, I resonate with that a lot. And it definitely sounds like such an amazing technology that we'll definitely go into more detail in a bit. Semarion's innovation really is centered around this SemaCyte cell multiplexing platform. I'm wondering if you can give us a little bit of a high level sense of what is it and why it matters, now informed by that spark that led to Semarion.

Tarun: At a very high level, and maybe the simplest and somewhat most fun way of describing what a SemaCyte is, is basically a microscopic micron scale flying carpet for cells. It's basically a mobile surface that carries a small clusters of cells on it, and it can be released into suspension and then handled by pipettes and robotic automation tools and dispensed really conveniently and easily, basically.

Tarun: I guess the why it matters is we had sort of looked at this process of going from, I guess, culture flask or growth flask to well plate. And over the course of quite a few years of this process being deployed in the pharmaceutical industry, we really felt that this sort of fundamental aspect of adherent cells being stripped and replated really hadn't seen a lot of innovation. And we thought there was potentially room here for an end-to-end solution that could deal with some of the embedded inefficiencies in adherent cell workflows.

Carli: Yeah, that sounds like it makes a lot of sense. And I cannot help but think now about the science of the SemaCyte deep behind it. And I'm wondering, how does it work in the scientific and engineering level?

Tarun: If you can imagine a single SemaCyte, it's a little carrier for a cell. Think of a really little box into which cells are able to fit and attach and remain happily, basically. That individual little box is made of a highly engineered thermoplastic, and it contains within it a very fancy little magnetic actuator, basically, as well as an optical barcode. So these are sort of all of the features that go into a SemaCyte.

Tarun: For show and tell, what I brought with you is a little Petri dish. So essentially, what we give end users are Petri dishes that have SemaCyte embedded in them. I'm not sure if you can see the little pattern there.

Tarun: Yeah, that little shiny pattern is basically an array of 50,000 SemaCytes that have been micropatterned onto the bottom of that Petri dish. So if you imagine that surface has just been broken up into a bunch of little tiles, each tile is a SemaCyte. And for when an end user wants to actually run their culturing workflows, basically, or run a cell assay, they take their cell suspension, simply seat it onto the SemaCyte.

Tarun: The cells drop on. Once they sort of hit the SemaCyte, they fall in and attach. Once they're attached, you just gently agitate the dish, and it releases the SemaCyte into suspension, because the SemaCyte are attached to the dish with a smart glue or a smart release polymer, basically.

Tarun: And at this stage, what you have is you basically have a suspension of little carriers for cells, or little flying carpets with cells attached. And the cells are in their happy adherent morphology. You haven't used any harsh treatments to get them off the surface, because they've taken the surface with them.

Tarun: And they can now basically go directly into something like a drug library for an assay. Or if you're interested in making large batches of these cells, what you can do is you can actually take that cell suspension and freeze it down in cryovials, which is typically how cells are stored in industry. So now what you have is basically cells adhered to a surface, frozen in a vial, that the next time you want to run the same experiment, you just pop over to the freezer, grab them, and you can go, because you don't have to wait for them to recover and reattach.

Carli: What kind of cell types or application could benefit most from this platform? Just considering that there's different cells that require different conditions. And I'm wondering about those cell types and what applications could benefit more.

Tarun: I mean, in terms of applications, we're really focused on, I guess, the end-to-end, in vitro discovery cascade, I would say. So we look at everything from target ID, all the way to optimizing leads. Although I'd say our sweet spot in terms of the way people run assays is usually screening to lead optimization.

Tarun: In terms of the types of cells that could work with the platform, it's really built to work with, of course, specifically adherent cells. So suspension cells or blood-based cells, not really our focus. It's really focused on adherent cells.

Tarun: So that's most typically the things you would think of are maybe immortalized cell lines from a variety of different indications. People are generally very familiar with cancer cell lines because a lot of research work gets done on them. But also I think more and more in this day and age, we're seeing more complex cells being used.

Tarun: So adherent cells that are basically derived directly from patient samples, perhaps, as well as things like IPS-derived stem cells or sort of cells that are derived from IPSCs, basically. And I think there's a lot of value added using a system like ours with some of these more complex scarce cell types, because inherently, SemaCytes make a lot of those screening processes more efficient, and it can also miniaturize those assays fundamentally.

Carli: That sounds magnificent. And I couldn't help but start thinking also about other applications and impacts that I've read a little bit about the SemaCytes, more specifically, its multiplexing capacity. And I was wondering if you can describe what that would enable for researchers.

Tarun: Yeah, absolutely, Carli. I think the multiplexing sort of capacity or cell multiplexing as a concept is what we think of as probably the most valuable and most interesting part of the SemaCyte platform. And really, I think we're trying to address what we saw as a pretty fundamental limitation of wellplate-based workflows, right? So microwells in which your drugs are contained for a drug library, for example, when you're doing a screening experiment. You can typically only add one cell type per well, because it's pretty hard to tell apart the identity of cells under a microscope.

Tarun: But often, you have workflows where researchers are looking at screening panels of cells, tens, in some cases, up to hundreds of different types of cells for essentially the same drug library. And that means they've got to do this in series for every single type of cell. That can be, I mean, A, just painful, but B, it takes a lot of time and it costs a lot, both in terms of reagent, but also in terms of the sort of personnel time spent on it.

Tarun: All SemaCytes within one Petri dish have the same barcode, but you can have multiple different Petri dishes with SemaCytes containing different barcodes. So if you had a panel of cells that you want to screen against a drug library, for example, you would take cell type 1 into Petri dish 1 containing barcode 1, cell type 2 into Petri dish 2, so on and so forth. Once the cells have attached, you just release them all into suspension and you pool all of those suspensions together.

Tarun: What you now have is basically a mixed cell population in liquid, but you can uniquely tell the identity of the cell on every single SemaCyte just by looking at the barcode that that SemaCyte contains. And that's basically a barcode that you can look at in a bright field image of an optical microscope. Yeah, so now what you have is basically this pooled cell population and depending on, I guess, the sensitivity of the specific assay you're running, we can look at up to 8 to 10 different cell types in the single well of a 384 plate at the same time. So you've now condensed your entire set of panel and dropped it all into one well simultaneously.

Carli: This is such a wonderful technology and I cannot help but think also about the folks who make it possible because we always see all these amazing technologies and you really are in awe with them because they're allowing us to really move the needle forward. But I'm really also interested in the humans that make that possible. So I was wondering, you know, developing something like this cannot be easy, especially with all the nuances and all the applications that it potentially has. So I'm wondering what were some of the toughest challenges that you you and your team faced when you were getting SemaCytes off the ground?

Tarun: I think certainly in the early days of building Semarion and building the SemaCyte technology, what we were doing is we were borrowing all of these advanced materials and fabrication methods from the semiconductor industry, right? And we were trying to deploy them in the cell biology applications. So all of a sudden we had to really validate and double check and make sure that any, say, variations in our manufacturing processes or design changes that we were implementing to make our product seamlessly integrate to end users was not affecting our biological outputs in any way because we're trying to marry a system of materials that cells typically haven't been used together with, essentially.

Tarun: And cells can be finicky little things and tend to be sensitive to the environments around them. So a lot of work and a lot of the challenges for us were really in trying to understand exactly how we could build this platform, a, to be deployed seamlessly into end users workflows, and b, to give them the biological outputs and endpoints that they were used to seeing, but significantly more efficiently.

Carli: I think that that challenge perfectly exemplifies how Semarion is such a great example of physics meets biology as well. And I was wondering, how did your team's interdisciplinary background shape this technology? I have already some ideas, but I would love to hear from you.

Tarun: I mean, it's absolutely fundamental to it. I think across our team we have people who identify as scientists, physicists, biologists, engineers. So it's basically people speaking a whole number of different scientific languages, right? And that makes, I think, for really interesting discussions and really interesting approaches to problem solving. So it means that we can sit in a room and you get scientific perspectives on the exact same topic from a whole bunch of different angles. And you really get to sort of explore that journey together.

Tarun: So it's been a really interesting and rewarding, I guess, experience building this common knowledge and sort of foundational understanding, and then being able to have all the different insights slot into that. And I think overall, that means that we're able to tackle problems in a much more dynamic and creative way than we would have been able to, were we not such a mixed group of scientists, basically.

Carli: Yeah, it sounds like it definitely would be. And not only for the scientific community, but also for all the scientists that partake in this. And going back to the interdisciplinary nature of Semarion, this highlights even more how much of a robust technology it is. And I was also pondering, you know, as I am talking with the CTO and co-founder of Semarion, I was wondering if there was any, what was the biggest learning curve for you personally, in taking Semarion from this idea that is incredibly groundbreaking. And if we would have even mentioned it a few years ago, I can hear already people saying, you know, that sounds like a great idea. It would be amazing if we could materialize it. And you folks actually did that. And I'm wondering, what was that learning curve for you?

Tarun: I think there's maybe, there's two points I'd like to touch on here. I think the first one is from the perspective of, I guess, being a scientist transitioning into turning your technology into a product, right? Scientists, I think, tend to take a science-first or maybe even a technology-first approach to their work. And what I had to mentally transition into was taking a product and application-centric approach to our work.

Tarun: And that meant we really had to, we had the proof of concept built, but in the early days, we had to do the legwork and go out there and speak to end users and pharma companies and biotech companies and really, really deeply understand what a lot of the challenges were that they were working with. And then how we would adapt the technology to address those challenges now that we knew that it could actually do what we needed it to do. So taking this sort of, yeah, this product and applications-first approach and making sure that we gave the end users what they needed, as opposed to just being technologists who could geek out on inevitably and infinitely optimizing and building their technology.

Tarun: There was a balance to be struck there, but yeah, changing it into the product-centric approach was a big learning curve for me. And I think the other part was what we touched on a little bit earlier, is when you have this diverse team of people approaching deep technical problems from a variety of different perspectives, getting them all to speak the same language and pull in the same direction is honestly, it's a really interesting but super rewarding sort of experience to grow through. And I think just going through that has led all of our team to grow a lot and it's been a really enjoyable and rewarding process.

Carli: Absolutely. And I cannot help but start thinking about what is ahead for you folks, the future of SemaCytes and the future of Semarion. And I'm wondering, where do you see then the SemaCyte cell multiplexing platform evolving in the new few years? Are there any new features or expansions that you folks are working on? It sounds like with such an interdisciplinary team, you folks are never short of ideas. So I'm wondering what has come up, if you can share.

Tarun: Yeah, ideas is certainly one thing we're never short of that I can attest to. But this is where I guess, refer to my previous answer, taking a product and application-centric approach really, really does matter.

Tarun: And so I think something that we're really excited about in the next couple of years is expanding into some of these more complex cell types and more complex assay types as well. So I think something that we've seen a lot of interest in the end user community for is things like being able to work with patient-derived cell material. Patient-derived cell material is scarce and it is hard to screen.

Tarun: And so if you can have a technology where you can do something like say, combine cell material from multiple different patients into one experimental condition, you can miniaturize and multiplex your assay and increase the breadth and depth of data that you can get from it. And something like that in the area of something like biomarker identification is an area that we're pretty interested in. Something else that we've seen a lot of interest in is in the area of neuronal cell cultures.

Tarun: So we've had a lot of people talk to us about the challenges associated with working with neurons as a cell type, but in general, the complex co-cultures that are needed to really be able to get a good modeling of disease-relevant phenotypes in a dish. And that usually means that you've got to do something like combine neurons with astrocytes, with microglia, and the ratio of those actually matters as well. And so these complex cell co-cultures are, I think, an area in which SemaCyte multiplexing could actually be pretty impactful in the future.

Carli: Absolutely. And for folks who are curious about learning more about the SemaCyte platform and specifically how they can address these bottlenecks that I'm sure many folks are experiencing in their labs, especially with applications like multiplexing, I'm wondering where should they go to get in contact with Semarion?

Tarun: Yeah, I mean, they can definitely drop by our website. We have a bunch of resources and application notes on there. You can also follow our LinkedIn for more updates from us. If people are based in Europe or the UK and they're going to be at the ELRIG Drug Discovery Conference in October, we'll be there. Come find us.We'll also be at SLAS in Boston in January. So yeah, do come find us there as well if you're interested in saying hi.

Carli: Yes, absolutely. And finally, to wrap up, I'm curious if you have any piece of advice that you'd shared with scientists or even entrepreneurs like yourself who want to bring new technologies into this world. You already have shared a lot of your process, but I'm wondering if there's one main piece of advice you would give our listeners.

Tarun: There's probably a lot of pieces of advice I could give about starting a technology company. I'll try to boil it down to one. The first thing I'll say is if that's something that you're interested in, it takes perseverance, but it is incredibly satisfying and rewarding to see the science that you've built go out into the hands of other people and being used with real impact. So as a scientist, if that's something that drives you, then I definitely encourage you to look at entrepreneurship as a potential route towards that, I guess.

Tarun: In terms of advice, the one thing I would say is in a small company where you have a foundation in sort of deep tech or deep science, there's often a lot of uncertainty. There's things like funding uncertainty. There's things like technical uncertainty. There's things like customer adoption uncertainty. So you kind of have to get really, really comfortable with uncertainty. You've got to get used to making decisions on limited information and not let yourself get stuck in a quagmire trying to find the perfect answer.

Carli: Absolutely. And I think that that would apply to so many folks. We should all be listening, even if we're not doing like actively science on the bench. That is something that we should all definitely take with us.

Carli: And that pretty much wraps up our time together. Thank you so much, Tarun. It's been incredibly inspiring to see and hear more about how Semarion is really pushing the boundary of what's possible in cell biology. And it's incredibly inspiring in so many different ways. I personally cannot wait to see how the SemaCytes are going to keep evolving. I am especially interested in those applications like multiplexing. So I'm definitely going to keep an eye out there. And I would definitely encourage all of our listeners to also do the same and continue seeing how the SemaCytes and Semarion are going to continue transforming research and discovery.

Tarun: Brilliant. Thank you very much, Carli. Cheers for having me. This has been a lot of fun. It's been great discussing Semarion with you and our SemaCyte technology.

Carli: And that was it for today's episode. That was a truly inspiring look at how smart materials are transforming cell-based research. If you'd like to learn more about Semarion and their work, check out the link to their website in the description box below. And if you enjoyed this conversation, don't forget to subscribe to Science In Real Time, share it with a friend or a colleague, and leave us a review or a comment.It really helps others discover the show.

Carli: Thanks for tuning in and we'll catch you next time on Science In Real Time.

Welcome to Molecule Talk. I’m Carli Reyes and this is where we break down specific papers, technologies, and ideas that are pushing biotech into the future. Today’s episode is about something very fresh — a preprint posted in June 2025 called CellCLIP: Learning Perturbation Effects in Cell Painting via Text-Guided Contrastive Learning. Now, before we dive in, a disclaimer: this work is on arXiv — which means it has not yet been peer-reviewed. The data and claims are preliminary. But the idea itself is provocative: it tries to link images of cells with natural language descriptions of perturbations. If it works, it could be a step toward making complex cell imaging data much more interpretable and useful across disciplines. So let’s break it down: what Cell Painting is, what CellCLIP does differently, why it matters, and what open questions we should keep in mind as this field develops.

Let’s start with the basics. Cell Painting is a high-content imaging assay. The idea is to stain different compartments of the cell — the nucleus, the mitochondria, the endoplasmic reticulum, the cytoskeleton — all with fluorescent dyes, all at once. You capture those images at high resolution, and then extract hundreds or even thousands of morphological features. The end result is what’s sometimes called a morphological fingerprint: a very detailed profile of how a cell looks and changes under different conditions. Why is that powerful? Because cells respond in subtle ways. A drug that inhibits a kinase, a CRISPR knockout, or a toxic chemical may each leave behind distinct morphological clues. If you can profile those patterns systematically, you can start predicting things like mechanism of action, toxicity, or even new therapeutic opportunities. The challenge? Interpreting those fingerprints isn’t easy. You end up with tables of features like “Haralick texture metric #7” or “radial distribution intensity in channel 4.” Not exactly intuitive.

Here’s where the new preprint comes in. The team behind CellCLIP asked: What if we could translate those morphological fingerprints into a space that’s aligned with how biologists actually talk — in language? Their approach is inspired by the CLIP model in computer vision, which stands for Contrastive Language–Image Pretraining. CLIP learns to connect images and text in a shared space. That’s why, if you show CLIP a picture of a dog, it can correctly rank the word “dog” higher than “cat” or “car.” CellCLIP borrows that same philosophy for biology. How it works: • On one side, you feed the model Cell Painting images from perturbation experiments. • On the other side, you feed it text — things like drug names, target pathways, or gene knockouts. • The model is trained so that the image embeddings and the text embeddings line up in the same shared space. In theory, this means you could ask the model: • “Show me all perturbations that look like a kinase inhibitor.” • Or the reverse: “Here’s a new image, find me the text description that best matches it.” That’s powerful, because it could make phenotypic screens not just about clusters of features, but about concepts we can interpret and act on.

So what did the authors actually demonstrate? First, they ran retrieval tests. For example: given an image of cells treated with a particular drug, could the model correctly pull up the right description of that drug from a pool? They report strong performance compared to other baseline approaches. Second, they evaluated mechanism-of-action classification. The model could better separate compounds by their mechanism compared to traditional embeddings. Third, they looked at generalization. Could the model handle both genetic and chemical perturbations? The preliminary results suggest it can — at least better than prior approaches. It’s early days, but the concept is validated enough to spark excitement.

If CellCLIP or similar approaches pan out, the implications are big. 1. Interpretability: Instead of staring at hundreds of abstract image features, you could anchor cellular changes to words — drug classes, pathways, phenotypes. That lowers the barrier for scientists to use morphological profiling. 2. Cross-modal integration: Imagine linking cell images with literature text, with omics datasets, or with drug annotations. Suddenly, you can navigate biology across different data types in one embedding space. 3. Drug discovery: This could make it easier to discover connections — for example, recognizing that a new compound “looks like” an immunomodulator in cell morphology, even if chemically it’s very different.

Of course, plenty of questions remain: • Can the model handle new drugs it hasn’t seen before? • How much does it really learn biology versus just memorizing labels? • What kind of text data is most useful — drug names, target pathways, structured ontologies? • And how well will it scale to the massive datasets we now have, like JUMP-Cell Painting? These are the kinds of questions peer review — and replication by other labs — will need to answer.

So, here’s the takeaway: CellCLIP is a fresh, preliminary idea that tries to bridge the gap between how cells look and how we describe biology in words. It’s not peer-reviewed yet, but if the concept holds, it could be a step toward making high-content imaging data more interpretable, searchable, and actionable.

Welcome back to Science in Real Time, I'm Carli Reyes and this is Funding Focus, your monthly lens on the business moves shaping biotech and drug discovery. Each episode, we spotlight the deals, policies and innovations that are driving science forward, from billion dollar pharma bets, to AI platforms raising capital, to even governments rewriting the rules of the game. This month's lineup really captures the full spectrum of how discovery is funded and scaled. Today, we'll cover: How Novo Nordisk's $550,000,000 leap into self replicating RNA The UK's plan to cement itself as Europe's biotech hub, Phenomic AI's $20,000,000 raise proving how imaging and AI are converting, and finally the bigger picture of how public and private fundings are shaping biotech's global ecosystem. Each of these stories shows how capital and policy are converging to accelerate discovery.

Let's start with our first story. Novo Nordisk has signed a $550,000,000 partnership with Replicate Biosciences to co develop self regulating RNA therapies for metabolic diseases. Three years ago, sRNA was still viewed as experimental, with most investments clustered in vaccines. By the 2024, early clinical data from smaller players proved the modality's durability, and now with this deal Novo Nordisk is really betting on scaling sRNA into mainstream cardiometabolic care, Another clear signal that RNA platforms have moved from niche to a core pillar of biopharma strategies. Across The Atlantic, the UK government announced a bold 1,120,000,000 industrial strategy to make Britain's Europe's top biotech hub by 2030.

So the timeline here matters. Post Brexit, so from 2019 to 2021, UK biotech saw a real slowdown in foreign capital. By the 2024, recovery was underway but investments still lagged behind The US. Now, this package that includes £600,000,000 for a national health data research service and £520,000,000 for biomanufacturing is meant to supercharge the rebound and create a self sustaining ecosystem. With this push, The UK is trying not to just catch up, but to set itself up as Europe's innovation hub for the next decade.

In financing news, Phenomic AI, a Toronto based startup, closed a $20,000,000 Series A to scale its imaging powered drug discovery platform. Back in 2022, imaging based phenotypic screens were mostly academic demonstrations. By 2024, pharma started to take notice, with a handful of early partnerships. Now, Phenomics' race shows the space has really matured into an investable commercial category. Phenomic combines high content cell imaging with machine learning to map tumor microenvironments and identify novel oncology targets.

Its success validates the broader industry shift towards AI and imaging workflows. What AI is to imaging is no longer a SI tool. It's clear that it's becoming an engine for modern drug discovery, as highlighted by these stories. So to close, why don't we zoom out a little bit? What ties these stories together is the interplay between the public and the private capital.

Three years ago, biotech was in a pandemic driven funding boom, with record breaking ventures all around. By the 2024, the cycle had corrected. Start ups tightened spends, and government stepped back into the gap with industrial strategies. Now, in mid-twenty twenty five, we're seeing something new public and private investment strategies converging. Government packages like The UK are designed to attract VC follow-up funding, while pharma's billion dollar deals are being de risked by early public sector investments in new modalities like RNA and AI.

Rather than operating in silos, capital sources are feeding into one another, creating an interlinked ecosystem that could sustain biotech throughout its next growth cycle. When you take a step back, what stands out this month is how different forces pharma strategies, national policies, start up capital, and even public funding are all moving now in sync. RNA is being scaled, imaging is being validated, and governments are actively positioning their economies around biotech growth. It's rare to see alignments across stakeholders at this level, but it may be exactly what's needed to push discovery into the next phase. As we look to September, two threats are worth watching closely.

00:14 – 01:39
Welcome back to Science in Real Time. I'm Carli Reyes and this is BioScopes, your weekly lens on biotech breakthroughs. Each week we explore three to five of the biggest stories shaping the field, from AI screening tools to new cell models to the imaging platforms that capture it all. This month's lineup really captures the convergence of biology and technology with a special look into organoid biology. Today, we'll be covering kidney organoids that mature on the same nine month timeline as human pregnancy, brain organoids revealing a surprising role in immune cells, a powerful AI called CellForge that builds virtual cells, and finally, heart organoids that sprout their own vascular networks. Each story highlights how discovery is becoming not just faster but more predictive and realistic. Let's dive in! First off, scientists have grown tiny human kidney models that don't just look like kidneys, they actually grow like them. The first long term fetal kidney organoid system. Most lab grown organoids develop too fast, like they are on fast forward, but these fetal kidney organoids unfold step by step over months on almost the same timeline as a baby developing in the womb.

01:39 – 03:07
And that's a very big deal because it means that researchers can finally watch kidney development play out in real time. They can see how stem like starter cells gradually turn into the many parts of the kidney, from filtering units to tubules that reabsorb water and salts to even the supporting cells that hold the whole structure together. And here's the twist: when the team blocked a key signal called Notch, think of it as a molecular switchboard, the whole system broke down. The starter cells started to pile up and while some tubules still formed, the proximal tubules which are essential for detox and nutrient recovery simply didn't appear. High resolution cell analysis confirmed that those cells never flipped the genetic switch to become proximal tubules. The big takeaway is that Notch acts like a gatekeeper of kidney development, and these organoids let us watch that gate open and close in real time. It's not just about growing tissue in a dish, it's really about modeling human organ growth with a level of accuracy we've never had before. That gives scientists a brand new way to study birth defects, uncover the genetic programming running during pregnancy, and even test how medicine might affect the developing kidney. In short, these organoids aren't just mimicking the shape of a kidney, they're actually capturing its tempo and its logic — a new window into how our organs take shape one week at a time.

03:07 – 04:56
Next up, scientists at UCSF used brain organoids — tiny “mini-brains” grown from stem cells — to uncover a hidden role for microglia, the immune cells of the brain. Most of the time, we think of microglia as the janitors of the nervous system. They patrol the brain, clear away debris, and protect against infection. But in these human organoids, they were doing something unexpected. Instead of cleaning, they were actually building, releasing a growth factor called IGF-1 that helps drive the production of inhibitory interneurons. Inhibitory interneurons may not sound glamorous, but they are incredibly crucial. They keep brain activity balanced, preventing circuits from becoming too excitable. Without enough of them, or if they don't develop properly, things go off kilter — and that imbalance has been tied to conditions like epilepsy, autism, and even schizophrenia. What the team found was striking: when microglia were present, more progenitor cells multiplied and matured into inhibitory interneurons. But when microglia were blocked from releasing IGF-1, that expansion stalled. In other words, microglia weren't just supporting brain development on the sidelines, they were actively shaping how many inhibitory neurons the human brain was going to build before birth.

04:56 – 06:56
And here's the twist: this doesn’t play out the same way in mice. Mouse microglia in the same brain region don't produce IGF-1, and deleting IGF-1 from them doesn't change interneuron development. That makes this a uniquely human mechanism, one we wouldn’t have discovered without organoid models that capture early brain development step by step. The big takeaway is this: microglia aren’t only the brain’s immune guardians. In the human brain, they double as growth promoters, expanding the pool of neurons that help keep circuits in check. That reframes how we think about healthy brain wiring and opens a new lens on disorders where that balance is lost. Next up, into the AI frontier. A new system called CellForge is letting scientists build predictive, virtual cells — digital models that simulate how living cells would respond to gene edits, drugs, or signaling molecules, all without lifting a pipette. Instead of one big algorithm, CellForge uses a multi-agent AI framework. Think of it like a team of digital specialists: one focuses on biology, another on data, another on coding. They debate, critique, and refine strategies until they converge on a model that best describes the science.

06:56 – 08:27
Give it raw single-cell multiomics data and a research question, and CellForge outputs a trained model — essentially a virtual cell you can experiment on in silico. In head-to-head tests across six datasets, CellForge consistently outperformed current state-of-the-art tools, sometimes by very large margins. That means sharper predictions for how cells behave under perturbations, from gene knockouts to drug treatments. Most importantly, it is open source. You can find it on GitHub at the link below. Think of it this way: if a lab notebook records what you've done, CellForge is more like a digital copilot. It runs parallel simulations, suggests what to test next, and takes some of the trial and error out of discovery science. For researchers, that could mean faster insights and a more direct path from data to discovery. And finally, to the heart. Stanford scientists have built the first vascularized human organoids — mini hearts and livers with their own branching blood vessels. Starting with pluripotent stem cells, the team used micropatterning and a carefully tuned mix of growth factors. That cocktail coaxed the cells into gastruloid-based organoids that didn’t just beat like a heart, but also sprouted hollow vessels running through them.

08:27 – 10:05
The result? A mini heart that looked and functioned like an embryo at about six-and-a-half weeks, complete with endocardial, myocardial, epicardial, and even neuronal cell types. The same recipe worked in liver organoids, showing that the vascular program is conserved across organs. One key insight: vascularization depends on Notch and BMP signaling — pathways that act as gatekeepers for vessel growth. Why does this matter? Lack of vasculature is one of the biggest roadblocks in organoid research. Without vessels, tissues can't grow large, stay healthy, or fully mimic human development. Now, scientists can actually watch early vasculature biology play out in real time — something impossible to study directly in human embryos. For drug testing, especially for cardiotoxicity, these vascularized organoids could act like living safety screens. And in the long run, they could open the door to personalized medicine — testing treatments on a vascularized mini-heart built from your own cells before they even reach the clinic. What ties these breakthroughs together isn't just the discoveries themselves, it's the infrastructure behind them. These organoids and AI systems have become a biological sandbox — a place where scientists can test ideas rapidly, ethically, and at scale.

10:05 – 11:17
Ten years ago, building a vascular organoid, tracking microglia in human brain development, or even simulating a virtual cell would have been completely out of reach. Today, they’re starting to feel almost routine. And that's the quiet story behind the headlines: a convergence of tools turning biology into something you can actually iterate on, almost like code. It's not meant to replace wet-lab science; it's meant to amplify it, giving researchers the ability to move faster, fail smarter, and see further. So, what’s next? If this month's stories tell us anything, it's that organoid research, AI, and high-content imaging are moving from proof-of-concepts to platforms. The coming years will likely be about scale: scaling up organoids so that they're larger, more vascularized, and closer to full tissue; scaling AI models so that they can handle more complex biology; and scaling access so that these tools aren't just specialized in labs but part of a standard toolkit. And that raises even bigger questions: how will regulators adapt to data generated in silico? How do we ensure models are representative of diverse human biology? And how do we balance speed with responsibility when the tools themselves are evolving so quickly? Those are the questions we'll keep exploring here in BioScopes. And that's a wrap for this week's episode. If you've been enjoying the show, don't forget to subscribe, leave a review, and share it with a colleague who loves science. Until next time, I'm your host Carli Reyes — thanks for listening and have a wonderful day!

00:15 – 01:55
Welcome to Science in Real Time, a podcast produced by Araceli Biosciences, where we share the stories behind the breakthroughs shaping the future of drug discovery. I am your host, Carli Reyes, and today I'm joined by Doctor Naina Kurup from Ginkgo DataPoints, a business unit within Ginkgo Bioworks. We'll be exploring the release of their GDPx3 dataset. So what is it, how it came to life, and the possibilities it opens for the life science community? Let's dive in. Naina, welcome to Science in Real Time. I'm very excited to have you here. I also wanted to start by thanking you for taking the time during what I imagine has been a very busy post-release moment. How has that been for you folks? So definitely thank you for having us, Carli. It's so great to talk to you and the rest of the Araceli team. Yeah, we released the dataset you’re talking about, GDPx3, around a couple of months ago at this point. And we also just last month released the dataset on HuggingFace. So it's been a lot of people reaching out to us to collaborate or get more information about the datasets, so it’s been exciting times. That sounds wonderful. And I was wondering if we can go ahead and start talking a little bit about the beginning of this dataset. I think that before we go and understand this dataset, I wanted to see if you can tell us a little bit about your role in Ginkgo DataPoints and what originally pulled you into this kind of work, especially since it's such a cool blend of biology and systems thinking.

01:57 – 02:53
Of course. Yeah, so I come from an academic background. Ginkgo is actually the first industrial role I took on. I joined Ginkgo a little over a year and a half ago, and my experience before this, through grad school and my post-doc, was mostly in using all sorts of microscopy techniques to analyze different aspects of neuroscience, including synapse formation, axon degeneration, and all sorts of dynamics within the cell. So I found out about the role at Ginkgo through a friend, and it seemed like a great fit for my expertise and the capability of building out a platform. At that point, Ginkgo was transitioning into a business unit, collecting data points where our goal is really to build platforms that serve as a way for generating data throughout this company.

02:55 – 03:57
Wonderful. That sounds fantastic. And can you talk a little bit about what is the difference between Ginkgo DataPoints and Ginkgo Bioworks? It seems like Ginkgo DataPoints is a subsection within Ginkgo Bioworks. Am I getting that correctly? Yes. I do understand this is a question that I get quite often. Ginkgo Bioworks is a pretty well-known name in the biotech industry. They've been around since 2008. Ginkgo DataPoints is a business unit that we inaugurated a little under a year ago, in August 2024, as part of Ginkgo Bioworks. And the goal is really the same. Ginkgo Bioworks and DataPoints want to make biology easier to engineer for everyone. Specifically, the focus of DataPoints is large-scale data generation for mammalian products, including…

03:58 – 04:20
Drug discovery, target validation, and other assays that can span a range of customers, including academia and tech-bio companies. We provide AI-ready datasets as well as for big pharma, and it's been a very exciting ride so far. I'm glad we get a chance to talk about it.

04:22 – 04:48
That's amazing. That sounds incredibly fantastic, and I'm very glad to start learning a little bit more about how Ginkgo Bioworks is expanding. That brings me then to talking about the protagonist of the day, the GDPx3 dataset. As I am sure our listeners have noticed, that name has been floating around a lot lately, especially in the systems biology and data science circles.

04:49 – 08:50
So, for people hearing about it for the first time, how would you describe it, or what would you say exactly is GDPx3, and why would you say it's such a big deal right now? Of course. So, as all of us are aware, the world is definitely moving towards a more AI-focused approach to many aspects of our lives, and drug discovery is no exception. One thing that all the AI models need is data. At Ginkgo Bioworks, we really want to enable large-scale data generation for such efforts. As a start, and to provide data for open science, we've decided to use our platform to generate datasets that are freely available. So the name GDPx3 really translates to “Good DataPoints Data Drop 3.” There’s a GDPx1 and GDPx2, and then we have GDPx3. GDPx1 and GDPx2 are datasets that primarily involve transcriptomic data. We’ve done large-scale compound library perturbations in different cell types and generated transcriptomic data using a technique called DrugSeq. GDPx3 is a complement to those two datasets. We used some of the same cell lines and some of the same compounds to generate high-content imaging data using cell painting. The hope is that GDPx3 can be used on its own for model training or drug discovery, or in combination with the GDPx1 and GDPx2 datasets that we released. That sounds wonderful, and I cannot help but want to get into the details of it, because I think that's where it really gets very fascinating. I was wondering, I hear a little bit about GDPx3, GDPx2, and GDPx1. What kind of data specifically would people be able to find within GDPx3? I know we have transcriptome metadata and perturbation info. How can you break it down for us? Of course. I think when we talk about GDPx3, it's great to talk about it in combination with GDPx2. In both datasets the fundamental perturbation is chemical perturbation. We use small molecules from a publicly available chemical library called the LOPAC library. We used a subset of those compounds and different cell types. The advantage is we used a mix of primary cell types as well as commonly used cancer cell lines. Then we dosed these cells at different concentrations of the small molecules and did a readout of either transcriptomics, which comes as GDPx2, or high-content imaging, which comes as GDPx3. So when we give you the GDPx3 data package, what you get is the raw image files, an explainer of how we acquired the data, and all of the associated metadata for each sample. Each sample would be a well in this situation, and you'd have information about what cell type it is, how long it's been treated, what concentration of the compound was used, and how many replicates we have of the same sample in our dataset. All of this will hopefully help you to use the dataset well. I cannot help thinking that this definitely doesn’t sound like just an “upload and go” situation. It's definitely something that requires a lot of preparation, and I can't imagine that there weren’t a few hurdles—not only scientifically, but also operationally—with something of this scale and magnitude.

08:51 – 10:08
I'm wondering what are some of the challenging aspects of getting this dataset into the shape that it is now. What comes to your mind when thinking about how this came to be scientifically and operationally? That's a great question, Carli. This was, to start with, an effort across a team of people, including imaging scientists like me, bioinformaticians, and scientific computing folks. The idea was that we wanted to showcase our ability to generate such datasets at scale and with rapid speed. Something about GDPx3 that we're quite proud of is that we were able to go from actually seeding the cells in the plates to getting the primary analysis of the data in one week. The perturbations that we've done include seeding cells at a certain density and doing compound perturbations at different concentrations, including positive and negative controls. We also imaged these cells at a 24-hour time point and a 48-hour time point. The goal here is to have diversity of data in both the concentrations of compounds and the time points that you get for this data.

10:10 – 10:19
That sounds wonderful, and I absolutely agree that a great team definitely does make the difference. I think this is a perfect example of how

10:20 – 12:18
You folks are basically moving the needle forward and creating a path for what science is going to be in the future. It is such an honor to be able to have you on the podcast. Naturally, this makes me think about what this dataset unlocks. Who do you think stands to benefit the most from a dataset of this nature? Is it mostly tailored for academic researchers, pharma pipelines, or synthetic biologists? Or alternatively, is it really for anyone just working at the cell level? What do you think? So we're definitely user agnostic when it comes to who wants to use our datasets. We've had download requests from both our website and HuggingFace, and there’s been a diversity of folks who have downloaded the dataset and asked us questions about it. We've had people from academia, people from smaller tech-bio companies, as well as larger pharma and biotech companies reach out to us for this dataset and for additional collaborations that we're excited about. So definitely whoever benefits is our user base. I think especially for academic labs that probably do not have the resources to generate such a dataset, this is really valuable. And for tech-bio companies that are starting out and want to see what kind of data Ginkgo can generate for them, this is a great primer to what we can do and how we can partner with them. That sounds marvelous, and I hope all of our listeners are taking note, because this is an incredibly good resource to have. This is also a great segue to a big player I wanted to thank and talk about a little with you, which is cell painting.

12:20 – 13:12
It is such a powerful and elegant technique, but I know that it's not without its challenges. So I was wondering: what did the cell painting workflow look like in this context? And is there anything that you learned from it? Of course. Cell painting has definitely been modified to adapt to different cell types and different questions. We modified the protocol a little to suit the workflows that we have. Something that really helped push the workflow forward was having an Araceli Endeavor microscope. Previously, efforts at Ginkgo and other places used other high-content imaging systems where getting cell painting data, which is imaging in 4 to 5 channels,

13:14 – 14:11
and doing around 4 to 5 sites per well, annotating a plate would take hours per plate. Whereas with the Endeavor, we've been able to generate the dataset in under ten minutes for one plate, which has definitely been something that drove us to get such a speedy, scalable workflow. We can do both the actual staining and imaging of all the plates in the same day, which has been so much easier to work with than with previous techniques. That sounds incredible. As a scientist myself, imagining a plate being read in less than ten minutes sounds glorious, honestly. So it really sounds like if we were to take a step back and think about how this dataset was generated—and in the timeline that it was generated, which I think is the key part here—this could have been generated before, but never as fast as it was with the Endeavor. So it seems like the Endeavor is a technology or infrastructure standpoint that is very promising for cell painting. Would you agree with that? Oh, definitely. I think so. This is just one of the datasets that we've drawn, and we have consistently delivered to customers in short timelines. Especially when you imagine the lab and looks-like scenario, you have things you want to test, and then we generate the data, and the workers come back to us with modifications or things they want to test. So having that turnaround time of a week from designing the experiment to actually having primary data has been really helpful, and the Endeavor has been key to getting to that speed and scale. And I'm also assuming…

15:08 – 15:15
this particular stage of the generation of the dataset is where some of Ginkgo’s internal muscle really showed up.

15:17 – 17:04
You folks are very strong when it comes to automation, biofoundry infrastructure, and even data engineering. Did that play a role in pushing the boundaries of what you were doing with this dataset? Definitely. Especially coming from an academic background, where I didn't have this sort of infrastructure to play with, this was eye-opening and a huge growth experience. Even though the cell painting assay was new to us, the existing infrastructure we had for automating all of our assays really helped drive this quickly. I also have to mention the support we had from our in-house LIMS system, where as soon as we design the experiment, we can ensure that all of the metadata is tracked—from cell seeding to perturbation to image data collection and analysis. Our metadata is conserved, and as an experimentalist, it’s something I don’t even have to think about or worry about. That has been a great benefit of being in the Ginkgo realm and having such infrastructure to work with. We also had a lot of help from the scientific computing team so that the data was seamlessly uploaded right after acquisition. You acquire the plate and then don’t have to think about the data until the analysis is complete. Then you have images that are passed for QC, which you can look at for downstream analysis. Without all of these things, I don’t think we would have reached the scale we have at this point, and for that I’m really grateful.

17:06 – 17:14
It's been such a pleasure to witness how you folks have been able to enable this sort of technology.

17:15 – 18:26
Really, being able to witness how everything comes together to create such an amazing tool that not only scientists will be able to use, but even beyond—you folks are certainly pushing the boundary. So I cannot help but think: where do we go from here? Is there a phase two coming for the GDPx3 dataset, or are you folks thinking about an entirely new dataset now that the infrastructure is in place? Tell me a little bit about the future—what it holds for you. Of course. Yeah, we have plans to do our own analysis integrating GDPx2 and GDPx3 data, and that should hopefully come out in the next couple of months. There’s interest from other groups as well as ours to combine transcriptomics and phenotypic high-content imaging data, along with metabolomic data and even data from agri-studies. Ginkgo DataPoints recently released an ADME service, so it would be interesting to see how we can synergize with those offerings as well.

18:27 – 19:14
GDPx3 was really about chemical perturbations. We were looking at a small molecule library, something we’ve been developing in-house and have data for customers on, but haven’t really had a big public data drop for. The key part is seeing how generic perturbations affect the transcriptome and the phenotypic space. That’s something we’re looking forward to sharing with the community next week. Of course. For folks who are particularly excited and want to start exploring the GDPx3 dataset right now, where should they go to find it? Of course. Our datasets are available through both direct download from our event site and through HuggingFace.

19:15 – 21:44
While people download the datasets from either source, they reach out to us with questions that we are very happy to answer. They also reach out to us with collaboration or partnership requests. I’d say keep that coming. If you’re more interested, reach out to our website. We also have a LinkedIn page where we constantly share what’s happening in the Ginkgo DataPoints space and how partners can collaborate with us. That’s definitely a space to look out for. And definitely just reaching out to anyone on the team—the DataPoints website has information about people on the team, so that would be a great resource as well. Of course, I’m happy to answer any questions if you’d like to chat with me as well. That's wonderful. That sounds incredibly amazing. And that basically concludes our interview for the day. It was amazing to have you, Naina. Thank you again so much for giving us a glimpse of what is happening behind the curtain with the generation of GDPx3 and beyond, and also giving us a taste of what we should expect from Ginkgo DataPoints in the future. It seems like there are a lot of great things on the horizon, and this isn’t just a dataset—it’s a whole new level of biological context. I, for one, cannot wait to see what you folks discover and how you make it possible. Hopefully we will cross paths again in the future. I’d love to have our listeners learn more about the other datasets you are generating. It was truly a pleasure, Naina. Thank you so much. Carli, I always—it's just a holdover from academia—I love to talk about the science that I'm doing. At Ginkgo DataPoints we are still very open to collaborating and promoting open science as well, so I’m happy to talk to anyone who would be interested. And thank you for having us on your platform. That's it for today's episode of *Science in Real Time.* A big thank you to Doctor Naina Kurup for sharing her insight into GDPx3 and the innovation behind it. If you enjoyed this conversation, don't forget to follow, subscribe, and hit the notification bell so you'll be the first to hear about our latest episodes. Until next time, I'm Carli Reyes. Thanks for listening.