Cell Painting at Max Planck

00:00 – 00:30
My name is Martin. I'm an application scientist at Araceli Biosciences. But the main talk today is being given by Fernanda, who is a postdoc scientist, biologist by training, and doing a lot of data science right now at the COMAS Screening Institute at the Max Planck in Dortmund. So please welcome Fernanda.

00:30 – 01:00
Hello everyone. Thanks a lot for being here today. I also would like to thank Araceli and Martin for the invitation. And so today I'm going to talk to you about the cell painting on the Araceli Endeavor system and some insights about that. So, as Martin said, we work at the Compound Management and Screening Center at the Max Planck Institute in Dortmund, and this screening facility is led by Doctor Sonja Sievers. And I would like to say just a few words about that. So there we do two types of screening. The first one is targeted-based screening, where you usually do some biochemical assays, you perform some hit selection, and then you look at the cellular effects that are caused on the cells.

01:00 – 01:30
And the other method is the reverse chemical screening, where you have a phenotypic screening such as cell painting. And then, after you find your hit, you identify your molecular target. So those are two valid approaches that we are using at COMAS. And there we also have a big compound management center, which is led by Carina Seitz. And we do have more than 60 collaborators all over the world. They can all submit their compounds to COMAS, and we do have 25K compounds from these collaborators. We also have more than 100K HTS diversity library, as well as commercial libraries, chemical probes, natural products—so all sorts of compounds which we can test.

01:30 – 02:00
Using these platforms—these two robotic platforms for high throughput screening that we have—and we just, in February, acquired the Endeavor system. And we are now preparing to implement it into our robotic system this summer. But we are already using it at the bench a lot. So I'm going to talk to you about the cell painting, how we do the cell painting at COMAS, then how we perform this benchmark with the Endeavor, the new system, and finally the outcomes. So the cell painting assay at COMAS is performed with the classical cell painting dyes. So we are using the six dyes which stain eight organelles inside the cells. And then, after we acquire the images with our system, we do the segmentation, the feature extraction, and of course the morphological profile.

02:00 – 02:30
And with this morphological profile, we can generate these profiles for a lot of different reference compounds, which we can use to compare to new compounds which have unanticipated mechanisms of action. Right. And then we have also clustering methods where we can put our new compounds and find what they are doing on the cells. And the analysis pipeline was developed by our former member and father of the Cell Painting Workflow—which is here today—so we thank Doctor Axel Pahl a lot for that. And, yeah, so in this talk I'm going to talk about—I have three questions, sorry. And so the first one: if Cell Profiler version affects our analysis. And this is because on our current workflow we are using Cell Profiler version three, and we want to move to the newest version, also to update our pipeline.

02:30 – 03:00
And of course because the Endeavor images, they are quite big, so it goes well with the newest version of Cell Profiler. Also we want to understand if the clustering with the four channels are enough, since our old system has five channels acquisition and the Endeavor has four. And also if the feature set should stay the same, because we do have a fixed feature set, and we want to see if this is working also for Endeavor. So we just—I'm going to talk about the details of the cell painting now. So we are doing these assays with U2OS cells. They are cell derived from osteosarcoma, and they are epithelial, extensively used in research. And we plate those cells using an automated equipment, and then after four hours of incubation we treat those cells.

03:00 – 03:30
And to do that we are using the Echo liquid dispenser, which can dispense really small amounts of the compounds. We already know that. And which is nice—that we can do—is treat 1, 384-well plates with 352 compounds per plate and 32 DMSO controls. And we can generate three replicates with this. Plate layout shifts, and the shifts—as you can see here—the DMSO control here in yellow, and then it’s shifting along the replicates. And this is quite nice regarding batch effects. So with that we can avoid a lot of batch effects.

03:30 – 04:00
So after we added the compounds, they are going to stay on the cells for 20 hours. And then we can proceed with the cell painting staining. So we are staining the nuclei, the endoplasmic reticulum, the actin, Golgi, plasma membrane, mitochondria, and RNA. So a lot of organelles. And then we were performing the acquisition with our prior imaging system, and it would take 3.5 hours per plate. So we usually do 12 plates every time we were on the new cell painting dataset. And, yeah, this is quite time consuming. And now that we have Endeavor, this goes down to seven minutes. It's insane. So, really cool that we can get this time down now.

04:00 – 04:30
And which is also nice is that the prior imaging system—we were acquiring nine sites for each well, and this covers 4.6 square millimeters. And now with the new system, we only—with four sites—we cover almost six square millimeters. So this is quite good. We are now having more cells also on our pictures. And as you can see here, this is the Endeavor field of view. Here our prior imaging system. And, yeah, it's quite different. We are really acquiring more cells with the Endeavor and with the same or better resolution.

04:30 – 05:00
And if you do a zoom also on the grayscale images—so, sorry, here I was showing the overlaying of all the channels—and here these are the channels in grayscale. And you can see the DNA, the ER (the endoplasmic reticulum here), the actin, Golgi and plasma membrane, and here the mitochondria. And one thing that is different in our prior imaging system is that we were acquiring RNA in a separated channel. Yeah, in its individual channel. And this is not happening anymore with Endeavor. But then we are going to discuss that on the results, how this affects.

05:00 – 05:30
Okay. So after we did the image acquisition, we can go to analysis. And this usually takes six hours per plate. We do run on an internal cluster. And the way that I set up the analysis is by doing one plate after the other. And then I start the Cell Profiler for each CPU that I have available to analyze each image. And, yeah, this is actually the bottleneck now, because we do this in seven minutes, but then the analysis takes six hours. So, yeah, this is what I'm also working on to enhance and make this faster.

05:30 – 06:00
Yeah. So Cell Profiler—we can configure a lot of modules, and what we are doing here is segmenting the nuclei, the cell, the cytoplasm, and then extracting all the features that we can for every channel and every object on the cells. And finally those features, they can be combined into a database. So, talking a little bit about the set of features that we have comparing Cell Profiler 3 and Cell Profiler 4: Cell Profiler 3—we had 1,700 features for all the channels, all the images of the objects. Sorry. And for the newest version we have 4,200. So that's a lot more.

06:00 – 06:30
By checking the different versions, we saw that the texture is now measuring four directions instead of one, and that the area-shape also has some additional features being measured. So that's why we have this difference. And for the Endeavor we have a bit less—so 3,300—because we are missing the RNA channel here. And then what we can do: we can aggregate this data. So if you take compound A, for example, we are going to take the single-cell measurements, aggregate into replicates, and then finally aggregate into a median profile—which we can then have the values for each feature with their respective median values.

06:30 – 07:00
So these features, of course—we have a lot of features, and we know that most of them can be redundant and do not add anything to our analysis. So we are doing feature selection, of course, and I chose this method that we were using in our old—the current—workflow that we run to perform this analysis. But of course you can choose different methods. So in this case we are taking the replicates for our compound—so the replicate one, then replicate two, so two different plates—and then you take the median cell area, for example, plot all the values and measure the correlation. So if the correlation is higher, this means that this feature is really highly reproducible between the replicates.

07:00 – 07:30
These are called the robust features that we selected only once and resulted in 579 robust features. And this was done a few years ago, and Axel performed that. And these robust features are based on our current imaging system, of course, and also on the current version of Cell Profiler, the oldest version of Cell Profiler, which is the 3.0. And then finally we can have the consensus profile. We can also calculate the Z-scores. This is how we usually compare the different samples. So you take each feature, and then you calculate the median of the DMSO controls and then the median absolute deviation to these controls.

07:30 – 08:00
And then, of course, you can have here, for example, nocodazole, and you can compare all different compounds that we tested with a reference compound, and then check what are the similarity—what is the activation values—for these compounds. And this is quite helpful. This Axel also developed a few apps that we can dynamically explore the data, which is quite nice. And then the induction values, they are calculated based on the number of features with absolute value higher than three, and this is divided by the total number of features. And we have a threshold of 5% to see if the compound is active or not.

08:00 – 08:30
And then the bio-similarity, which is basically a Pearson correlation between the compounds, and we do have a threshold of 75%, or it can go up a little bit more if you need to be more precise. So if you have, for example, compound A, compound B, and you plot all the feature values and then you check what is the direction of those features, they can be highly biosimilar or less biosimilar. So this is our database. These are the reference compounds. This is just an illustration of how many compounds we have tested—at least the reference ones—and these are the biological clusters that we have. Not all of them are published yet and validated, but they're in the process, of course.

08:30 – 09:00
And we have already measured 5,400 compounds, which are referenced ones. Of course there are more, but then from these ones 30% of them are active in 2 and 10 micromolar concentration, and from those 80% have biosimilarity higher than 75%. So we can, yeah, classify at least 30% of the compounds. And for this task with the Endeavor, we have a pretty small—I would say—dataset, which we are working on expanding now that we have the system on site.

09:00 – 09:30
So for this test I had to do some filtering. So I'm using only compounds that have more than 75% biosimilarity and compounds that are referenced, of course, which have known mechanisms of action. So—and then here you can see our published clusters. So here is a UMAP of feature reduction, and then you can see all the different biological clusters that we have. And, yeah, I would like just so you pay attention to the DNA synthesis and the aurora. So those are both—they are compounds that can act on the cell cycle and this belongs to both of these clusters.

09:30 – 10:00
And we want really to separate them, because sometimes they can cause phenotypic alterations on the cells which can seem similar—for example, enlarged cell or multinucleated cells. And what we want on the Endeavor analysis too is to have, for example, those two clusters well separated. So, yeah, also how we annotated the compounds from the images with the compounds from Endeavor is taking the ground truth labels which were defined with our previous imaging system. And finally, how did I analyze this data? I did some UMAPs, and then, of course, I used k-means clustering, which I think is the most straightforward one. You can also choose other methods, of course, of unsupervised clustering, to compare the ground truth labels with the predicted ones.

10:00 – 10:30
And then I ran the experiment at least 100 times so we can get a mean Adjusted Rand Index value, which is equal 1 for a perfect match or −0.5 for a discordant cluster. Now, the first question is: the Cell Profiler version affects image analysis? And to do this analysis we found the 579 features—robust features—which are equivalent in the newest version of Cell Profiler. And then I used the same images as a source, and we hope that they should be highly correlated. Of course, if you plot the cell area from the same images, same source, you would expect them to correlate, I would say at least 0.9 above, because this should stay the same, right?

10:30 – 11:00
And this actually happens. But for some correlation features, they do have low Pearson scores. So on Cell Profiler you can add different modules to your analysis, and this module called Measure Colocalization does have a few discordant features. So, for example, even negative related ones. And this can be connected to some changes on how they process the images on the new version. So this is explainable, but, yeah, this is just something that I wanted to bring up if anyone is using Cell Profiler and wants to know. And, yeah, now for the next two questions that I'm going to answer together: so clustering—four channels—is enough? And should the feature set stay the same?

11:00 – 11:30
So for Endeavor, here you can see the different clusters, and actually it does pretty well. You can see, for example, the aurora and the DNA synthesis cluster—they are well separated. The other ones, they're a bit closer, but this we also see with the Cell Profiler 3. And if you compare with our prior imaging system—yeah, we have, I would say, similar results. These ARI values are pretty close, and we also have a nice separation between the different clusters. But this imaging system was including the RNA channel, right? So what if we remove this channel and check the performance?

11:30 – 12:00
It actually—I would say—stayed the same. It's pretty close too, which indicates to us that this RNA channel is actually very redundant. And I will show you why. Because if you do an overlay of the endoplasmic reticulum and the RNA channel on our prior imaging system, you're going to see the overlap. So here in orange-yellow, inside the nuclei—so clearly this SYTO 14 stain is leaking also to other channels. The same thing happens with the Endeavor. We are not acquiring exactly the SYTO 14 on Endeavor, but when we acquired the ER and the actin, Golgi, and plasma membrane, we still get the same pattern. This is nucleoli inside the nuclei.

12:00 – 12:30
So I would say, going forward, we would keep doing the same panel of stainings, because this just makes sense, and you can still have all the information here. Using Endeavor with four channels—you have fewer channels, so this is good. Just an observation. And what if we add more features? I think you'll notice that we went from 3,300 features to 251, and it's quite a low number, but they all are reproducible between the repeats. And if we add more features, actually it also doesn't—it doesn't make a lot of difference, just a little bit. On those two clusters, they are now closer together, but still I would say it's good.

12:30 – 13:00
Basically, cell painting is very robust. You can do a lot, but you need to have the right set of features, of course, because if you try to use—the question someone was asking—if you try to use the 579 features that we use on Cell Profiler 3 to aggregate the data on Endeavor, we actually got a pretty random cluster. As you can see here, nothing's clustering. We have an ARI of .06. And so, yeah, it's very important when you change imaging systems to check if the same set of features works or not—if you need to do again the feature selection. So this is, going forward now, what we want—to actually get into a consensus of how we are going to analyze the data and which features we are going to use with the Endeavor.

13:00 – 13:30
And also, if you take a look on the feature types: Cell Profiler can export, for example, area-shape, texture, intensity, radial distribution, granularity, correlation, and neighbors. These are the main, I would say, measurements that Cell Profiler can do—the different modules. So if you take a look, the Endeavor selection of features—we have most of them belonging to the area-shapes, so related to the morphology of the cells, and the second place the texture. And then our current pipeline—we have the texture as the main source of the features, and then the other ones. Then you can get into a whole discussion about if also you change the system—so now you have different cameras, different objectives—so this, of course, will impact on your feature selection.

13:30 – 14:00
Another thing that we tried was to do this simulated binning. I just want to reiterate that this binning method actually is an underestimation of what the microscope can do, because if you take, like, the real binning on the hardware of the microscope, this is better than the simulated one. So this is even an underestimation—will result in good results. So we can see that still we have this nice separation between the clusters, and then we can easily acquire these images on 2×2 binning, which reduces a lot the storage that we need to have. So this is very good. So we do have more, of course, data that we are acquiring now, and also with 2×2 binning, and so stay tuned for updates.

14:00 – 14:30
So, what we learned: we compared the prior imaging system with the new. We know that moving to Cell Profiler 4 is going to happen, because now, of course, you want to update your pipeline, and also we need that for the new images from Endeavor. We know that four channels are enough, so we want to keep—I would say—acquiring SYTO 14, because we still have the information about the RNA, which is good. And the feature set is yet to be determined. When we acquire more data, then we can actually get into a consensus about that. And then, of course, 30 times faster—the Endeavor. So it really helps us on the day-to-day tasks that we have in the lab. So this is quite nice.

14:30 – 15:00
The area coverage also helps a lot. So we do acquire more cells, which is, you know, important for a more robust profile, morphological profile. Also we are acquiring fewer channels, and it does not affect the Cell Profiler as a performance. And finally, the expanded usage. So we are actually also using the Endeavor for other assays—so, for example, glioblastoma assays—that I'm doing some live-cell imaging. And it quite helps on the day-to-day basis, because we just go there, take a preview of your plate, and actually I'm using basically every day now. So it's quite good. And, yeah, I want to acknowledge, of course, all my colleagues. Without them we could not do this work. I also, of course, acknowledge Axel for the apps and the cell painting workflow, and also our funding agencies, of course, and the COMAS and the Max Planck Institute.

15:00 – 15:30
I have five more minutes. In case you wonder what this Endeavor is she's mentioning and what it looks like—and we actually had a big announcement yesterday—we launched two more products here at SLAS. The Endeavor Pro is what we have at the COMAS, and we now launched Endeavor Ultra, which can, in some cases, go twice as fast. And we have a more budget-friendly Endeavor Core that is a bit slower, has less of the imaging cores inside, but gets to the same image quality. It's the same easy-to-use software.

15:30 – 16:00
And with these we have, of course, the Endeavor Acquisition software, which is—I hope you would agree—easy to use. We have the Clairvoyance analysis software, which was not used for this, but can be used for a lot of other assays. And what we introduced with the Endeavor Ultra now is the ClaireRT for real time, which gives you a quality readout while you are acquiring the images—so while the plate is still in the machine, you get a readout of your cell count, your signal-to-background, all these kinds of things, focus quality—but you want to validate while the plate is still in there and maybe take action on these quality readings.

16:00 – 16:30
So, use case for the Ultra: of course, four minutes per plate. I used to say the Pro is the fastest high-content imager on the planet. Now the Ultra is. Use case—of course you have a huge library; you have all these CRISPR things to edit genes; you have all these compounds to screen, or you're making your own compounds. It's the fastest thing we can offer. Of course it's automation-ready to the SiLA2 interface, like the other two units. But this would be the fastest offering for the best QC that we can bring to you. That's the new system.

16:30 – 17:00
Then the Endeavor Pro is actually a system that was used for all the cell paint, and it's a system I have live at booth 221... 612—says Endeavor multiple times, because we have three of them there, and one of them is a live unit. And I can show you the speed and the image quality. You want to come and go there. And of course, application here—live-cell assays—because the plate goes in, you do a fast scan, plate goes out, back into the incubator; cell painting for medium-sized libraries, I would say—these kind of things is what you would use the Pro for.

17:00 – 17:30
And then the Core, with the reduced imaging units but still giving you the same quality—more budget friendly—maybe if you have a lot of small runs that you have to do, or if you're just developing the assay and don't want it on automation, you want a walk-up system for that, or if you're doing QC after you have found your leads and you need to validate and then come back and only have, I don't know, two plates a week—then you don't need the Ultra, but we have the Core for that. Gives you the same images. You can't tell what system they're from. It's the same sensor. It's just the underlying infrastructure that makes them a lot faster. That's it for my side already. And thank you for being here.

17:30 – 18:00
So, in summary, the Endeavor family now consists of the Core, Pro, and Ultra. Core is great for smaller labs or assay development, Pro is for medium-sized libraries and more demanding applications, and Ultra is for the fastest, largest-scale screens with added real-time QC. All three provide the same high image quality, use the same software, and are automation-ready. The difference is really throughput and speed, and with Ultra you get ClaireRT for real-time monitoring.

18:00 – 18:30
ClaireRT, just to explain a bit more, stands for Claire Real Time. It continuously analyzes every image as it's being captured, so you get immediate feedback about data quality. You can monitor metrics like signal-to-background ratio, focus quality, saturated pixels, and nuclei counts. Results are displayed in heatmaps for quick interpretation, and if too many wells fail the criteria, you’re alerted while imaging is still in progress. That means you can stop, adjust, or fix issues right away, instead of waiting days for an analysis team to report problems.

18:30 – 19:00
This saves a lot of time and resources. For example, one customer mentioned that in the past, it could take them days before their data analysis team told them a plate’s data was unusable. So, they used to rely on “quick and dirty” spot checks during acquisition. With ClaireRT, you don’t have to compromise anymore — it provides objective, automated quality control during acquisition itself. That is a major improvement over the way imaging systems have traditionally been used.

19:00 – 19:30
And to put numbers on it: the Endeavor Ultra can image a 1536-well plate in under four minutes, with submicron resolution and large well coverage. Combine that with ClaireRT, and you get both speed and confidence in your data. In fact, some customers told us they are now able to stain and image all their plates in a single day — something that used to take them several days with older techniques. That’s a huge productivity boost.

19:30 – 20:00
So just to recap the customer benefits: more data in less time, higher confidence because of built-in quality control, and faster iteration cycles because you can trust the results right away. For cell painting in particular, this means larger datasets can be generated, analyzed, and interpreted more efficiently, and researchers can accelerate discoveries without bottlenecks in imaging or QC. That is really the future of high-content imaging that we’re working toward.

20:00 – 20:30
Let me also briefly touch on how these systems fit into automation workflows. All Endeavor models are ready for robotic integration through SiLA2, so whether you’re running standalone experiments at the bench or integrating into a fully automated screening facility, you have the same interface. That’s been important for sites like COMAS, where the system can operate both ways — at the bench today, and integrated into robotics this summer. It’s a smooth transition, not a separate setup.

20:30 – 21:00
Another point is flexibility. Because Endeavor systems are modular and compact, you can start with one and expand as your throughput needs grow. Core is a great entry point, Pro covers most screening needs, and Ultra is the high-end option. But all three are compatible with the same workflows, analysis tools, and automation interfaces, so data is comparable across systems. That way, labs of different sizes can collaborate more easily without worrying about inconsistencies in image quality.

21:00 – 21:30
For image analysis, while we’ve been using Cell Profiler extensively for cell painting, the data can also feed into Araceli’s own analysis software, Clairvoyance, which is designed to handle large imaging datasets efficiently. Clairvoyance works seamlessly with Endeavor outputs and offers advanced analytics, including machine learning-based clustering and classification. So, users have options: they can stick with open-source tools like Cell Profiler or adopt Clairvoyance for additional scalability and ease of use.

21:30 – 22:00
To conclude this part: the main takeaways are that Endeavor dramatically reduces imaging time from hours to minutes, provides larger field of view and more cells per image, integrates real-time QC with ClaireRT, and supports flexible automation and analysis workflows. That combination makes it possible to scale cell painting and similar assays far beyond what was practical before. This is why we believe Endeavor — and especially Endeavor Ultra — represents the future of high-content imaging.

22:00 – 22:30
I also want to thank the teams that made this possible: the engineers who built the system, the collaborators at COMAS who helped validate it, and the data scientists who worked on the analysis pipelines. Without this collaboration between biology, engineering, and informatics, we couldn’t have achieved the advances I’ve described today. It’s really been a multidisciplinary effort, and we’re excited to see how researchers around the world will use these tools in their own discoveries.

22:30 – 23:00
Looking forward, we’re continuing to expand the capabilities of both the Endeavor hardware and the software that supports it. For example, adding more real-time metrics into ClaireRT, expanding compatibility with cloud-based pipelines, and optimizing binning and storage approaches for ever larger datasets. The goal is to make high-content imaging not only faster and higher quality, but also more accessible and scalable for labs of all sizes.

23:00 – 23:30
So whether you’re a large screening center running hundreds of plates, or a smaller lab just starting with phenotypic assays, there’s an Endeavor configuration that fits. And importantly, data quality remains consistent across configurations. We’re committed to ensuring that image quality and analysis reliability are never compromised, no matter which model you choose. That consistency is what makes collaboration and comparison possible across different sites and projects.

23:30 – 24:00
And with that, I’d like to wrap up the formal presentation. Thank you again to Araceli Biosciences for the opportunity to present, thank you to COMAS for the ongoing collaboration, and thank you to all of you for your attention today. We’ll now transition into discussion and questions, but before that I’ll just emphasize again: Endeavor is not only about speed, but about enabling science — helping researchers get more done in less time, with higher confidence in their results.

24:00 – 24:30
So this is very good. So we do have more, of course, data that we are acquiring now, and also with two by two binning. And so stay tuned for updates.

24:30 – 25:00
So what we learned: we compared the prior imaging system with the new. We know that moving to Cell Profiler 4 is going to happen, because now, of course, you want to update your pipeline, and also we need that for the new images from Endeavor.

25:00 – 25:30
We know that four channels are enough, so we want to keep, I would say, acquiring SYTO14, because we still have the information about the RNA, which is good. And the feature set is yet to be determined. When we acquire more data, then we can actually get into a consensus about that. And then, of course, 30 times faster—the Endeavor—so really helps us on the day to day tasks that we have in the lab. So this is quite nice.

25:30 – 26:00
The area coverage also helps a lot. So we do acquire more cells, which is, you know, important for a more robust profile, morphological profile. Also we are acquiring fewer channels, and it does not affect the Cell Profiler as a performance. And finally the expand usage: so we are actually also using the Endeavor for other assays—so, for example, glioblastoma assays that I'm doing some live cell imaging—and it quite helps on the day to day basis, because we just go there, take preview of your plate, and actually I'm using basically every day now. So it's quite good.

26:00 – 26:30
And, yeah, I want to acknowledge, of course, all my colleagues—without them we could not do this work. I also, of course, acknowledge Axel for the apps and the cell painting workflow, and also our funding agencies, of course, and the COMAS and the Max Planck Institute. I have five more minutes.

26:30 – 27:00
In case you wonder what this Endeavor is she's mentioning and what it looks like—and we actually had a big announcement yesterday—we launched two more products here at SLAS. The Endeavor Pro is what we have at the COMAS, and we now launched Endeavor Ultra, which can, in some cases, go twice as fast.

27:00 – 27:30
And we have a more budget friendly Endeavor Core that is a bit slower, has less of the imaging cores inside, but gets to the same image quality. It's the same easy to use software. And with these we have, of course, the Endeavor Acquisition software, which is—I hope you would agree—easy to use.

27:30 – 28:00
We have the Clairvoyance analysis software, which was not used for this, but can be used for a lot of other assays. And what we introduced with the Endeavor Ultra now is the ClaireRT for real time, which gives you a quality readout while you are acquiring the images—so while the plate is still in the machine you get a read out of your cell count, your signal to background, all these kinds of things, focus quality—but you want to validate while the plate is still in there and maybe take action on these quality readings.

28:00 – 28:30
So use case for the Ultra, of course: four minutes per plate. I used to say the Pro is the fastest high content imager on the planet. Now the Ultra is. Use case: of course you have a huge library, you have all these CRISPR things to edit genes, you have all these compounds to screen, or you're making your own compounds. It's the fastest thing we can offer.

28:30 – 29:00
Of course it is automation ready to the SiLA2 interface, like the other two units. But this would be the fastest offering for the best QC that we can bring to you. That's the new system. Then the Endeavor Pro is actually a system that was used for all the cell paint, and it's a system I have live at booth 221—612 says Endeavor multiple times—because we have three of them there, and one of them is a live unit. And I can show you the speed and the image quality. You want to come and go there.

29:00 – 29:30
And, of course, application here: live cell assays, because the plate goes in, you do a fast scan, plate goes out back into the incubator. Cell painting for medium sized libraries, I would say—these kind of things is what you would use the Pro for. And then the Core with the reduced imaging units, but still giving you the same quality, more budget friendly—maybe if you have a lot of small runs that you have to do—or if you're just developing the assay and don't want it on automation, you want a walkup system for that—or if you're doing QC after you have found your leads and you need to validate and then come back and only have, I don’t know, two plates a week—then you don't need the Ultra, but we have the Core for that.

29:30 – 30:04
Gives you the same images. You can't tell what system they're from. It's the same sensor. It's just the underlying infrastructure that makes them a lot faster. That's it for my side already. And thank you for being here.