Evotec SE (EVT) Earnings Call Transcript & Summary

Good afternoon and good morning, everybody, to our first virtual Capital Market Day in 2022, which is actually the third virtual market -- Capital Market Day in the last 2 years. The theme is PoS up, which stands for AI and ML precision technologies driving probabilities of success up. You can find the supporting slides for this presentation at our website at evotec.com. A few housekeeping items before we start. You can see on the screen on the left side a question mark, which you can use at any time to ask questions to send them in, and then we will answer them in our Q&A sessions. And in addition to that, I would also like to mention that the webcast will be recorded and will be available as of tomorrow on our website. And last but not least, there is information set forth in this presentation, which is -- which contains forward-looking statements. And this is outlined in the cautionary language on Page 2. And with that, I'll hand over to Werner Lanthaler, CEO of Evotec.

Thank you very much, Volker. Welcome to all of you from sunny Evotec. Today, let's talk about what really makes us in our industry. How do we bring probabilities of success up to make earlier relevance available to make drug discovery processes more efficient and more predictive into the future. As you know, Evotec stands for evolutionary technologies. So this is core of what is our mission, to integrate latest technologies into the drug discovery and drug development process to bring probabilities of success up and to exclude random processes in our industry because they only increase costs and don't help patients. If you go to Page #4 of this presentation, you should see that the agenda will be conducted together with a fantastic team that is working for Evotec. So I'm so proud that I'm here together with my management team, Cord, Craig and Enno, and that we are supported by the best of scientists that you can imagine globally when it comes to the application and to the strategic relevance of AI machine-driven processes in drug discovery and development processes. So I'm here with Uwe Andag, Head of metabolic diseases. I'm here with Christiane Honisch, leading our EVOgnostics efforts. I'm here with Nele Schwarz, one of our top scientists in our stem cell & regenerative biology. I'm here with Paul Walker, a true expert when it comes to safety prediction. I'm here with Linda, one of the key leaders of our biotherapeutics efforts and who is leading now Just-Evotec Biologics. And with Steffen Grimm, who is leading our RNA efforts. On Page #5, you see the outlined agenda where basically 2 aspects, after I have updated you on our action plan 2025, are important. What are precision technologies to bring probabilities of success up and what are processes to bring novel targets via an integrated full suite of AI and machine learning tools into truly commercial manufacturing processes. And if you go forward one more page to Page #6, you should see that it has always been the mission of Evotec to bring the industry closer together. This is one of the key contributions that makes us the most efficient collaboration partner in the industry because we discover medicines in very difficult to treat diseases in a large network of partners. And this network of partners is increasingly data-driven to focus resources on processes where we can truly improve probabilities of success. And with this, we are building not only the platform for the sharing economy in research and development, but also, as one element of our business model, the largest royalty pool in the biotherapeutic industry. This is so critical for us because it gives us a very long-term view on our strategy and on our vision and our mission. If you go forward one more page, you see how we build our strategy over a very long time horizon. And it is great to see and great to say that Action Plan 2025 is in full swing and is even accelerated by our platforms that we are building in precision medicine. And with this, we can also confirm our targets that we have set ourselves for Action Plan 2025 by going above EUR 1 billion in revenues by the year 2025, which is a consequence of a double-digit growth of this company every year in the last 10 years. We do this on a highly profitable basis, and we do this with more investments in research and development and more capacity and expansion investments than ever before. And again, with the key goal not only to build the best platforms, but also to create the largest co-owned pool of assets under our royalty pool that we are building. When you look at the power that Evotec brings to bear, 5,000 employees by the year 2025 and more than 4,000 as we speak. And with more than 1,700 Ph.D.s on our platforms, we are probably the largest group with the highest density of scientists at -- on all platforms that you can imagine. And with this, when we talk about AI and machine learning, it's a company that really knows how this works in action, and that's what you will see in the next couple of minutes. When you go forward, you see that for us, integration of technologies into an innovation hub is key. This innovation hub now represents R&D efficiency platforms fully integrated along the full management chain up to commercial manufacturing. This also represents the right solution for the right modality for every drug target. And with this, we are today the go-to partner for biotech companies, for foundations, and of course, for strategic partners in the pharma world to create novel precision drugs. If you go one page forward, you see that once we have put our platforms in actions, the output is amazing. And here, just a few KPIs that illustrate to you that there are more learning curves happening on our platforms than anywhere else in industry these days. And when you are able to educate your algorithms with more learning curves than anyone else, you will be able to have the best algorithms available in the industry. And that's why, for example, highlighting our iPSC platform with more than 440 billion derived cells and data points here is creating data input into our algorithms, which is basically uncopied anywhere else in the world, or if you look at the power of our Just-Evotec Biologics platform, where so far, we have a 100% success rate in what we are building in our J.POD, that is, for us, the best starting point to really break into the market here in an [ unparalleled ] world. And of course, by having more than 130 co-owned assets under our belt, Evotec insight of this co-owned pipeline is just starting. Yes, it's also true that, when you go forward one more page, building our co-owned pipeline also sometimes sees setbacks. And I'm, of course, fully aware that all of you have seen the news that our P2X3 was handed back by Bayer to basically Evotec because they didn't want to pursue this asset on the risk-benefit analysis that they have made. We are, as we speak, in discussion with Bayer to take the asset back, and we'll then discuss of how to optimally bring this forward. But one step back, and if you look at the existing pipeline, you should see that there is no reason to stop the vision of creating this royalty pool. On the contrary, you should see that our precision technologies have shown us that we can accelerate the building of this royalty pool into the future. And if you look at the next 12 to 24 months, you should see here more than 15 data points coming out of our iceberg of already co-owned assets that will have significant value impact to the upside and that will bring our co-owned pipeline forward. And if you look at this page, on Page #11, you see that the opportunities on this iceberg of co-owned drugs is really fueled not only in one disease area but in multiple disease areas where we have excellent starting point. This ranges from neuroscience, via oncology, into metabolic diseases, via inflammation & immunology, and of course, also into the world of infectious diseases. That's one aspect of this iceberg. The other aspect of this iceberg is that, as you see here, every data point that is generated is educating all our platforms and makes all our platforms better across all indications. So if you translate this concept forward, you should see on Page #12 that our scenario play going forward into year 2025 is fully alive by building more than 170 co-owned assets by the year 2025, by having more than 130 co-owned assets in our pipeline already and by accelerating this strategy via more R&D investments as we speak and more partnerships that are about to come still this year and going forward to make the portfolio of partnerships even bigger and better. If you translate all of this, and this is now just our EVT Innovate segment, you see that we are running an innovation platform illustrated on Page #13 that ultimately, we'll not only have super input into algorithms and learning platforms but also is a great business. And this great business is not only illustrated by co-owned assets, is also illustrated by profitability coming through research payments and increasingly also -- via royalty rates where we are confirming that we still expect the first royalties to come into Evotec by the year 2025, but it's not a massive royalty pool that, for example, other assets could have brought. It will be smaller royalties to come, but nevertheless in '25, we expect this to start and then to continue going forward. This is the Innovate segment, and don't forget, this is copied on the other side with platforms that are built on our base business Execute side. And with this, we have created a unique business strategy, which we translate into a long-term plan illustrated on Page #14, which we call Action Plan 2025. This is a clear strategy in place where by achieving more than EUR 0.5 billion in sales in the year 2020, we are very confident that we will achieve more than EUR 575 million by the year '21, which was the midpoint of our guidance. And if you look here at the illustrated bars going forward, you'll see that we think that double-digit growth is possible into the future because the demand on our platforms looks excellent. And I'll say this because 2022 has started, despite some really big macroeconomic challenges, in a very nice and a very good way for Evotec, which you will see once we report our quarters in the future. So you will see that Action Plan 2025 will shift our revenue composition more into the biologics field, but all segments, as they're represented here, are growing very nicely. And if you go one page forward, let me again confirm our goals for Action Plan 2025, but now bring you into what we want to illustrate today to you when we say that we are the true company that cares about returns on investment when it comes to making more precise medicine. Every day, you see multiple sources of innovation popping up. You see multiple trends that are illustrated where science is illustrated to you. The beauty of a company like Evotec is that we can truly integrate these novel sources of innovation, this novel data pools into drug discovery and drug development pipelines and into drug discovery and drug development platforms. So we can sound out the world of science what truly makes sense to be integrated in industrial processes. These industrial processes are represented here with R&D efficiency platforms, precision medicine platforms, Just-Evotec Biologics and right modality drug designs. And on these platforms, we can integrate the data that we see and we can aggregate the data that we see, and then translate these data pools into drug discovery and more precision-driven medicine processes. If you go one page forward, it shows you the power of this because when you talk about data today, it is a highly fragmented industry where many players are collecting from early data to late-stage data but the integration into drug discovery and drug development processes is truly happening at companies like Evotec and where we are able to aggregate and integrate from many different sources as we speak. You will see that the consolidation of this industry will continue because data has to come together in consistent and high-quality controlled places. That's the only way how data makes sense, and you will see this illustrated by the example of Uwe and Christiane shortly. And if you go one page forward, why do we do all this? It really brings us back to the core of what we want to achieve. We want to have more accessible drugs or more patients available. And with this, we have to find ways to reverse the situation that costs for drug discovery and drug development are going up and that commercial returns are decreasing. The only way to do this is to bring probabilities of success up. And if you can change this equation only by 10%, 20% or 30% or 40%, you should see that the leverage is enormous. We talk -- because we talk about an industry that is spending more than $100 billion a year in research and development, and just imagine, if you are increasing probabilities of success here by 30% or 40% and can then deploy this to better medicines in the future. And with this, if you go one page forward, let me show you that we are illustrating to you now a few examples of how to make molecular patient data available for better patient stratification, how we show you that iPSC-driven disease processes are fully in action and how we make efficiency and safe predictions integrated into the drug discovery process. These are elements where we are using AI and machine learning within Evotec. And this comes on top of processes where we are already, as an industry, working, of course, with all tools in AI and ML, as you will see. It is so exciting that I couldn't tell you how excited I am that we are living in a world where more technologies will come to patients very soon, and that Evotec is one of the core companies to do this and accelerate this. And with this, let me hand over for the first presentation to our CSO, Cord Dohrmann, who brings you into the overall framework of how we bring probabilities of success out.

Good morning and good afternoon to everybody on the call. Today, we want to talk about probabilities of success, and talking about probabilities of success means that we also have to talk about failure rates. And although the industry failure rates are well known, we usually try to ignore them. Page 22 is a reminder of a few key statistics. About 95% of all started drug discovery projects fail before they even enter the clinic. Unfortunately, this 95% failure rate is most likely an underestimate as preclinic failure rates are not really systematically tracked and reported. During clinical development, still over 92% of projects fail on their way to market. This means that not even 1 out of 10 projects that enter clinical development makes it into the market. Another reason why this number is fairly sobering is the fact that it pretty much stayed the same over the last 10 years. Furthermore, even after drugs enter the market, they can still run into major challenges which are usually safety related. Actually, over 32% of drugs encounter safety issues after they have entered the market. The fact that drug development success rates have not improved is of particular concern in light of the fact that the cost per NME launch continues to rise. The cost per NME launch has now reached approximately $3 billion. For this reason alone, it is of great importance to develop platforms which will improve probabilities of success throughout the drug discovery process. Attrition rates during clinical development are of particular concern as drug candidates that have made it into the clinic have already undergone a very rigorous selection process and lots of money has been spent, bringing them to this point. For this reason, we want to have a closer look at current probabilities of success during clinical development, which are shown on Page 23. The left panel shows the probabilities of success during clinical development in 2021. What immediately jumps out is the fact that the likelihood for a compound to transition from Phase II to Phase III is particularly low. Actually, less than 30% of drug candidates make the transition into Phase III after successfully completing Phase I studies. This indicates that despite all efforts during preclinical development, trying to make sure that we have identified the right target and compound for the right indication, in the end, over 70% of all projects fail at this point of development. This also indicates that we predominantly fail when it comes to disease relevance. Ultimately, this also means that we are not doing a very good job in the preclinic to test targets and compounds for their disease relevance. Current tools to select and validate targets as well as in vitro and in vivo models of disease to test for disease relevance are not only suboptimal, but in many areas, wholly insufficient. Another concerning statistic is on the right panel, which compares overall clinical attrition rates in 2026 with attrition rates in 2021. Despite all efforts to reduce attrition rates, the overall trends indicate that clinical attrition rates are actually going up. It is quite clear that we need to do better. Page 24 summarizes Evotec's approach to improve probabilities of success. At Evotec, we have been building a precision medicine platform, which more effectively uses a patient-centric approach. It is very much focused on delivering drug candidates with superior efficacy and safety profiles with a particular focus on disease relevance. The first component of the precision medicine platform are our molecular patient databases. Molecular patient data is instrumental to come to better understanding of disease etiologies and its molecular mechanisms. Molecular patient data helps us to essentially redefine health and disease on a molecular basis. We have been building molecular patient databases that allow us to more clearly stratify patient populations according to molecular phenotypes rather than just symptoms alone. The second component of this platform are patient-derived disease models. Here, we are using iPSC technology to directly model disease in culture, which also allows us to directly profile compounds for disease relevance at the very early stages of the drug discovery process. And the third component are our PanOmics and PanHunter platforms, which allow us to comprehensively profile and quantify disease processes. Our proprietary PanOmics platform allows us to generate comprehensive omics data at unprecedented throughput and quality and whereas our PanHunter platform allows us to analyze these high dimensional omics data sets with support of AI and machine learning tools. These unique platforms enable Evotec to more diligently test and select targets and compounds in regards to their disease relevance, but also their safety profiles. This will significantly contribute to improve probabilities of success, in particular, during clinical development, but also in the market. Page 25 is a reminder that omics during drug discovery does not end with a genome sequence, but that the genome sequence is only the start. Implementing omics-driven drug discovery is still facing considerable challenges. A major challenge is that we are still relying too much on genome sequences. The reason for this is that it is more easily scalable than other omics technologies. However, the biological insights that can be extracted from genome sequences are limited as these only give us an idea about the predisposition to disease rather than the actual disease status. In order to measure disease status is as important to directly analyze the disease tissues. And here, it is really transcriptomics and proteomics which are of critical importance. The transcriptome and proteome of disease tissues tells us what is going on inside of disease cells and how to potentially intervene. Another major challenge is that transcriptome and proteome analysis are more difficult to scale than genome sequencing. High-throughput transcriptome and proteome platforms that deliver high-quality data and are affordable are not yet widely available in the industry. The third challenge is that transcriptome and proteome data is significantly more complex than genome sequence. With transcriptome and proteome analysis, we are not only quantifying the expression of thousands of genes, but we are doing this in the context of different cell types at different stages of disease or even drug exposures. So in order to bring omics into the mainstream of the drug discovery process, what is fundamentally needed, a high-performing and cost-effective omics data generation platforms. At Evotec, we have built unique and proprietary omics data generation platforms, specifically for transcriptome and proteome data generation. And these platforms are called ScreenSeq and ScreenPep, respectively. These platforms are not only highly versatile when it comes to what cells and tissue types can be analyzed, but even more importantly, these platforms operate at unprecedented scale and deliver unprecedented quality of data. I just want to mention a couple of metrics here. Our transcriptomics platform operates at high throughput in 384-well format and still quantifies up to 15,000 genes. This is a resolution that is usually only achieved by deep sequencing. Similarly, our proteomics platform is able to process about 40,000 samples annually, quantifying up to 10,000 proteins based on single-shop proteomics. So what makes Evotec's platforms unique is really their overall performance. They are unique in terms of the scale at which they generate highest quality data and they are doing this in a highly cost-effective manner. On Page 27, you can see that on the basis of our high-performance ScreenSeq and ScreenPep platforms, we are generating huge amounts of transcriptomics and proteomics data that need custom-built infrastructure to manage its data sets and analyze them. In order to be able to do this effectively, we have developed proprietary data analysis tool specifically geared to connecting and analyzing huge omics data sets from clinical and pharmacological data. This tool is called PanHunter and it is a unique user-friendly software tool that enables bench scientists and data scientists to independently analyze that data, while statistical and AI machine learning analysis tools are running in the background. Data analysis tools can easily crunch huge amounts of omics data that will be essential for the drug discovery process in the future. Still just as important are reference data sets or databases that are essential for proper analysis. For this reason, it is important to really master both cost-efficient, high-quality omics data generation and AI/machine learning supported data analysis. Together, these kind of platforms will improve the probabilities of success in the drug discovery process. The demand for omics data analysis tool is steadily increasing. Since 2012, we have been building PanHunter with an industry-leading team of bioinformatics experts, data scientists and software engineers. Since then, we have used PanHunter in many of our collaborators -- very -- collaborations very successfully. In the last 2 years, we have clearly seen an acceleration of the use of PanHunter, not only internally but also externally with many of our partners. This is clearly driven by the exponential growth of omics data generation and the need to be able to analyze these data sets as well as generally the fact that omics-driven drug discovery is generally on the rise in the industry. We believe that PanHunter is poised for exponential growth here in this field in the future. I already mentioned that Evotec employs AI/machine learning tools for the analysis of omics data. And this is, of course, not only true for omics data-driven analysis here, but really throughout the value chain. Page 29 shows that Evotec employs AI/machine learning in essentially all aspects of drug discovery and development. This begins with the profiling and -- of patient data, drug screening as well as general design of small molecules and antibodies, but also those in the preclinical environment and clinical development. Today, we have selected 2 areas where we want to discuss the use of AI/ machine learning-driven data analysis in a bit more detail. One is the analysis of patient data and the other area is safety prediction. These 2 areas have become essential for drug design that have a higher -- to design drugs that they have a higher likelihood to have superior efficacy and safety profiles in patients. This brings me to my last slide, Page 30, and the end of this introduction. In what follows, you will see that the drug discovery paradigm is changing. The way we use omics data will change the way we conduct drug discovery and generally change how we look at data and how we analyze it. The 3 pictures here show how a single transcriptomics readout is used in a high-throughput screen of a small molecule library. The data cannot only be used to make sure that we select the most disease-relevant compounds, which are shown in dark green on the panel on the left. But furthermore, the same transcriptomics data can be used to determine the mechanism of action of these compounds as well as it can be used for predicting potential safety issues. The seamless integration of omics data throughout the drug discovery process, in conjunction with AI/machine learning-driven data analysis, is changing the drug discovery paradigm and will become the most important driver to improve probabilities of success in our industry. With this, I'd like to thank you for your attention, and this is where I hand over to my colleagues, Uwe Andag and Christiane Honisch, who will present our efforts in building highest-quality molecular patient databases and how these can be influenced or how these will influence our drug screening efforts as well as patient selection in the clinic. The stage is yours, I guess, Uwe.

Yes. Thanks a lot, Cord. Thanks for the nice introduction. And good morning, good afternoon, everybody. We are now presenting you an overview of our E.MPD, Evotec's molecular patient database. This database is becoming more and more a key component, actually, of our daily research and drug discovery efforts at Evotec. E.MPD will redefine patient populations according to disease mechanisms rather than symptoms, which is, in our mind, the entry point for precision medicine. Using human data as a starting point for drug discovery programs will clearly improve our probability of success in the future. On the next slide, you can see that we have started to build up this database with an initial focus on kidney diseases. And since then, we are expanding the E.MPD by collaborating with high-quality patient cohorts in Evotec's key therapeutic areas that are listed here. Our multi-omics platform which is called EVOpanOmics as well as Evotec's proprietary software solution, EVOpanHunter, are central components that are required for generation of this very unique molecular patient database. Here, you can see some impressive numbers that give you an idea of the power of this database. By now, we have access to samples and data from more than 15,000 patients which results in a total amount of more than 200,000 samples and more than 200 billion data points that we have generated so far. And this is really constantly growing, as Cord said already. It's important to understand that Evotec is taking a holistic multi-omics approach rather than focusing solely on genetic analysis of patients. Cord also alluded to that. So this multi-omics approach is leading to detailed understanding of critical cellular mechanisms, whereas genomics analysis has a strong focus on underlying genetics only. In addition to the omics data that we are generating on patient samples, very rich clinical records of 50 to up to 500 different categories contribute to stratification of patients and proper understanding of the omics data that we're generating. Out of more than 30 different diseases, we have filtered a number of highly valuable targets and progressed those towards target validation and drug discovery already. This slide summarizes the buildup of the E.MPD in the kidney disease space where we actually have started with the NURTuRE consortium in the U.K. back in 2018, followed by another U.K. cohort, the Salford Kidney Study in 2019. With the addition of data and samples from the QUOD, Quality in Organ Donation, biobank in 2020, we have added a significant number of high-quality, healthy donor references to our database. Last year, we were able to add the German CKD cohort to our portfolio that almost doubled the number of kidney disease patients that are part of the E.MPD. This GCKD cohort is a longitudinal study over more than 10 years that now contributes significantly to our understanding of key events and pathways of kidney disease progression. As you might have seen, we have added another cohort in collaboration with the University of Bristol to the E.MPD yesterday. This is a well-defined international cohort with samples from Asia, India and Africa that is focused on the nephrotic syndrome. Now what are we actually doing with all this data and samples? Well, we are combining high-quality clinical data with multi-omics data that Evotec is generating again with its EVOpanOmics platforms. With our EVOpanHunter software solution, these very rich data are then QC-ed and analyzed, enabling the following value chains. So first of all, development of first-in-class therapies, our core business, as you all know, but also biomarker discovery and clinical trial support, these data are used for. And last, but not least, these data have the potential for next-generation patient stratification as Christiane will show in a few minutes. Altogether, this deep understanding of diseases is going to significantly improve the probability of success for all our programs. You might wonder why Evotec is taking such an effort generating all these data in-house since there must be, and there are, plenty of data available in the public domain. So this slide nicely summarizes the reason for taking this huge effort and the strength of E.MPD versus the use of publicly available data. Summarized are key parameters and steps that significantly contribute to the outcome of an analysis of human data sets. As you can see, E.MPD has superior quality for all of these parameters and steps. Engagement with clinicians prior and during a clinical trial is important for proper study planning and execution. And the clear understanding of the data as well as tight QC of the clinical data set is essential for proper patient stratification. Furthermore, high-quality data generation, preferably on one and not multiple platforms, will strongly improve the outcome of the analysis, which actually is the basis for efficient and successful target and biomarker identification. This slide nicely summarizes how limited understanding of data can severely affect the outcome of an analysis. The lower panel shows an analysis on real data that our scientists at Evotec have generated using the E.MPD. In this case, rich information from the clinical records help to stratify the data sets according to sex, and the QC of the omics data made sure that low-quality data sets were erased. The analysis showed a clear progression of one of the populations, the red curve, you can see here. In contrast, as you can see on the upper panel, doing the same analysis really on identical data sets, but lacking a few key information would lead to a completely different outcome. So lower quality input and lack of relevant information leads to a situation that makes it basically difficult to perform a proper comparison of populations and even raises the risk to run a totally wrong comparison. For example, male or female versus bad quality samples that should not even be part of the analysis. So this will consequently lead to a completely wrong interpretation of the data or a very high signal-to-noise ratio in the final data set, which makes it impossible to identify the relevant population, the spread curve amongst the others. So in summary, the use of public domain data sets can be very challenging and might lead you on a completely wrong track, whereas the E.MPD is going to improve probability of success, as you can see from this very nice example. So on the next slide, as mentioned before, an in-depth molecular profiling of patient samples is constantly feeding our E.MPD, and DNA analysis like [ Sniperase ] is used, for example, to identify polygenic risk scores of certain diseases, the transcriptomics data, deliver key targets and mechanisms that are driver or predictor of diseases on mRNA level. So again, one step closer to the final biology here with the mRNA. And then in addition, we are using our high throughput and state-of-the-art proteomics platforms to confirm and validate a certain hypothesis on the protein level, which is the most important piece in the end. All the omics data are then combined with clinical records using our EVOpanHunter software. Since all of this is fully integrated into Evotec's drug discovery infrastructure that has been introduced earlier today, Evotec is able to progress interesting findings from these data sets from the bedside to the bench and then actually back to the bedside. This slide summarizes the process for one of our internal drug discovery programs. As you can see on the top left figure, we have identified a strong link of a particular target to disease progression in patient samples. In this case, the target expression is negatively correlated to eGFR, so a measure of kidney function. In our in vivo models, we could then confirm that the correlation of target expression in kidneys to disease severity shown here for the ZSF1 rat model of diabetic kidney disease. Furthermore, knockout mice where the target gene has completely been deleted have improved kidney function in the CKD model as shown in figure 3. These very promising data sets prompted us to develop an inhibitor of this respective target that apparently is a key driver of the disease. The optimized molecule showed very nice in vivo efficacy in several models of kidney disease. And one example of such protective effect is shown in figure 4. The molecule is now on its way to the clinic. So we are basically closing the circle back to the bedside. I hope this is giving you an idea of the power of molecular patient database that enables us to start the whole drug discovery in human, which is a clear advantage over historic drug discovery attempts. Obviously, several other companies in the field are sharing the vision, too. E.MPD has attracted quite a number of pharma companies that entered into strategic collaboration with Evotec on this database. With the Bayer collaboration, we are using parts of the database for better understanding of translational aspects at late stages of our joint drug discovery programs. Our joint venture with Vifor called NephThera as well as Novo Nordisk, Chinook and very recently Lilly have joined our journey on drug discovery and development based on the E.MPD, too. This is not only generating immediate return of our investment, but in addition, it significantly increases the probability of success in this area since we are having multiple shots on goal together with our partners based on high-quality human data sets. As mentioned earlier, we are not only using the E.MPD to feed our drug discovery pipeline, but also to work on new concept in the patient certification area. And Christiane is going to tell you more about this. So over to you, Christiane.

Yes. Thank you, Uwe. Good morning, good afternoon to everyone. Beyond the use of the E.MPD to feed our drug discovery pipeline as just presented by Uwe, the combination of rich molecular data sets with the patient clinical data in our E.MPD puts us into the unique position to accelerate changes in the health care paradigm from the current disease management based on intervention to a health management focused on prevention before or at the onset of disease. Let's have a closer look into what this means to our patients. Today, as predicted -- as depicted, sorry, in the upper parts of Slide 42, the vast majority of patients see the doctor only when they have disease symptoms. Depending upon the severity and the clarity of the presentation of these symptoms, the doctor provides an initial diagnosis and initiates a set of consecutive diagnostic tests of noninvasive or even invasive nature. This usually leads to a one-size-fits-all treatment and reactive rather than a proactive intervention, which is especially for chronic diseases, resulting in today's sky-rocketing expenditures in health care. In the future, the rich molecular patient data sets we are generating today in our E.MPD are going to provide omics-driven personalized health and disease monitoring as shown in the lower section of the slide. During their regular health checkups, patients will then benefit from a health stratification based on these extensive data sets. For example, we have blood draw, blood transcriptomics screen and machine learning analysis. We will be able to place patients on a map of health and disease. This leads to a much earlier detection of disease predispositions and latent disease states and facilitate disease prevention, less invasive treatments and prolonged life expectancy at prevention health management that will come with a significant savings in health care costs. The foundation of the new paradigm of omics-driven personalized health and disease monitoring in the area of kidney disease is our world-leading, most comprehensive molecular profiling based on blood transcriptomics. Blood transcriptomics reflect key drivers and mechanisms of disease. A comparison of limited blood sample numbers in the public domain depicted here in gray in the graph, and the large blood sample numbers in our E.MPD with the NURTuRE, GCKD, longitudinal and healthy control cohorts show clearly how deep our insights and knowledge in the kidney disease space are going to be. In EVOgnostics, we have shown a new way to define patient populations and to place patients on a map of health and disease based on molecular disease profiles, especially broad transcriptomic data in combination with clinical parameters from our E.MPD, building EVOgnostics' artificial intelligence and machine learning models. As shown here on Slide 44, from left to right, we generate omics data, especially transcriptomic data, from the patient's blood using our EVOpanOmics platform, combine them with clinical data and analyze them via machine learning in our EVOpanHunter to predict patients' individual risk for disease and disease progression from low to medium, to high disease risk. We focus on early disease relevance to improve probability of success in personalized disease monitoring and show a tracking of these disease -- of disease progression. On Slide 45, we exemplify this EVOgnostic machine learning models in kidney disease and in our NURTuRE patient cohort. The NURTuRE cohort reflects the complexity of kidney disease etiologies on the left of the slide and the associated differential diagnostic challenge. Over 30% of chronic kidney disease etiologies remain undiagnosed and listed as unknown etiology and lack disease-specific treatment strategies. On the basis of transcriptomic data, clinical data and the EVOgnostic machine learning models, we start to be able to redefine these patient populations and specific disease etiologies according to disease mechanisms rather than disease symptoms. As shown in the middle of the slide, we are able to assign CKD risk scores to each patient for disease probability and progression monitoring and stratify patients into CKD risk books. The health to disease map on the right is composed of the blood transcriptomes of over 2,000 CKD patients. One blood transcriptome test per patient resulting in 1 data point within the blood. Patients stratifying nicely by disease progression from healthy or no disease as depicted in blue, to manifest the disease depicted in orange. With the EVOgnostic machine learning models in kidney disease, we have the unique opportunity to replace the diagnostic paradigm in chronic kidney disease of a multitude of individual tests with one molecular disease profile in one omics test. And we can, as shown on Slide 46, even get deeper disease insights down to the single blood cell level resolution for subpopulations or individual patients which require further stratification and monitoring. An example of the resolution of the patient's blood transcriptome down to the individual blood cell types is shown on this slide on the right. For certain kidney disease etiologies, genetic predisposition, which can be inherited, might add to the risk of developing the disease. We can read these genetic predispositions, as shown, for example, in the Slide 47 from our omics data, and combine this genetic predisposition with the blood transcriptomics CKD risk score shown in the previous slides to increase the accuracy of our patient health and disease maps and ultimately support diagnostics test precision to even further increase probability of success. The Evotec translational molecular patient database is central to our data-driven drug discovery pipeline, our patient population stratification, clinical trial design, precision diagnostics and precision medicine. The E.MPD expansion will accelerate multiple therapeutic areas for us such as kidney and liver disease, inflammation and fibrosis, oncology, neuro-inflammation and infectious diseases. We are and will add retrospective, prospective and longitudinal cohorts in all of these areas to further pave the way to true data-driven precision medicine and increase our probability of success in understanding and treating complex diseases. This brings me to the end of my presentation. Thank you very much for your attention. Let me now welcome Nele Schwarz and Cord Dohrmann to the stage. They are going to take you into Evotec's iPSC drug discovery.

Thank you, Christiane, and thank you, Uwe, for demonstrating how instrumental high-quality molecular patient databases are to create the future generation of precision medicines. In the following section, we want to talk about patient-derived disease models which are the best way to transfer this molecular disease knowledge and know-how into -- from the databases -- molecular patient databases, into the drug discovery process. Evotec has built an industry-leading iPSC-based drug discovery platform, which covers over 300 patient-derived cell lines and 15 different cell types. Currently, we are generating over 500 billion iPSC-derived cells per year. And these are used then to run over 20 discovery-stage programs where we screen over 1 million compounds annually. Most importantly, we have brought our first compound, EVT8683, which originated from an iPSC-based disease model into the clinic last year, and we expect more to come. As you can see on Page 52, iPSC-based drug discovery is changing the paradigm and that we are increasingly aligned on patient-derived disease models rather than traditional rodent models to assess disease relevance for targets as well as compounds. Although animal models will still be required, they won't be the primary selection criteria, especially when it comes to the question of disease relevance, nor will they play a role in selecting patient populations for clinical studies. At Evotec, we have been investing into iPSC-based drug discovery since 2020 -- 2012, initially, through a collaboration with Harvard University in the motor neuron space. Since then, our portfolio of cell types has constantly grown and now also includes cell types relevant for eye diseases. These cell types cover retinal pigment epithelial cells, photoreceptor cells as well as retinal organoids. Similar to other cell types, this offers the opportunity to take a new step at highly complex eye diseases such as age-related macular degeneration, and my colleague, Nele Schwarz, will tell you more about this in a couple of minutes. Before we get to this, though, I would like to briefly describe conceptually how iPSC-derived disease models enable the screening of compound libraries with a primary focus of selecting disease-relevant compounds that are reverting a disease phenotype back to the more healthy state. Page 54 shows a screen where we used iPSC-derived cells to model a disease process in culture, which we can then still use in a high-throughput screen. With an iPSC-derived cell-based disease model, which recapitulates the molecular disease phenotype, which was originally observed in patients, we can now screen compounds directly for their ability to revert this disease phenotype to a healthier state. Through the use of our high-throughput transcriptomics platform, we can assess the profile of individual compounds in a highly comprehensive manner. This means we can assess if the drug candidates we are screening fully revert the phenotype to the healthy state, if they only partially revert the phenotype or not at all or do something else that is detrimental. Modeling disease through human patient-derived disease models is the best way to ensure our drug discovery programs are really disease relevant. This will significantly contribute to improve probabilities of success, especially when it comes to testing drug candidates in Phase II clinical trials, where failure rates are currently the highest. Page 55 exemplifies that our iPSC-based drug discovery platform has already delivered very significant commercial value. One example is our collaboration with BMS in the field of neurodegenerative diseases. This collaboration has only been initiated at the end of 2016 and has delivered a broad pipeline of projects and milestone payments. The first molecule from this collaboration entered the clinic in the second half of last year, and we expect others to follow. 2021 was a particularly successful year for this collaboration, generating, in total, revenues of USD 75 million. And furthermore, we have already had a good start in 2022 with a further expansion of this collaboration based on another USD 15 million expansion payment. Page 56 shows that Evotec's iPSC platform keeps growing, not only in the field of small antibody drug discovery -- small molecular antibody drug discovery, but also in the field of cell therapy. We believe that the use of iPSC-based disease models for drug discovery purposes and cell therapeutic approaches is still at its very beginning and will become increasingly mainstream in the industry. Beyond neurodegeneration, in ophthalmology, there are many more areas or enough opportunity such as neuro-inflammation, neurodevelopmental disorders, chronic kidney diseases, as Uwe mentioned before, and the same counts also for iPSC-based cell therapy approaches for which there are a number of opportunities that are listed here as well. Having said this, today, we want to focus on the use of iPSC-based drug discovery in the context of ophthalmological indications. And this is where I hand over to my colleague, Nele Schwarz, who will take you through a project she pursued with her team for quite some time. More recently, this project attracted [ a big ] pharma partner, but she will tell you all about this. Nele, it's over to you.

Thank you, Cord. Yes. Following on from Cord's overview of our iPSC capabilities, I will go now into more detail on how we can utilize the iPSC platforms to drive drug discovery forwards in ophthalmology with a special focus on AMD. AMD, or age-related macular degeneration, is a very common retinal disease which affects approximately 200 million people globally. And due to an increase in the world population and the higher life expectancy, this number is set to rise to 288 million by the year 2040. However, the treatment options are currently severely limited. We distinguish between two types of AMD, dry and wet. The wet form describes the disease form where immature blood vessels grow into the retina, where they then tend to burst and cause retinal bleeding, which further aggregates vision loss. And here, the therapeutic strategies are intravitreal anti-VGF injections and laser therapy, for example, but these approaches have significant downsides. Intravitreal injections generally pose a risk of causing infections and inflammation in the eye as well as retinal detachment. In addition, the period of the effectiveness of these treatments is not long, so that this is not a long-term treatment option for patients. There are currently no treatments for the far more common form, the dry AMD. Therefore, we urgently need to identify suitable targets and therapy options to treat patients with AMD. And the reason why developing therapies for AMD has been so difficult is that this is a complex multifactorial disease. The main disease driver is age but there are also environmental factors such as smoking, and so far, unknown genetic links, for example, AMD accumulates through family history. The challenge to develop therapies for this complex disease is further confounded by a lack of predictive in vitro and in vivo models as well strong genetic heterogeneity of the patients and the absence of a clear genetic disease link. The majority of drugs in the clinic and in clinical trials currently target wet AMD but since the vast majority of AMD patients suffer from the dry form of AMD, this means that probably 180 million people currently are without any therapeutic option to slow this progressing eye disease. For dry AMD, though, there are currently 2 drugs in late-stage clinical development, which provides a great opportunity to identify targets and develop compounds for all -- for early, intermediate and late stages of AMD. And we believe that by using iPS cell-derived models, we have the edge here. As you know, iPSC models have become widely used models in academic and industry research as well as in drug discovery, primarily because they fulfill the so-called rule of three. This means that using human iPSC-derived cells provide a system that confers high relevance already at the screening stage. Patient-derived or CRISPR edited cells also provide a cell intrinsic, non-artificial disease trigger, which will be more relevant to the respective disease that one is studying. And finally, using the system, one can achieve a translational readout with a functional manifestation of the disease, which also enables screening of multiple different donors and potentially subsequent patient stratification for later clinical trials. And what we have done at Evotec is to take advantage of iPS cell models and combine them with our highly innovative platforms, some of which you've already heard about such as our PanOmics platform and some of which you will hear about in the speakers that come after me. And using the synergy of these platforms in combination with our iPSC modeling expertise enables us to raise iPSC disease modeling and screening to a new level that provides a powerful tool to discover new therapeutic targets. And this is precisely the strategy we will take to develop therapies for AMD. So we will use our iPSC-derived retinal platform and an unbiased phenotypic screening approach to identify new targets that drive processes towards retina protection. This system provides high translatability into the clinic as we are already working with the target cells for AMD and we also have established this in a highly scalable and robust assay format to enable high-throughput screening. And using a highly diverse compound library, coupled with AI and PanOmics, enables therefore the discovery and deconvolution of novel disease relevant targets to deliver ultimately first-in-class therapeutic options to patients. And as Cord has already alluded to, we have recently entered into a collaboration with Boehringer Ingelheim where we will use the combined power of the iPSC and PanOmics platforms to identify and validate new therapeutic approaches in ophthalmology. This collaboration comes with a significant upfront payment as well as an FTE-based research payment with milestones and royalties. And of course, following the successful setup of the iPSC ophthalmology platform, we don't stop here. We will further expand our capabilities in the coming months. For example, our sustained focus on AMD will lead to an expansion of our in-house iPS patient cell line bank as well as the development of more complex disease-relevant assays such as drusen formation, for example. We will also strive to integrate available patient data sets into PanHunter to dig deeper into disease pathways and identify potential biomarkers. In addition, we have also successfully generated our first retinal organoids, which show a fully stratified retinal structure with all the relevant cell types, including photoreceptors. And these models will be used to study diverse ophthalmic diseases, for example, genetic retinopathies, which mainly affect the photoreceptors. But these models also open the door for cell therapy approaches, for example, in the area of photoreceptor transplantation. And this concludes the iPSC session. After a short video clip, I will hand over to my esteemed colleague, Paul Walker, who will highlight our capabilities using AI and machine learning to predict drug-induced liver injury. Thank you. [Presentation]

Thank you, Nele. Today, I would like to walk you through an example of using AI/ML in the prediction of drug-induced liver injury. Next slide, please. We are building the largest transcriptomic safety database in the world made up of a known drug-induced liver injury compounds and marketed compounds. The database is also being expanded across many high-risk organs such as cardiac and kidney. This project combines organotypic in vitro organ models using Evotec's iPSC capabilities and our high-throughput transcriptomics. Further to this, we are able to analyze greater than 15 organs from preclinical safety studies using the same platform. This information is combined using AI to better predict safety issues, helping prevent costly late-stage failures and help lead our programs away from target and off-target mechanisms of toxicity. Next slide, please. As you can see in the pie chart on the left, drug-induced liver injury is a leading cause of drug attrition in both preclinical and clinical studies. Further to this, 18% of drug withdrawals from the market are caused by DILI. In this example, we show how by combining our safety database with AI allows us to proactively derisk DILI in early drug discovery. Next slide. We continue to build our DILI database using known DILI compounds from the FDA Liver Toxicity Knowledge Base and marketed compounds, plus toxicology mechanistic information from additional tool chemicals. The overall aim being to enhance our accuracy of DILI prediction and to understand the mechanisms of action to aid risk assessment and lead our chemistry away from the safety liabilities. We take extremely complex high-throughput transcriptomic data, covering tens of thousands of differentially expressed genes and hundreds of toxicity mechanistic pathways. Combining with AI, we then -- we have then turned this information into a visual risk assessment tool where we can look at dose response information against the maximal concentration observed in humans. This allows us to understand the DILI risk of novel compounds in development. As you can see on the right, this has been assessed against multiple in vitro liver models. This technology has only been available since 2020. Therefore, the probability of future success will go up with this technology. Here, we show the dramatic improvement we see by combining our safety database and AI. This is the current gold standard image-based approach in primary human hepatocytes. Using a set of 128 DILI reference compounds gave an overall accuracy of 70%. Taking exactly the same compound, human-relevant liver cell models, high-throughput transcriptomics combined with AI modeling derived a vastly improved accuracy of 86%, thus giving us confidence in using this data in early safety assessment. We are continuing to build on this success by expanding our safety database, combining human relevant organ model to tackle safety issues across different organ types, in addition to an enhanced mechanistic insight in preclinical species studies. With this, you have seen holistic improvement in our safety assessment capabilities using AI/ML and high-throughput omics technologies. Thank you for listening, and I would like to hand over to Volker.

Thank you, Paul. And we will have the chance now for a brief Q&A session. We will be having a much longer at the end of the event, but we thought of dealing with the first questions already now. And I think it's good to start off with a few strategic questions. Many of you spotted that we anticipate still a small portion of 2025 revenues coming from royalties and are asking for the thinking behind that. And I think that's a question for Werner.

Yes. Of course, we have, at this stage, smaller products in regional efforts going on. For example, our GABA inhibitor is in China Phase III trials, which we expect to translate into the market in the coming years. And also, don't forget that we will be able to charge on certain technologies, for example, linked to our iPSC platform, royalties for using these platforms, which will also be permanent income. So with this, again, it will be starting slower than initially expected in 2025, but also, don't forget, there was never an expectation of any royalty rates in our projection to go beyond EUR 1 billion in our Action Plan 2025. And we still then see afterwards the full pipeline delivering its power in the years '25, '26, '27 to come when it comes to marketed products and royalties, and that's why we are confirming not only the action plan, but also division of royalties.

Thank you. And also, related to the Action Plan 2025, question related to the trajectory of revenues and EBITDA goals. Can they be achieved organically? Or does it include necessary or possible takeovers?

You will see today, and have already seen before the break, that we have never been so thankful and powerfully equipped with our platforms and technologies as we are today. So the best investments that we can make is expand our precision medicine platforms, our human molecular patient databases, our iPSC platforms, our proteomics platforms. Our platforms that are really driving what you will then see in the future AI/ML integrated drug discovery or, for example, what you will see in a few minutes, our RNA platform. So from an organic perspective, yes, Action Plan 2025 is absolutely possible because the platforms are there, and we will expand some of these platforms only with, I would say, smaller rounding up additions and M&As where necessary. But of course, and that's, as you know, something that I have to say, we will never exclude any opportunity where we can create shareholder value. And if this is an M&A, then, of course, we'll also do it. Having said that, we feel that with more than EUR 800 million cash in the bank and a highly profitable business, the company is in a strong situation to evaluate these opportunities in all dimensions.

Thank you. And then we have two questions related to royalties. One directly linked to the Boehringer deal and a comment here on royalties to -- related to iPSC assets. If we can elaborate a bit deeper here. And the second question, more strategically, also linked to the overall theme of today. If we were able to show that probabilities of success go up, would that change the structure of royalties going forward? [indiscernible], I guess.

The power of our platforms, I think it can be best illustrated by the fact that we are able to achieve, in this very early stage of the drug discovery process, already significant royalty rates. And for example, if you know that standard royalty rates for many assets that are partnered in Phase II stage are still only at small double-digit rates beyond 10%. And when we, for example, in our partnership on protein homeostasis with BMS are able to generate double-digit royalty rates already on the platform, that shows you how much our partners value the intellectual input, but also the power of these technologies. And that's why, yes, royalty rates are, for us, an important indicator for the power of innovation, and we are able to generate high single digits, sometimes double-digit royalty rates depending on how we balance upfront, how we balance research payments, how we balance then marketed royalty rates on size of markets and so on. So there's an individual escalation going on there. But in general, all the precision medicine platforms are equipped with royalty-bearing assets, and that's fantastic for us. When it comes to -- if we are -- it's not a question, if we are improving probability of success. What you are seeing today is making you witnesses of a paradigm shift that we are shifting probabilities of success. And that's again why we are able to ask for royalty rates at this early stage of the drug discovery partnerships already. Otherwise, it would clearly not be possible. And I think, yes, it needs some more proof points. And yes, it needs also a bit of experience with many partners to do that. But I will tell you, once you have done a PanHunter via transcriptomic process, you will never want to live without it. Once you have done an iPSC model and you have seen patient relevance early on, you will never want to work without it again. And that's I think why this will be very sticky processes where you're not only increasing probabilities of success, but where you're also making processes much more elegantly, networked together with our partners.

And another one on commercials of E.MPD. In the light of extensive clinical data sampling and unique insights derived, how's Evotec monetizing its work to develop novel diagnostics? Maybe that's one for Christiane and Werner can chip in afterwards.

Yes. So the path forward to leverage, this is leveraging our disease insights, which we get from the molecular patient database, and to turn this into biomarkers -- into biomarker panels, ultimately omics test in partnerships and provide content-driven improvements to diagnostics and to diagnostics service providers and manufacturers. So supports the diagnostic entities in the markets with our disease insights, which we get from our molecular patient databases.

It comes down to the convergence of what you're seeing here that the split of pure-play diagnostics companies to pure-play therapeutic companies will go away once you understand the power of molecular patient-derived data. And that's why Evotec doesn't intend to become a diagnostics company. Don't misunderstand that. Evotec will be able to understand, and with this, empower people to work with molecular patient data in certain disease areas. And that's a very easy way then to start commercialization. And if you look at what Uwe has presented to you, the series of only kidney-based disease deals that we have made in the last 5 years shows you that basically our partners understand here, it's more elegant and efficient for them to access our molecular patient databases and pay us research fees, pay us milestones and pay us royalties than to build this on their own because what is true waste in the industry that everyone does this in a fragmented way alone. And that's why I was hinting to this in industry, data only makes sense if it is curated, high quality, and it is in an unbiased and in a large pot, available for many and then shared.

Thank you. And I think there's room for one more for Cord, in that case, dealing with a better understanding of what we saw in the past in the industry. Nearly every therapeutic area that was shown on Page 23, I guess, shows decline in clinical success rates over the last 5 years. What is the driving force behind this decline?

Well, I think, unfortunately, the drivers are multifold, and it's, of course, a major driver, as I alluded in my presentation, is the fact that we are still relying on really inadequate disease models largely. But on top of that, there are, of course, the competition is steadily increasing, standards of cares are going up all over the place and regulatory requirements are increasing. So I think that every drug discovery project that you start nowadays is based -- is in a much higher competition and exposed to many more risks than it used to be. And the fact that once again that we are still relying on disease models that are essentially in many areas, ancient and have proven to be not very predictive or helpful in developing drugs is, of course, also detrimental. And this is where we see the major negative impact once again in the clinic, especially when it comes to Phase I studies where we test compounds or drug candidates for the first time for their efficacy in patient populations. This is really where we fall down. And I think this is one of the areas that the industry needs to focus on, and we are trying to do that within our own efforts here with our physician medicine platforms and approaches as we outlined.

Thank you, Cord. So we're now switching gears. After the next short clip, Steffen Grimm will lead us to -- through the topic of RNA biology and we can continue with processes, bringing probabilities of success up with -- starting with that session. [Presentation]

Good morning and good afternoon to the call. My name is Steffen. I'm the Group Leader for the RNA biology here at Evotec. I have an exciting opportunity to build up the RNA targeting platform here, which is really applying a core strength of Evotec to a new biology. If we move to the next slide, this platform that we build is clearly of commercial relevance. It enables the field of targeting RNA biology here at Evotec. And the field really of RNA biology targeting was somewhat neglected over the past 10, even maybe 20 years ago. But there is an increase in high interest in the field to this RNA biology. And that's, of course, to part due to the very successful examples of mRNA therapeutics, but that's not all of it. In addition, there are some pillars, and really the high interest of big pharma as well into RNA targeting platforms that is showcased here by our Takeda collaboration that we have already entered, but we're clearly convinced that there is more to come. Looking at some indicators of our platform, we have quite a range of projects that have a significant resource needs even at Evotec. So you can almost think that we are a small company within inside Evotec that are dealing with getting really a new biology to rule on novel medicines. And we already employ a significant part of integrated technologies, and we are expanding further in this field, and we have gathered a unique experience through wide range of different RNA biology that we have already tackled, and we are looking forward and are very well positioned to future growth in that area. Looking really into the targeted genome, starting from the DNA level going down to really the drug targeted proteins. You can see that almost 75%, 80% of the DNA is copied to RNA, but only small part ends up in proteins. And even a smaller part of that is finally drugged on the protein level. So that makes up 2 baskets, if you will, of RNA targets that we can tackle. One is really proteins that are hard to track on the protein level, which potentially can be tracked on the RNA level, and the second basket is really non-coding RNAs. I'm not saying that all of the 75%, which are copied from DNA to RNA are drug targets, but looking into one example as the large non-coding RNAs, say, 8, 9 years ago, there was a handful of functionally described large non-coding RNAs. Today, there's more than 100, and the community is looking into this field continuously, so this target space is ever expanding even from just the academic contributions into that field alone. At Evotec, we do have a holistic view also to the RNA targeting here. We do have Antisense or siRNA technologies that really at the core hybridize with the aim to down or knock down protein targets, but we also have a fully developed platform on small molecules, which, in that sense, not only can downregulate the protein, but they can tackle modulation, degradation, translation or transcriptional efficiency and other means and are really maybe the more versatile tool at this level for targeting RNA. The core strength of Evotec, as I mentioned, is the development of small molecules targeting to the RNA targeting platform. And it offers a way to have a core strength developed to novel disease mechanisms. With really the success also on small molecule targeting RNA, for instance, the [Indiscernible] FDA approval and other literature examples, we clearly can show also in the literature that chemical beauty compounds can bind highly selectively and with high affinity to RNA structures. And we, at Evotec, have our platforms built up as toolboxes. We are combining technologies and assays in a toolbox that allows us to not rely really on one workhorse assay, which limits yourself into what you can actually target on the RNA level, but we're combining different technologies in toolboxes so we can have a wider array of targets to address to. This is shown here on the slide where we have toolboxes for getting the structure or validating the function of RNA elements and can pick the right species that we then put forward into the drug discovery process. This is also, to some degree, an overarching theme for the RNA biology. We rely on the unique combination of technologies, really. We have naturally grown experience on RNA biology from RNA production to characterizing RNA species to its interactions with compounds, and we connect these all with artificial intelligence and machine learning that connects really and allows us to build a world-class database that will, in the end, connect RNA structure to its function to the chemical space that can inhibit regulatory rules and show a therapeutic effect. This unique position, which we really have for the small molecule targeting platform, offers us to really offer a full spectrum of technologies along the whole value chain of drug discovery from the start of RNA production to target validations, into screening, into IND capacities. Our workflows, which are already streamlined from anyway to Hit ID to IND, we can deliver excellence and high-quality data for our clients and all the way along the value chain. And looking at this really in action, looking at affinity, looking at structure, looking at therapeutic effects, what we can show here from our programs that we have that we do show our compounds that we find have affinity and selectivity for RNA. They do change the structure of the RNA confirmational space. And they do show therapeutic effect in a cellular environment, showing really the therapeutic effect that we are looking for. We are already excelling here in different areas on RNA targeting. Some of our Innovate projects, clearly a different biology -- or targeting different biology, we were able to move the show the cellular target engagement as seen on the previous slide. We can classify 2-tier structure activity relationship in these programs, but also pushing our major collaborations to enter the Hit ID phase here for the small molecule programs for our ASO, siRNA programs we even have some of our collaborations also at the pre-IND or IND phase forward. But of course, we don't want to stop here. We have a clear path forward in the maturation of the RNA platforms, looking really to enlarge our product pipeline, broadening our target areas, clearly, for instance, such as novel approaches where we combine bifunctional molecules to RNA degraders, where we on the bifunctional level have experience on the protein level. We want to bring that experience from the protein level to the RNA level, and yes, create a novelty to this bifunctional molecules as RNA degraders, but we're also learning from the functional elements that we are targeting on the RNA level when -- can fuse the small molecule world to the mRNA therapeutics by introducing switchable mRNA therapeutics and really fusing these 2 worlds, again using a core strength of Evotec and apply it to a new novel biology. With that, I'd like to thank you for your attention. And after the short film sequence, I will hand over to Craig. [Presentation]

Good morning, good afternoon, everyone. It's really a pleasure to join you here from Seattle, where I'm spending the week with an inspiring Just-Evotec Biologics team. In this section, we'll now turn our attention to the application of cutting-edge technologies and exploitation of data science to the process of medicines discovery and development, and particularly in small molecules and biologics for probability of success. I firmly believe there is a sweet spot of performance in drug discovery and development, which lies at the intersection, the combination of high-quality data generation, the ability to extract predictive power from that data with the AI/ML tools, and there's still very human elements of track record, deep knowledge and expertise. Evotec has a long established track record of success in its integrated drug discovery and development in partnerships. And I believe they are well known and well respected as one of the leaders in this field, heavily driven by our industry seasoned intradisciplinary experts on all our teams. But today, I'd like to introduce you to E.INVENT-AI, our internally developed suite of AI, ML and experimental cutting-edge molecular design tools, which in the hands of these industry experts, allows us to really hit the sweet spot of drug discovery performance over and over again. On the next page, we have some very high impact numbers, which reflect our position in this really exciting field. Evotec's own Alex Heifetz edited the recently published book on AI in Drug Design, which was coauthored by many Evotec scientists, along with other leading contributors across the industry. With 20,000 downloads in the first 2 months or after publication, this is clearly a hot area for practitioners in the field and well beyond. Our E.INVENT design suite is widely used daily by community of molecular architects and drug designers to generate new insights, predictive models and drive projects faster than ever, typically 30% to 40% faster and more efficiently than benchmarks. This results in a high output of inventive steps, as illustrated with 62 new patents filed with Evotec named inventors in the past 2 years. That's roughly a pattern every couple of weeks on average. On the next page, we can all appreciate, I think, that having a high-performance design engine like E.INVENT is what you might call necessary but not sufficient for successful drug discovery and development. So we've nested E.INVENT into our well-established highly integrated drug discovery and development infrastructure where we can combine these cutting-edge tools with expert decision-making, drug hunting know-how and disease area insights and improve translational power that you've already heard about from my colleagues earlier in the call. This is really what differentiates successful and high-quality drug discovery and development from simply cool technology, and I've tried to lay this out on the next page. Many of the tech companies in this field do absolutely have good tools, but most like really high-capacity and high-quality data generation capabilities such as described by Uwe earlier, and therefore, rely on generally published sources to drive their engines. In addition, in-house capabilities for cutting-edge experimental data to support molecular design such as 4D NMR, structural biology, Cryo EM and coupled with computational power to exploit it are rarely found altogether. This becomes even more marked as the asset requirements and the predictive tools reflect criteria further along the drug discovery process such as ADME, developability, predicted human dose, physical form and indeed predictive DILI that you've heard out from Paul. With all of these capabilities and capacities available to our drug hunting teams at Evotec, we feel we have a marked advantage in real drug discovery and development. But the design engine, which generates high-quality and inventive steps, is absolutely important. So on Page 90, we've summarized some of the key features of the E.INVENT-AI design suite. Now to succeed in this field, you need really quite high-quality generative tools, which use AI tools like SLERP, along cellular molecular generators like reaction vectors. And we've also taken the step to quote decades of Evotec's expert medicinal chemistry experience into well-established and productive transforms. These generate in silico ideas, they need to be filtered through multidisciplinary predictive models for activity, physical chemistry, admin safety and so on. Especially early on, it's important to design molecules that are specifically conceived to inform better models of prediction. For this, we use basin optimization tools. And last, but crucially very, very powerful, experimental data like protein structure insight and small molecule for the confirmation gives us the experimental data into how the drug can be energetically locked for optimal activity and the lead drug disposition. Putting this suite in the hands of experienced molecular architects and drug hunters drives advanced and efficient problem solving and inventive state. So how does this work in a project context? On Page 91, we have a schematic representation of this. So ideas for targets and therapeutic concepts can come from, of course, a wide variety of sources. And exploitation of combined data can give rise to specific and novel insights for targets or mechanisms. From these starting points and using a combination of richly populated global models and precise local models as appropriate to the problem, rapid iterative design cycles allow improved prediction and determination of as many integrated readouts as possible, which, in turn, delivers improved drug profiles with a minimum time and resources and with the highest probability of downstream success. So to bring this to life for you, I'd like to rapidly describe our recent project as an example. This project started with a very simple query from a partner. We had a small set of molecules, around 30, some of which were really very potent, and they wanted to know if they had a preclinical candidate in hand. Application of predictive tools, such as early-dose demand and generation of precise data points to inform the project, rapidly highlighted that the series in itself was not candidate quality due to safety observations and also our predicted very high human dose. By building a data set Evotec around these first few molecules, we then apply basin optimization and found a very sparsely populated area of the data, but that which breakthrough into solving the potential safety liability by reducing CNS exposure just within a few months. And on the next slide, the full project story is laid out. The red data points on the left represent the first few molecules brought to the table by the partner. The orange data points represent the basin inspired insight into a more attractive subset of chemical space. And then application of new E.INVENT, confirmational insight and downstream predictive tools led to further advancing the profile in green, and then with final polishing and prediction of dose demand, allow the identification of compound 5 with a very attractive profile of very high potency, but differential concentration in the target organ compared to CNS and a very low predicted dose, which in fact, statistically enriches long-term safety and probability of success in itself. All of this was done in less in -- around 350 compounds and less than 12 months of effort. This is really the power of the current state-of-the-art and fully integrated drug discovery and development. I'm really proud to be doing this caliber of work every day at Evotec. But even now, I think it's just at the beginning of what's possible in the future. So now, I'm handing over to Linda, you'll hear that we have a correspondingly powerful AI-enabled suite for the discovery, development and manufacture of biologics. Linda?

Hope you can all hear me. Nice to meet you all. I'm Dr. Linda Zuckerman. I'm the Executive Vice President of Just-Evotec Biologics and the Global Head of Bio Therapeutics. Just-Evotec is based in Seattle, Washington, and I'm happy to be here this early AM to share with you our vision on how we have been executing over the past year. Just was founded in 2014 by a highly experienced team, expanding the field of protein process and manufacturing sciences. And they helped complete that mission that started back in the '80s with the approval of the first bio therapeutics, human insulin and OKT3. Since then, the founders and our staff today have been trying to solve the scientific and technical hurdles that have been blocking access to life-changing protein therapeutics, hence the term EVOaccess. We are accomplishing this task using AI, ML and NLP-driven technologies for protein therapeutic discovery, design and combining that with state-of-the-art manufacturing technologies. Last year, we spoke about our plans in the future. And today, we want to share with you how we are successfully executing on those plans. So if you can please go to Slide 96, I wanted to share with you a few of our key indicators of our success. These numbers span across our capabilities from discovery with the anticipated diversity of our J.HAL library all the way through our CGMP manufacturing, where the success rate has been extremely high. I'll point out that our -- you'll hear more about our impressive process and product design platform throughout the top, and it is perfusion based on continuous manufacturing and results in extremely high-mass demands. As you can see from just a few runs at a very relatively small scale, we can generate tons and tons of kilograms of antibodies as needed. If you go to the next slide, at Just-Evotec, our promise is broken down on the traditional silos of biotherapeutic product development, again using a foundation of AI, ML and NLP combined with powerful data capture tools. Thus, the data is shared back and forth across the platform and the continuum, allowing optimum product development and performance, both for ourselves and for our clients. And you'll hear an example of that in the case study today. Our J.HAL discovery tool uses generative adversarial networking, or GAN, to buy a highly converse stage display library that's biased for developability and other desirable properties. Our new PHH offering created from our HAL library is a much better alternative to traditional bispecific and multi-target approaches to treating diseases. Our J.MD, or molecular design platform, is a state-of-the-art AI, ML, NLP-based and silica tool, but as you'll see, powers very strong down selection of multiple candidates as well as improved molecular design of antibody and antibody-like projects. When coupled with physical and mass spec analysis, this results in the identification of the most optimal drug therapeutic candidate and can ameliorate key residues that can cause aggregation, instability, poor PK, immunogenicity and the like. Our JP3, or cell line development capability, focuses both on upstream and downstream processes, yielding extraordinarily high mass output. This is achieved through intensification and highly optimized downstream processes. Our platform can also increase product quality, providing the best fit for both traditional mAbs and more complicated molecules like bispecifics, fusion proteins and multispecific, which is where we see that the industry is headed today. Our J.POD approach to perfusion based continuous manufacturing leads the field in state-of-the-art. It is not uncommon for us to achieve 6 to 10 kilos per 500-liter run. This results in a highly cost-effective approach for manufacturing and is in line with our mission of making biotherapeutics accessible to all. Taken together, we'll achieve our mission of global access for biotherapeutics. Our clients, importantly, can engage us at any point in the continuum, and we've had some strong partnerships listed on the right-hand side of the slide over the past year. So just to get into a little bit more detail on the next slide, we offer a highly competitive offering in the drug discovery field. We have a license to the ally humanized transgenic mice. And for superior speed and developability and flexibility, we also have our own offering of the J.HAL stage display library, as I discussed earlier. Clients can also give us sequences and have us design and improve them, removing the manufacturing liabilities. And we did this quite a bit during the pandemic. We did a lot of down selection of antibodies against COVID using our NLP round-up tool and our Abacus tool. And again, this provides clients with the very best shot of bringing a well-behaved molecule into manufacturing, both reducing technical risk, increasing the probability of success and creating a high yield. Our JP3 cell line focuses on both upstream and downstream, yielding extraordinarily high mass output. The continuous approach reduces all times, as you'll see later in the top and increases product quality, making this the best fit for all types of therapeutic modalities that are antibody-like. We can also take clients materials into our own platform using their partnered cell line, and we have our own proprietary cell line, which you call CL72. And of course, as you've heard over the years, we've just opened in the past year our J.POD manufacturing suite out in Redmond, Washington and completed CQV runs and our online running client projects. And we're very excited to be offering this to the industry today. If you can go to the next slide. We know that this is a very competitive field. We want to take a few minutes to see how we stack up, which is quite well. We have a leading monoclonal antibody discovery platform that builds in CMC readiness, which is a highly unique offering in the field. We could also provide an under one roof solution from discovery all the way through IND. And you'll be hearing more about these types of programs over the next few years. Our perfusion-based approach allows us to edge ahead of the competition for carbon reduction and the best quality molecules. This is because our platform, although renew relatively in the industry is leading the field and highly mature. Now as I mentioned, my predecessor, Jim Thomas, last year, who is now retired, and this is for the last time according to his wife, spoke a lot about what we were building and where we are headed. And today, I'm going to share with you for the remainder of my talk a case study that not only illustrates the power of approach, but how we are truly executing on that promise. So if you can go to the next slide. In this example, we enabled a client with a panel of over 200 antibodies against an infectious agent. The antibodies have been cloned from convalescent patient sera, and many have similar binding properties to their target and suitable efficacy. So how do you choose? Well, this client turn to us to evaluate this very large panel using our AI, NLP, Abacus and RANDY app tools. We took the panel and down selected very quickly to the top 4 and then combine that with our powerful analytical suite of -- that's based on multi-attribute mass spec and a suite of biophysical analysis and then recommended the best 2, which we could take into our JP3 cell and cell line development platform. If you go to the next slide, I will highlight that. In parallel, we took these 2 research cell banks in, and we were off to the races. We used our high-throughput tools and real-time analytical suite to develop a highly intensified upstream process for both antibodies. Using our [Indiscernible] approach, we were quickly able to get to a clone with a single round of cloning, and we also leverage our small -- our scale down models of perfusion to quickly assess how the cells would behave at scale, thus allowing us to move directly into GMP in less than 5 months. If you go to the next slide, I'll show you some data from that. On the left-hand panel, we have viable cell density measured in millions of cells per ml on the Y-axis. And on the X-axis, we have days of culture. All of these data are from our 3 leaders scale-down models. There are a few points I'd like to make. Traditional fed batch holds the cells at a much lower density as you do not continually feed them. They extend the media throughout the culture day and thus make out a fixed -- make a fixed amount of antibodies. By comparison, we use perfusion-based technology, which allows the cells to be fed every day while we simultaneously bleed off the antibody drug product. This allows for the cells to remain happy, viable, and we got much better quality because the antibodies are not sitting around in the proteases of extended media and dead cells during the culture time. What you also see from this graph is that as the cells grow and we continue to leave them, we can tune the culture day from 10, 12, 14, and as you'll see later, much longer. This allows you to very quickly increase your yield as you see in the panel on the right-hand side. Where you have a fixed amount for fed batch; at 12 days, you get much, much more. And then at 15 days, you can see by just increasing the culture day by only a few days, you can increase the yield by several kilograms, thus underscoring the power of this approach. To date, we have made about 18 kilograms from these 2 antibodies across just a few runs, and we plan on extending the culture day to get more and more as needed. We've run these antibodies across both of our manufacturing sites in Seattle at our J plant facilities, and more recently, in our Redmond facility and found them to be comparable. And this is an important point because our vision, as you'll see later, is to have a clonal network of J clubs, and it's important that we can maintain the quality across all of that for our clients. So all in all, a highly successful demonstration of our capabilities. So where do we go from here? I've shown you a 15-day culture -- cell culture duration in what we call our hybrid platform. However, we are actually aiming for 25 days or more in a fully end-to-end platform, which means we automate all of the steps, and we are on track with that vision. If you go to the next slide, you'll see here that we have just completed our first fully end-to-end continuous process for a late-stage product that had greater than 25 days of production. And as you can see that the yields can be very high in our hybrid process today, but as we scale up into our 1,000-liter bioreactor run fully end-to-end and extend the culture beyond 25 days, we are quickly able to make over 35 to 40 kilos from just one run, thereby showing the power of our technology. If you go to the next slide. As I mentioned, the data that I shared with you was from both our sites, J.Plant and J.POD here in Redmond, Washington. I'm happy to share with you as you probably have heard in some of the earlier communications that we are going to begin groundbreaking of our second J.POD site in Toulouse, France. Thus, we are truly initiating our global network. If you go on to the next slide. This shows that our robust growth over the past years has been fueled by our successful execution, delivering on our promises, and we are getting meaningful traction in the marketplace as a therapeutic antibody of choice. COVID-19 continues to be both a challenge and an opportunity. And to keep our trajectory, we will need to continue to build out our Redmond facility for maximum efficiency as well as execute on our plans for building J.POD in Toulouse, France. We hope at this facility, and we are on track to be operational in 2024. And last point is that we are also in the last stages of completing our commercial readiness, so that we can become a fully functional from first-in-human all the way through commercial site and a meaningful difference to clients in the marketplace. I'm very excited to have been able to share this story with you today. And I'd like to hand it back over to show you a short film sequence before Werner provides some closing remarks. Thank you for your time. [Presentation]

Thank you so much, Linda. It's about the power of technologies to bring probabilities of successor, but it's also about the power of people to bring probabilities of successor. With this, in the last 120 minutes, you have seen how exciting it can be to truly disrupt by bringing probabilities of success up in our industry. Novel AI/machine learning technologies will make us much more comfortable with probability and uncertainty. And if you're more comfortable with probability and uncertainty, you know better where to invest, where to focus and where to deploy the energy to make drugs available for patients. That's the mission where we are on, and it is fantastic for us that on that mission, a great team is already working, and this great team will be expanded by a new Chief Business Development Officer who will join us as of 1st of May. Matthias Evers has spent his life in trying to become not only an excellent scientist, but also someone who is deploying his knowledge of technology into innovation processes in the industry. With this, we are very happy to welcome Matthias as an excellent addition to the team where we feel that the power of our technologies through him will be even better networked into our partnered network in the industry. With this, let me round up what you have seen today and let me tell you that this was only a small potpourri of what we can and will show you in the future behind every technological capability and every capacity that we have built within Evotec. If you have time and if you go to Page #8 of your presentation once you reflect on this day, you should imagine that behind every logo there, we have best-in-class capabilities, most inventive people within our organization to drive the industry forward. And with this, we feel that we are in a strong position to set up the pace for Action Plan 2025, which we comfortably confirm today. We're in a comfortable position to build out our R&D efficiency platforms even further with novel partnerships to come and expand -- partnerships to come, and also with a year of strong capacity investment as you see it in 2022 and 2023 on our strategic road map. Our precision medicine platforms have breakthrough potential that allow us to co-own drugs into the future via royalties and milestones to come. This is an essential building block for our co-owning strategy, which all ends with the idea to create a royalty pool in the long run on top of a highly successful and highly profitable business. And doing the right thing for the right target with the right modality is best exemplified by, for example, what Linda has shown you with our Just-Evotec Biologics platform for antibodies and biologics. We do this in all modalities. But with our Just-Evotec Biologics, I think we are ready and probably even ahead of schedule to fulfill the mission to bring biologics to all, and that's the slogan behind EVOaccess. Let me stop here and open for questions. And let me, before I do this, thank my team of co-presenters here, let me thank everyone who helped in the background to organize this day because it was Anya, it was Gabi, it was Volker and their teams who really helped significantly to make this an almost Netflix-like experience of how great science can be and how it can flow. And let me with this open the floor for questions.

Thank you, Werner. Yes, we hope to keep track on that expectation management. So I think a few questions centering around the efficiency platform first, addressing Craig, in particular, if we can provide the audience with a few examples of efficiency gains, benefits in speed and efficiency. And also related to that, we have outlined that the average time saving is 35%. What would be the benchmark for that?

Thanks. Thanks for the question. We've -- on a number of occasions previously, we've highlighted the latest benchmarks for -- for example, the cumulative cost to IND, cumulative cost to first GLP dose and the time line that it takes to get there. And the industry benchmarks have been remarkably stable over many years, not really improving, which is an interesting entailing tail in itself. So typically, benchmark-wise, it takes 5 to 6 years to go from any new target idea through to an IND or a first GLP dose step. And the benchmark costs are somewhere around $80 million -- $70 million to $80 million, which includes the cost of attrition. And that's important to be in the mind given that the theme of this is about improving the probability of success in reducing attrition. Typically, portfolio-wise in Evotec, we can accelerate that by 30% to 40%. That's what I referred to. And the cumulative cost of a portfolio project set in Evotec is 40% to 50% reduced compared to those benchmarks.

And maybe a follow-on as you are on the line already related to the evolution of discussions with partners in recent years and even months, based on the overall market dynamics that we saw and also the even more integrated external innovation model that we're running.

Yes. It's really interesting because as we've grown in capacity and also the comprehensive nature of our offering has developed, the increase in the number of discussions that we have with partners at a portfolio level rather than a single project level has gone up dramatically. And what's really interesting, I think, and better relevant for the AI and machine layering theme here in this presentation is that, that portfolio view where we're conducting multiple projects for partners simultaneously allows us to learn across the project portfolio by creating a chamber of data -- a data chamber for the partner. And this allows us to learn from one project to the next. It allows us to accelerate the whole learning process and the learning loops for the whole portfolio. That allows us to increase probability of success within the portfolio view for that partner and exploit the data with machine learning tools for the benefit of the partner and also for the platform. So this step-up in volume and portfolio view is really the clear direction of travel for us and for our partners.

Thank you. Before we get to the 2 largest blocks, mix in total and also just first 2 questions on E.MPD for Uwe and Christiane, I guess, and then sells for Nele and Cord. So first on E.MPD, for kidney disease, it's clearly very large, and this effort is now well established. What is already placed in other disease areas? And what is the time line to similar capabilities as an kidney? And secondly, we also stated today that E.MPD is one of the largest and highest quality molecular databases today on a global scale. How do we measure quality and how do we benchmark the database against other proprietary ones, not the public databases? Maybe Uwe and then Christiane.

Yes. Thanks a lot for this question. Start with number one. So yes, you are totally right. And as I said during the discussion and the presentation, we started with kidney disease back in 2017, '18. But as you probably have seen in the presentation, we have also added already liver diseases like NASH, NAFLD to the portfolio with the QUOD organization. So that's the next step where we have access to thousands of patients and data and samples there. In addition, we have reached out and we are in really advanced negotiation process with other cohorts in the liver disease, metabolic disease area, NASH again, but also inflammatory diseases. And Christiane highlighted a number of of these upcoming cohorts and collaborations on her last slide. So inflammatory diseases, you probably have seen our investment into IMIDomics very recently. That's an equity investment. But also in that area, we will be active very, very soon. And if I say very soon, this is going to happen this year for inflammatory diseases, liver diseases like NASH, et cetera. Other areas, again, Christiane highlighted oncology, neuro and infectious disease as well. So Christiane, anything to add there?

Yes. Maybe just on the infectious disease side, it's clearly that there is an unmet need, and that's really our goal to look at diseases where we have unmet needs in the diagnostics space where we can support diagnostics companies, but also enhance our treatment discoveries, which we are having ongoing in -- within Evotec. So we are looking into treatment monitoring in the infectious disease area, building molecular patient databases with foundations in the future. So that's another area where we are in the short, mid- and longer term, potentially extending into.

Thanks, Christiane. Then the second question on this quote, E.MPD is one of the largest and highest quality molecular database globally. So this quarter is actually, to a great extent, based on direct feedback from both our academic collaborators, but also pharma partners who are giving this very positive feedback. In particular, the pharma partners, as you might imagine, that are always performing a deep due diligence on our data are very impressed by this database. And again, the database is taking a holistic view and not only a focus on genomics analysis, as Cord also said earlier today. So this is a pass that many others have taken, genomics only, but we are really going much, much deeper with our multi-Omics, PanOmiccs approach. Now I should also probably mention that we have looked into a number of biobanks and cohorts and finally only picked those to be added to our E.MPD that come with sufficient relevant information and also high quality. As you can imagine, there are huge differences between all these biobanks and cohorts. And also, as I said before, we have been looking and still doing so into public domain data sets that many times come with very limited information. And yes, that's always a limit for data analysis, as you can see in my presentation. So I hope this answers the question. Christiane, anything to add?

On the -- maybe on the platform side. So as we are acquiring in a lot of these cases the data on the transcriptomic side, on the proteomics side, we are able to standardize the quality of these data, and thus, enabling our PanHunter tool to really get meaningful insights into these disease areas. And as you have seen from the data graphs, which we have presented, really control the output of the data, which we get, and thus, the insights, which we gain.

Thank you. So then we move on to cell therapy, in particular. Given the recent clinical durability issues, earlier patient relapse and things like that, we saw from CAR-T approaches, what can your platform add to the mix?

Yes. Thanks. I think with our platform, we can enhance the capability to find treatments across all modalities. It doesn't have to be small molecule based. I think the eye is a very good target organ also for other modalities such as AAV treatment, for example, or azos because the eye is very easily accessible. And I think these treatments can only be modeled in our retina in a DISH system that we are approaching, and therefore, will greatly reduce attrition or the risk of failing in the clinic later.

And on cell therapy related to our diabetes program, what progress have you made on your diabetes stem cell project in the last 3 years since you had the right standard back from Sanofi? When might we see back the asset in the clinic? And can you give us an update on any discussions on finding a partner? I guess that's for Cord and Werner.

Let me start. The evaluation of options is ongoing as we speak. We are still more committed probably than ever before to what we now call our CureBeta initiative because we made significant progress on both elements of that; one, on understanding the cells; and second, how to apply these cells into the human system. And with this still in 2022, you will see a lot of progress on CureBeta. And maybe without saying too much, I hand back to Cord on why we are so positive on that.

Yes. So essentially, it's -- I think we are still very much on track on moving forward here. We have made great progress, especially on the device end of things. And so I think we -- overall, we are very optimistic that we will achieve key value inflection points in 2022 and also that these key value inflection points then should ultimately translate into partnerships or third-party financings of this particular program.

Okay. And then we move on to Omics and all these aspects. First, I think a question related to the experiences in the past. In genomics, validated drugs doubles probables of success that's known, also, the effect of biomarkers having a positive impact on probability of success. So why is the industry not more proactive here in adopting these approaches? And is it still the case that we still don't know enough about biology in that case, in that context?

Maybe I take that question. So I think it's an excellent question because as Volker, as you said where we post the question, everybody is sort of aware of the power of Omics technologies and how helpful they can be. As I tried to outline in my presentation, though, Omics technologies have been used all over the place in the drug discovery process. However, they have been used sparingly mainly because they come with significant costs attached to that. And essentially, especially when trying to move Omics technologies, in particular, transcriptomics and proteomics, into the mainstream of the drug discovery process, it means drug screening, in particular, in profiling drug candidates, in preclinical disease models. Here, once again, these technologies have been employed, but rather sparingly and mainly because of simply the -- it's difficult to get the right throughput, and it's very difficult then also to get the cost proposition right. As you can imagine, for example, doing a proteomics experiment on profiling a drug candidate is very significant -- comes with significant costs attached to it. And this is exactly what we focus at Evotec, increasing the throughput, maintaining highest quality of data and overall bringing down the cost in order to bring these technologies into the mainstream of drug discovery. And ultimately, the second challenge is and -- was and is actually that you're, of course, then generating huge amounts of Omics data. And unless you can effectively crunch and analyze these data sets, the -- spending the cost and generating them is simply not worthwhile. And here, we are just getting to the point, I would say, nowadays that we have sufficient amounts of reference data sets that allow us to really interpret highly complex and high dimensional Omics data sets. And also with the use of -- without the advent of AI, machine learning technologies, all of this wouldn't even be possible. So I think this is really a field where we are still at the very beginning. And as I said in the -- during the presentation, ultimately, I think the whole industry now is starting to move into these areas. But I think if you would ask around in the industry how many companies are capable of conducting transcriptomics analysis and 3 84-well format and really screening hundreds of thousands of compounds using this as a primary readout, there will be very few companies who can actually claim that they can do this. The same is true when it comes to using proteomics as a primary readout for drug screening purposes. We are not aware of any other company that has done that so far, and Evotec is really here at the forefront of things. So I think overall, the overall dynamic continues to be that more and more Omics data will be generated. This will drive an enormous demand for being able to analyze it and yourself being for any company active in the space, being able to generate a high-quality Omics data on their individual programs and being able to analyze this.

And I think there's one question closely related to that. We're still sticking to the organotypic indication definitions of disease areas. And the question is, does our data and model systems not go beyond these organs definitions? What's our prediction how this will challenge the industry?

I think generally, when it comes to integrating Omics data into the mainstream of the drug discovery process -- and here, I'm -- particularly, I want to point out when we talk about this, we're talking mainly about transcriptome and protium data from disease tissues. We are actually moving beyond organotypic definitions of disease. We are moving actually into the molecular description of disease and also stratification of disease via molecular phenotypes, because ultimately, if you do not use this stratification, you will always run into the same issues that even though you -- the same organ may be affected by a certain disease, the process that leads to the disease may be completely different from one patient to another patient population. And the only way to get to a really very definitive stratification is via molecular mechanisms. And these will then point to new intervention points for specific patient populations. So ultimately, the short answer is we are really moving, and I think the whole industry is moving to a molecular description of disease phenotypes. And this will ultimately also define patient populations.

And maybe last one on Omics before we then look over to Just and Biologics manufacturing. Given that there's still a number of years out before Omics-based drug development leads to a product in the market, have you back tested PanOmiccs and PanHunter with unsuccessful drugs or drugs that were only successful in certain subpopulations.

What we clearly can see here and we -- the back testing, I think -- actually, Paul was presenting a very nice example of this, which essentially involved building in the world's largest molecular profile for DILI prediction, for example, where we essentially use the whole panel of compounds that are known to have DILI issues from market experience. And we can definitively show that we can recapitulate or identify these issues even in the preclinic very, very clearly. And this indicates that we don't have to fly blind nowadays anymore. We can actually predict safety issues at an early stage without having to go through extensive clinical trials, and even bringing drugs to market and then only realizing that there are safety issues attached to them. And the same is true also for efficacy prediction. So I think the only way all of this works is actually by benchmarking or creating benchmark databases that involve molecules with known profiles and then basically use these as reference data sets for future predictions. Yes. So essentially, in a short way -- short answer is yes. So they're constantly cross-testing sort of drugs with known efficacy and safety profiles, these are sort of the foundation of the databases in order to make predictions for the future for the next generation of drug candidates.

Thank you. The remaining questions related to PanOmiccs and PanHunter will be then answered after the Capital Markets Day. We will get back to you. That's a promise. So now a few questions left for Linda on the J.POD, in particular, starting off with that and capacity here. How many production lines we are now running in Seattle? And by when do we plan to have full capacity in the facility? And actually, what CapEx would be related to that in order to reach $250 million in revenues by 2025? Maybe that's then for Werner, the second part.

Yes. I can start. Today, in Washington and across our 2 facilities, we have 2 by 500 and 1 by 1,000. We have 1 by 500 in Seattle and 1 by 500 in Redmond and 1 by 1,000 also in Redmond. We're planning to put our next train in to be online in 2024, and it's expandable all the way up to 6 by 1,000 in Redmond. And then I can let Werner answer the second part. Perfect. Or I can answer. It's up to you guys.

Maybe Craig, it's better if you would...

Perfect. Thanks for the question. So as Linda said, we -- one of the beauties of J.POD and the whole concept of the podular approach and the trains approach is that we can rather responsively build out capacity as the demand swings in. And so this means that Linda's forecast is absolutely credible plan. But in the event of even faster demand, then we kind of respond very dynamically and get more trains online relatively quickly within like a 6- to 9-month lead time. The addition of an additional train within the infrastructure because the infrastructure is all day and the space available to expand, the addition of -- the cost of an additional train is somewhere between $15 million and $20 million for CapEx.

Thank you. I have to say the CapEx is going up these days when it comes to prices quite nicely. So that's also a dynamic that we are facing in this industry, but that was always kind of something that we factored in -- while, again, this podular system makes it so elegant to do this because you synchronize your CapEx investment also with your partners who are there once you do -- once they sign up to work with you in the long run. That's maybe one part. And the other part, we are in full time line to start and make the groundbreaking of J.POD who -- which represents around $200 million investment in Toulouse, which is not happening in 1 year, but will happen up to the year 2024. And that also indicates to you how much capacity and demand that we see coming.

And last question on market dynamics and biologics manufacturing demand for continuous manufacturing and discussions with potential pharma companies. James who sent this question sites biologics competitors who suggest that some beta companies are adopting the technology rather slowly. And how does our continuous manufacturing platform stack up against competitors?

I can address that. We know from -- when from our sciences out in the field giving talk that continuous manufacturing is a wave of the future. The companies are headed that way. We've seen competitors start to build some of that capability. So we are out in front of it as we also have been discussing with regulators. So in the future, we will see, I think, that all of biologics manufacturing will probably eventually shift to this. Whether it's fully end-to-end or hybrid up to a point, it just is more efficient and cost effective. But there's also been large pharma, as you mentioned, has made substantive investments into stainless steel, very large tanks. And of course, there will probably always be a place for that as those investments have been made. And some molecules may be -- that may be okay because there's not a high mass demand. But we think with the advent of the hundreds of biologics and antibodies that are starting to go through regulatory processes, and as we know more about the biology of the indications that we'll continue to see this area grow and grow and grow as it have robust growth in the past. And we're happy to be at the forefront of it.

Okay. I think with that, we have reached the end of today's Virtual Capital Markets Day, 2.5 hours. Hopefully, very informative for you. We haven't answered all the questions that were coming in, which I think is a good sign. And with that -- and the promise that we will get back to you and answering this -- sorry, these questions, I'll hand back to Werner.

Yes. First of all, let me thank you for your interest. Let me thank you for understanding that the power of technology that we are generating is only there for one reason because there are still 3,300 diseases that need cure, and we are fully dedicated to this mission at Evotec together with our partners to deploy these technologies for cures. But what we should all see only if we bring probabilities of success up and if we make faster processes available in this industry, the chances of doing this during our lifetimes increase because they have to increase. And with this, let me thank not only you as our partners in the outside world supporting Evotec, but let me especially thank our more than 4,000 people working for the company who do an amazing job to integrate the best of science, into the best of technologies, into the best of partnerships to make more cures for patients. Thank you so much.