Evotec SE (EVT) Earnings Call Transcript & Summary

Welcome. My name is Werner. I work for Evotec. Welcome to our first Virtual Capital Markets Day, the R&D Autobahn to cures. For this presentation, you can find supporting material like this presentation online and www.evotec.com. Thank you to start with that you are participating to discuss a bit deeper with us some of the topics of our business model where we want to go deeper with you, that you learn more about our platforms and our long-term strategy that we want to discuss together with you. On the left-hand side of your screen, you can see a question mark. This is the button to push when you have a question throughout this presentation, just do it, and we will make sure that your question gets answered either in the Q&A session that follows the presentation or that it gets answered later after this event, if there wouldn't be enough time. Let me also tell you that this presentation will be recorded, and the webcast will be available on our website afterwards. All the information in this presentation contains forward-looking statements, I want to make you aware of this. If you go to Page #3 of this presentation. Let me again ask you for joining in this call with one idea, that we have come together here to discuss about Evotec. But more importantly, we have to discuss the future of R&D. I'm here together with our leader of integrated drug discovery, Karen Lackey, and our leader of Just-Evotec Biologics, Jim Thomas. And of course, I'm here with our management team, CSO, Cord Dohrmann; our COO, Craig Johnstone; and our CFO, Enno Spillner. Our Head of Corporate Communications, Gaby Hansen and Volker Braun, our Head of IR and ESG, will support throughout this presentation later when we come to the Q&A session. Maybe a few words of introduction to Karen and Jim, as some of you might have not met them before. Karen has an outstanding career in pharma and biotech, especially, she had an outstanding career at GSK. And Karen joined Evotec about 3 years ago to then take over our integrated drug discovery organization. Jim also has an outstanding career within pharma and biotech. Jim held leading roles within Immunex and then Amgen. Jim is founder of Just, now, the President of Just-Evotec Biologics, and Jim clearly is one of the key thought leaders in biologics discovery up to manufacturing processes globally and in the whole industry. With this, we have come together here as a team, to introduce you to the Autobahn of R&D. If you go to Page #4 of this presentation, the Autobahn is an infrastructure toward research and development and the Autobahn is something that is a symbol for us. Because it has a clear direction, it enables fastest possible speed and you have various [ lines ] available on the Autobahn. These lanes on the Autobahn for us are modalities. And most importantly, on the Autobahn, there is no detour. Getting an infrastructure together with dedicated teams of brilliant scientists, that's what brings winning teams together, and this is what brings winning infrastructures and teams together on our Autobahn. For lack of better terminology, we so far, call it Vrooooom, this Autobahn for R&D. If you go to the agenda on Page #5, it should just highlight how we want to spend this morning or afternoon with you. Where most importantly, we want to really not only go through our business model, but also show you how data-driven processes make precision and efficiency in our industry completely transforming of what we currently do. If you go to Page #6, let's take one step back. And let's take one step and think about November 5 of this year, which was, again, another important wake-up call for this industry. Aducanumab would have been the first new Alzheimer treatment in nearly 2 decades, but the FDA said that there was not enough evidence of its effectiveness to slow down cognitive decline. After more than 15 years of research and after clinical and preclinical spend of more than EUR 4 billion consistency of evidence for disease-relevant was just not there to convince the experts, but also not to convince the patients. 6 million people in the U.S. and roughly 30 million people have globally Alzheimer disease. This number will even double in the next 20 years. So no question that our industry has to do in a much better job to serve patients going forward. This is what's driving Evotec. For us, R&D precision and efficiency are not only a skill, it's an attitude that every Evotec employee brings to every partnership that we built. We want to dramatically expand and accelerate access for better drugs for more patients. On this point, let me also mention that this industry today probably does a better job than ever before because it is this industry in times of COVID that has come together stronger and better than ever before, and it is this industry that in record time, and you have seen in the last days, has delivered data points that will be novel vaccines, has delivered novel antibody data points that will be novel therapeutic antibodies and also we will see novel antivirals that basically will show that this industry can provide answers to the big problems of health care on this planet. So I think we are really at the starting point of a massive era where health care and the health care industry will show to society how much value we bring and what we do in order to serve patients. If you go to Page #8, we very often say that it is just the beginning, and we mean it. Our mission is clear, and it is in full synchronicity with what society needs. We put drug discovery ideas in leading technologies across all modalities into action. And with this, we enable to accelerate the development [ of medicine ] with our partners. So when you see our purpose statement, research never stops, this is really the motivation for us to make diseases disappear. And when we say disappear, it is truly the motivation of our partnerships and our technologies to not only symptomatically treat diseases, but we want to make the more than 3,000 different diseases that are currently out there that are untreatable, truly go away. If you go to Page #9 of this presentation, you'll see an explosive mix. The mix of development costs that are skyrocketing. And at the same time, peak sales of drugs that are going down is clearly not good. The good news is that R&D investments continue to go up. But at the same time, given the 2 equation elements that I've told you before, the IRRs are falling dramatically within the last decade. So this is the grand challenge, where Evotec has put its business model right in the center of it. The grand challenge of R&D is to reverse this trend and increase IRRs for novel drugs. The biggest driver for this negative trend is that many drugs are failing with very high costs in late clinical trials. Because very often, disease relevance from the beginning is not visible, and that's why people spend way too much money and time in order to then find out that products don't work. If you come to Page #10, you see the DNA that Evotec is built on. Evotec has built data-driven processes on the basis of a data-driven organization. Here, we are creating value chains from the very beginning of drug discovery up to the manufacturing of commercial drugs. These data-driven processes allow us to make discovery insights very early. And if we have genetic correlations, we can double success rates. If we can stratify and have proper biomarkers for certain indications, we can even increase probability of success threefold. If you then can combine these precision tools. And here, Cord give you an insight into our PanOmics and PanHunter tools that we are bringing forward here. If you can combine this with the clarity of experiments that bring you forward on this R&D Autobahn, and this will be the focus of Craig, Karen and Jim's presentation, you can see how to create much more efficiency on a drug discovery process forward. Here, of course, the efficiency tools of today are artificial intelligence and machine learning tools that we basically apply throughout the whole process in drug discovery across all modalities. More precision combined with higher efficiency, this is ultimately the equation that allows us to help our partners and Evotec to reshape the R&D return curve. And this reshaping of the R&D return curve ultimately is the entry point for our co-owning pipeline because people are very happy to share with us if we help them be faster and better at more efficient processes to the market. We very often say multi-modality is reality. And it is the most fundamental change that we have done in the last decade and especially in the last 5 years, that we have fully embraced the idea to create from a small molecule core-driven company, today a company where all modalities are available in an unbiased fashion on Evotec's platforms. Why is this so important? Because we can apply now the best modality for any given new target and bring this target forward together with our partners. If you go to Page #12 of this presentation, please keep in mind whenever you are thinking of Evotec, whenever you think about our global footprint and our global network, this Autobahn today has more than 3,400 employees driving R&D projects forward. 3,400 scientists and employees at Evotec, this is an almost unparalleled R&D power that we can put behind problems of today. And we are definitely, from a quality perspective, unparalleled in industry. And with this, we also have the ambition, the power, but also the qualification to basically tackle every R&D problem of today. Our business model that many people often ask us about. It is very simple. We have a unique business model that is combining best of both worlds. We run a very focused, highest quality research and development and intellectual process generation process that we call Evotec Innovate. And at the same time, we have built a scaled partnering service business. This allows us to have the highest efficiency combined with our partners, and it allows us to have the best learning curves in the industry. And on both axes, we can basically make progress. And if we combine the output of both axes this is the material that allows us to create co-owned pipeline products with many of the partners that you will see soon. Very important for our strategy is that typically we do not sponsor or finance clinical trials, but we love to co-own clinically upside from co-owned product. Our building material for co-ownership typically comes from, first, that we create unique intellectual property or knowhow that we can offer a starting point for first-in-class and best-in-class products. Secondly, our platforms increasingly become the starting point for us and our partners to have more precise starting points for certain indication areas, and this gives us co-ownership to certain projects. And thirdly, significant efficiency gains for higher precision through our technologies allow us to better achieve competitive positions for drugs that we work on. If you look at Page #14. To some of the financial key performance indicators of the last years, you see that the model works. I think it's fair to say the model works extremely well. But it's just the beginning of massive, massive value creation that we will put on top of our profitable business that is profitable every day, once these value-creating co-owned products also will deliver significantly higher milestones and ultimately royalty rates. Page #16 summarizes the model of how we look at it today. It holds significant value potential, and there are 3 doors that basically allow us to form partnerships with co-ownership ideas behind them. If you look at the 3 entry doors that we want to focus on today, you can see that our platforms, like our induced pluripotent stem cell platform, our protein degradation platform, our PanOmics platform, our PanHunter platform, our health, human antibody and also our high-value integrated drug discovery insight, are platforms and services that basically define starting points for co-ownership. Of course, individual drug targets that we come up with out of Evotec Innovate can form co-owned partnerships. And of course, also our operational ventures and academic bridges, and company formations that we are building can be the starting point for co-ownership. With this, every transaction that you see from Evotec is not the ending point, but it's the beginning of risk-free or risk-free upside to the future that we are building. You see on Page 18, just a summary of selected portfolio highlights of our deal-making of the last 5 to 10 years. But what is very important for you to remember and to notice, every transaction is the starting point of future leverage of our assets together with a partner. And this allows us to be not linked to one partner only, but to create a massive portfolio with basically an endless number of potential partners who leverage this portfolio together with us. And when I say endless, it's at least as large as the more than 3,000 diseases out there, where we want to and where we can create partners. This makes the strategy so unique that we, first of all, has a big vision. But secondly, that we have a very clear tailor-made way of creating alliances that meets the demands of our partners and to satisfy the demand of Evotec. Because the demands to form these partnerships are different. For example, our first typical partner group are large and mid-sized pharma partners who bring their key motivation to work together with us from the need to have more innovation, higher speed and massive execution power. Why? Because large pharma partners will only make money with their drugs if they're the first or the second to the market. Here, massive execution power and innovation power of Evotec really helps to accelerate these processes. Our second group of partners are typically biotech companies and increasingly virtual biotech company formations who need higher quality infrastructures, who need the perfect work plan to come with a high-quality guidance to value points and very often to their initial clinical entry points. And our third partnering group, which is growing very, very fast, are disease-focused mission-driven foundations who basically want to follow their mission and, with academic partners, use Evotec's platform to translate best science as fast as possible behind their cause. This portfolio is delivering already significantly -- here's just a few selected numbers, but there is cash that has been generated from upfront, way north of 200 million already. The portfolio today holds way north of 7 billion potential milestones. And -- and this is the most important point -- more than 100 assets own royalty rates that ultimately will come to Evotec's P&L, once we bring these targets together with our partner to the market. Page 19 gives you a summary of what a typical blueprint transaction looks like. So here, you can illustrate that, first, we invest into IP building or knowhow creating processes. And after a typical time of about 3 years and about EUR 5 million spend, we go out to the market and [ form a license ]. What happens at this point in time is that we get very valuable market feedback that allows then either to form these alliances or to reshape the project that we are running. But what is typically happening is we create when we create a partnership a free [ auction ] with a lot of upside going forward and a lot of leverage going forward together with our partners. Selectively, we also trade in that we say we take lower upfront versus the idea of higher milestones at the back end. And that, again, is a testament to the fact that we can create tailor-made partnerships. Very important for us is portfolio thinking and, with portfolio thinking, moving the best of a portfolio forward in the most efficient way. If you look to Page 20, one of the central themes of our strategy to build a portfolio has always been to create not only one horizon where we have small molecules, but to create 3 horizons by basically having small molecules, biologics and novel gene and cell therapies. So these are the 3 horizons that have come together where you now have more than 118 assets that we co-own. When we say co-own, these assets are de-risked, but nevertheless, from the very beginning, we always have skin in the game and have skin in the game of these partnerships. And that's what success is bringing, that basically 2 partners are coming together, both have skin in the game and develop a portfolio forward. If you look at the first horizon, the second horizon and the third horizon here, you can see that this portfolio is not only growing, but it's growing at much faster speed than ever before. Page 21 shows you 100 highly attractive co-owned individual assets. And if you look at them, you should be aware of one partnering basically rule that we have built. Only if a partner is progressing an asset that we co-own at the optimal speed, he is allowed to keep the asset. If for whatever reason, our partner would slow down or not continue development an asset would fall back to Evotec and would give us the opportunity to repartner or reshape such asset. This is the guarantee that at every moment in time, every asset on the Autobahn is moving at the best possible or optimal speed. And that's how it should be on the Autobahn. People should go as fast as they can with optimal speed. Discovery is, of course, a high-risk area, and we are fully aware of that. And Page 22 shows you that we have created a massive discovery co-owned pipeline already. But also our partners are aware that attrition in discovery is real and that's the reason why we are, at every alliance, really trying to create a cultural fit first, where the idea of filling fast in order to optimize value generation is built into all these discovery partnerships that we have made here. Most importantly for you to notice is that at this stage, we are creating significantly more new alliances than that we are losing because of attrition. When you go to Page #3 (sic – Page #23) of this presentation, let me tell you that I'm so happy that you will see that the clinical pipeline visibility will gain strongly in the next 12 to 24 months. This is the case because now we are starting to harvest what we have seeded in the last 5 to 10 years, also with clinical visibility. But most important is that these clinical upside points will come with minimal or no clinical cost to Evotec. If you go to the next page, you basically see, and we will start now to give you a few examples of all 3 horizons. Horizon 1, which is gene and cell therapies, so just to remind you of the fact that Evotec was one of the first companies in 2012 to systematically invest into induced pluripotent stem cells. Here, we have built a platform which has more conserved cell lines than any other competitor in the industry or on the platform. Here we have built more assays than anyone else, and here, we have clearly defined the benchmark lighthouse of induced pluripotent stem cells. When it comes to induced pluripotent stem cells, on the next page, you should see that we are so excited about this technology because it gives you one thing from the very beginning, and this is disease relevance. And this disease relevance, we can translate into either cell therapies or into drug discovery projects. When it comes to a portfolio building approach in gene and cell therapies, you should see on Page 26, just a snapshot of how the currently more than 20 projects that are ongoing, where not all of them are partnered by moving forward if you just would give this a 2-year view going forward. One highlight could be here that next year, the first novel -- completely novel discovered iPSC-derived target could influence the clinic out of our BMS iPSC collaboration. And then you will see that we are also moving multiple on quality projects forward within our cell therapy approaches because this is the conviction that we want to build and want to just underline that cell therapies in immuno-oncologies go extremely well together. And with this, our IPC platform becomes a competitive tool. And of course, the third disease area that we are focused on, and this has been a historic focus of Evotec since the acquisition of DeveloGen, is diabetes and cell therapies in diabetes. Everything that we do in this area is called CureBeta and is bundled in our CureBeta initiative. Page 27 shows you why we are so convinced about this paradigm shift in diabetes. Because today, you see a market which is growing massively in diabetes, it's almost like a pandemic, as many of you know. And the treatment modalities that we have are basically still insulin only. The long-term side effects of insulin are highly underestimated. [ Beta ] server generation would be the obvious choice not only to address this massive market opportunity, but also to develop a cure into a disease area where we could massively improve the quality of life. Page 28 shows you only one beta-driven illustration of our belief here in our very, very strong ongoing discovery project here. With a beta cell implant or potentially with a beta cell infusion, you can observe the normalization of blood glucose levels. This is, of course, what you ultimately want to achieve, that you normalize blood glucose levels with a beta cell infusion or with a beta cell transplant. This combined with our GMP capabilities basically combine a process and a product, and that makes cell therapies then truly efficient to bring them forward. As mentioned before, we are here at this stage in late preclinical development. We expect this project to enter the clinic latest by 2022, and in '21, we will define our business model going forward, which will either be a new pharma partnership, or which will be a leveraged risk share capital formation into a CureBeta initiative. When it comes to the second horizon of our portfolio, let me switch here a little bit into one area where novel biologics are not that popular so far -- biologics in infectious diseases. So here, we have detected and analyzed multiple diseases where, with novel biologics, a significant health burden could be addressed completely differently than in the past. Chikungunya, for example, is one of these widely unknown diseases that represent a horrific health burden. This mosquito-borne virus is very prevalent in the tropical areas and recently you have also seen cases in Europe. Chikungunya is often misdiagnosed, but most importantly, you have nothing to treat chikungunya with. If you now can come up with a novel treatment, you have something which makes you not only highly attractive to the market, but also makes you highly attractive to the WHO and the FDA who has given chikungunya to the priority voucher list where such a voucher alone would represent a value of between $80 million and $100 million. You see the strong data behind chikungunya on Page 30, where the idea to develop a prophylactic and a therapeutic antibody at the same time, was so attractive that the NIH joined forces this year together with Evotec and where we have last week initiated Phase I clinical trials together with Duke University fully sponsored by the NIH. This also shows you a fantastic synergy between Just-Evotec Biologics, who will be the designed -- sorry, the designated commercial manufacturing partner for this product going forward. And it also shows you a blueprint of how stockpiling for products that should be ready for use once they are needed, could look like in the future. Another biologic, which we have in our second horizon under development, which is highly attractive, is hepatitis B. In hepatitis B infections, we do not go for any symptomatic treatment or not for any vaccination, which is out there, we ultimately go for cure. There are currently no immunomodulators in the clinic for chronic hepatitis B that, aside from interferon, has demonstrated clinical proof of concept. Interferon leads to a functional cure in less than 10% of the patients out there. So that's why a cure by a novel biologic would address a significant market, as you can see on Page 31. Page 32 shows you that basically, with our HBV [ uptime ] program, we would create a bifunctional biologic, which is engaging a validated anti HBV target in interferon, together with a novel target CD40. This together would basically induce B and T cell responses for anti-HBV responses. What could be imagined is a once per month treatment, which ultimately, after a limited period of time, would lead to HBV's functional cure. We are also here in late preclinical development and expect Phase I to be initiated in '21. Maybe of note, also here, Just-Evotec Biologics is the designated clinical and commercial manufacturing platform. The third horizon of our co-owned pipeline are small molecules. And of course, within this third horizon, you have often heard us talk about our P2X3 antagonist now called Eliapixant. This is together with Bayer. In Page 33 shows you the fantastic complementarity of disease know-how and capabilities and capacities of these 2 partners. That's why our partnership that was initiated in the year 2013 has led to a full portfolio of novel small molecules that are now in clinical development and that are moving along. This partnership model was so successful that we have also rolled it out to chronic kidney disease and to other disease areas in women's health like PCOS. Page 34 illustrates to you not only the high unmet medical need for one of the applications of our P2X3 antagonists, chronic cough. It also shows you that this is an area of medical need, where currently you have no treatment available. And with this, the more than 15 million patients would be very willing to pay a fair price for such a drug. Page 35 shows you why we think that the data that we have at this stage presented a better set -- that our partner at this stage has presented to the outside is fully justifying the fastest possible development to the market. You see here a drug where all taste side effects have been basically disappearing after the treatment. And you see a drug where the activity and selectivity is very high. And with this, at least from all that we see within the competition, this could be the best-in-class drug, and that is also the motivator why Bayer has initiated here a very large Phase IIb trial in October of this year. We are very happy. And also, we are very happy today that we can announce to you that 2 more large indications are already on their way with Eliapixant. One of them is endometriosis, and you know how bad currently the diagnosis, but also the treatment of endometriosis is. And another Phase II, which is underway, is in overactive bladder. Overactive bladder represents one of these diseases where with aging, basically, the market is growing every day. So this is just a snapshot of many of the ongoing portfolio aspects out of our 100 co-owned assets. And if you now go to page number 38, let me bring you again into this business model of creating a massive co-owned clinical pipeline. And let me bring you here a bit into the numbers. Because what you see here are 3 points in time, where we basically show the evolution that is reflecting also attrition from the past to the presence. And where you see how this co-owned pipeline seeding phase is leading to a harvesting phase into the future, and we are today at a fantastic starting point. You can also see here that, of course, our R&D investments have not been growing dramatically over the past and, with this, you can see a very efficient translation of our R&D investments into this co-owned pipeline. This is especially highlighted on Page #39, where you can see that over the last 10 years, with very focused R&D investments, we have generated substantial revenues that are today visible within Evotec Innovate and that come from R&D payments from our partners, from milestones. But what you can also see is that Evotec Innovate, despite the fact that we do not, at this stage, have any product incomes already shows a fantastic EBITDA, which is fluctuating around the 0 line despite the fact that we are investing around 45 million of R&D this year alone. And of course, if you translate this forward, you can see this co-owned value going up every day. With this, the EBITDA hockey stick effect, will latest come into the company once you have substantial royalty rates that basically fall to our bottom line directly. And this motivates us always to say that this is just the beginning of what we are doing, and you will see significantly more of that into the future. And when I say beginning, it really a lot is -- really a lot of that is driven by our fantastic platforms that have been built. And a lot is coming from the really fantastic hands of the team that Cord is guiding. And with this, after a very short video clip, I will hand over to Cord who will bring you into 1 or 2 platforms that really will change the way we think about drug discovery of the future. Thank you so much from my side here. I hear you then later. [Presentation]

Good morning and good afternoon to everybody joining us here today. Let me start by saying that Evotec does not only have a unique business model, as just described by Werner, but also a unique approach to precision medicine. This unique approach covers various components, which we assembled over the past 10 years to form one coherent precision medicine platform. This platform is ultimately the foundation of our Innovate business segment, which drives our co-owned product pipeline. Before I dive into describing our precision medicine platform in more detail, I would like to make a few more general remarks to set the stage. The health care industry is constantly evolving. And just like in many other industries, incremental change is not good enough to stay competitive. Much more importantly, incremental change is not good enough to deliver on our common goal, that is to develop effective treatments for diseases, which are currently untreatable. And better treatments for diseases where drugs are available but are not as effective or as safe as they should be. The competition is very fierce in this space, and we are working in highly attractive markets. On the flip side, we are working in an industry where project failure is part of our daily business. The increasingly rising costs for drug development, which are currently eclipsing $2.5 billion annually per drug -- per development, are a constant reminder of this. It is clear to everybody that we need to do better. Based on many technological advances and new biological insights, the opportunity to change the odds and improve the success rates has never been as clear and as achievable as today. Actually, at Evotec, we believe that we are living in a golden age of technological advances, and I would like to mention just a few examples here. First of all, sequencing. Today, sequencing is feasible at almost unbelievable throughput and costs that are unimaginable just a few years ago. Sequencing the first genome was a huge task, which took $2.7 billion and almost 15 years to complete. Today, we can sequence a human genome within 1 day for just a few hundred dollars. Moreover the ability to conduct single cell sequencing opened up a new world, single cell sequencing redefines cell identities as well as the disease status of individual cells. But it is also -- enables the profiling of drug candidates with much more granularity than ever before. Another example is iPSC technology has changed the world when it comes to disease modeling. Patient-derived disease models have become the new gold standard in profiling drugs in the preclinic. Genome editing is a revolution in the field of gene therapy, but it has also a great impact on many aspects of drug discovery. And last but not least, artificial intelligence and machine learning tools are changing the way drug discovery is conducted across the whole value chain. There's hardly any process that is not affected by machine learning or artificial intelligence, and you will hear more about this later from my colleagues. The diligent routine incorporation of these kind of technologies into the critical path of drug discovery will not only improve overall productivity, but most importantly, improve the quality of drugs in development. With this, I'm moving to Page 44. Here, I would like to discuss a few Omics technologies, which are critical in this context before I take a deeper dive into Evotec's approach to precision medicine. Generally, Omics technologies are widely available and often used. However, they are not routinely applied to the critical path of drug discovery and development. For example, when it comes to genome sequencing, we have simply not sequenced nearly enough genomes and effectively connected these to medical data sets to learn what the average genome tells us about our own health. All we have learned so far is that gene mutations do not perfectly predict therapeutic outcomes of drugs. And that most conditions that are caused by a combination of low penetrance variants in the genome, which often lie outside of [ coating region ]. This means that genome sequences only give us a glimpse of our predispositions to disease. They do not measure the disease status or disease progression. To measure disease status and disease progression, we need transcriptome and proteome data. Transcriptomics and proteomics allow us to directly measure how a genome interacts with the environment in the context of an organ, a tissue or a cell. As transcriptomics and proteomics are unbiased and comprehensive readouts, they are crucial for a better understanding of disease processes and, in particular, disease-relevant molecular mechanisms. Unfortunately, these technologies have not been scaled to the same extent as genome sequencing, but constant progress is being made here as well. Higher throughput transcriptomics and proteomics will allow a more routine use of these technologies across the drug discovery value chain. Overall, we actually do see a dramatic increase in the use of Omics technologies. The 2 key drivers here are lower costs to generate the data and the adoption of machine learning tools to support the analysis of big Omics data. As you can see on the next page, the pace with which we are generating genome, transcriptome and proteome data is constantly accelerating. And in the last 3 years, has been growing exponentially. Actually, we have generated more Omics data in the last couple of years than in the past 10 previous years combined. This is really a staggering development if you think about it. If this development is any indication of the future, we will be facing a tsunami of Omics data coming from patients. The sharp increase in the generation of Omics data is being driven by 2 important developments. The cost to generate Omics data is constantly decreasing. The best example here is the cost of sequencing, which has come down by several orders of magnitude. From thousands of dollars per megabase, it is now a fraction of $0.01. Together with improved throughputs, this massive reduction in sequencing cost is driving a dramatic increase in the sequencing of genomes, but also in particular of transcriptomes. A major motivator to generate more Omics data than ever before is the fact that we have developed machine learning supported tools that enable us to work with these huge, high dimensional data sets. These machine learning supported tools make it possible to deal with not just hundreds but thousands of Omics data sets in parallel. On the next page, I would like to make 2 points. First of all, at Evotec, we believe that we are still at the very beginning of the position medicine megatrend. And secondly, at Evotec, we also believe that we can play a significant role here in delivering on its promise and develop more effective therapeutics and also more precise companion diagnostics. In many ways, it is really hard to believe that we are still living in a world where 90% of all drugs only work in 50% of all patients. This inefficiency causes enormous physical and social suffering, not to mention the associated economic burden. The estimated annual cost of ineffective treatments is considered to be at EUR 350 billion annually. Precision medicine is the best way, if not the only way, to improve on this. For the last 10 years at Evotec, we have been building what we believe will be the precision medicine platform of the future. This platform has a number of essential components which have to be brought together in order to work properly. First of all, we are building proprietary molecular patient databases, which are required to understand the molecular mechanisms of disease and thus lay the foundation for precision medicine approaches. Connecting molecular patient profiles with clinical metadata really redefines health and disease and will also measure disease progression more adequately than ever before. Secondly, we have built an industry-leading iPSC-based drug discovery platform, which can today model diseases in about 15 cell types. Patient-derived cell-based assays are crucial to more accurately model disease. That's why iPSC-based assays have become the new gold standard to profile drug candidates in the preclinical setting. Furthermore, we have built a proprietary multi Omics data generation platform called PanOmics. It is industry-leading when it comes to throughput, robustness, cost efficiency. And in particular, in the -- this in particular in the field of transcriptomics and proteomics. Finally, we have built an artificial intelligence and machine learning supported Omics data analysis platform, which is called PanHunter. PanHunter can deal with Omics datasets and is also able to put these in context on relationship with preclinical and clinical meta data sets. PanOmics and PanHunter together are uniquely suited to support Omics-driven precision medicine approaches throughout the whole drug discovery value chain. So we believe that the identification of disease-relevant molecular profiles in patients is fundamental for most precision medicine approaches. Once disease-relevant molecular patient profiles have been defined, they can be used for screening and profiling of drug candidates in patient-driven disease models. And ultimately, they can also be used as biomarkers to monitor disease progression and during clinical development as well as in the market. So this brings me to my next section. What does Evotec's PanOmics platform look like? As you can see on the page, we can cover -- on next page, we can cover all major Omics platforms. However, our particular focus has been to move transcriptomics and proteomics into the mainstream of the drug discovery process. For both, transcriptomics and proteomics, we developed proprietary platforms with unprecedented performance in regards to their throughput and, thus, associated costs while maintaining high standards when it comes to data quality, robustness and reproducibility. We achieved this by tinkering with all aspects of the entire process optimizing each and every step from sample preparation to data capturing and data management. And very importantly, we automated as many steps along the process as possible. Higher throughput drives more data generation and thus the challenge to be able to analyze these high dimensional Omics datasets. Specifically, for this purpose, we have built PanHunter, a multi Omics machine learning supported data analysis tool, which allows data scientists to work with these huge amounts of data in a very user-friendly fashion. Now on the next slide, I would like to make the point that the term transcriptomics covers many different kinds of transcriptome analysis. It covers bulk RNA sequencing, high throughput RNA sequencing, but also single cell RNA sequencing and spatial transcriptomics. And although single cell sequencing and spatial transcriptomics are important new innovations, it is really high throughput transcriptomics, which is game-changing for a number of reasons. High throughput transcriptomics is necessary to build large molecular patient databases effectively. To conduct, for example, single sequencing analysis of spatial transcriptomics for tissues of 10 or 100,000 patients is currently simply not possible. Only high throughput transcriptomics enables drug screening and higher -- higher throughput and lead profiling, providing a single but unbiased and comprehensive readout that monitors all positive but also all negative effects of a drug candidate. High throughput transcriptomics will also change the way we run the animal models. Transcriptome analysis of the most -- 20 most important organs in an animal will essentially provide an almost complete view of most drug effects in an animal model in a highly quantitative manner, and thus, we can essentially create a transparent animal model. For the reasons I just mentioned, we built a high throughput transcriptomics platform called ScreenSeq. It is shown on the next slide and fulfills all criteria. It is run in a 384 well high throughput format, so we can run screens of up to about 100,000 samples or compounds, and this more than covers the requirements of any typical screen. The detection limit is around 15,000 genes, which is also more than what's needed for most purposes. And ScreenSeq works for most tissues from animals as well as humans. So essentially, it's bridging the gap between the preclinic and the clinic. And finally, all of this can be done at reasonable costs considering the vast amount of innovation that transcriptome analysis provides for each and every compound screened. To illustrate the point that our ScreenSeq platform is industry leading, we have extensively benchmarked it -- its performance against all commercially available platforms. And this is shown on the next slide. For time reasons, I won't be able to go into great detail here, but as you can see, ScreenSeq's performance is clearly superior in essentially all key parameters. With this, I would like to move to Evotec's proprietary proteomics platform, which is described on the next slide. And as the proteome provides any more important information on the status of a cell or tissue, we also worked hard to improve the throughput of our proteomics platform while maintaining highest performance on the proteome coverage. Similar to ScreenSeq, we tinkered with every aspect of this platform to achieve unparalleled throughput while maintaining highest quality standards regarding proteome coverage and reproducibility. We have already partnered this platform with BMS and, together with BMS, we are using this proteomics platform to screen a compound library using whole proteome analysis as the primary and only readout. This is probably one of the most exciting projects we are currently running and, to our knowledge, has never been done in this industry before. I'm very convinced that you will hear more about this groundbreaking approach in the future. Once again, to illustrate the unique performance of our proteomics platform, we extensively benchmarked it against competing platforms. And I won't have time to go into great detail here as well. But once again, ScreenPep performs better than all other commercial platforms in the industry on all key parameters. In the following section, I would like to switch gears and present a few examples how we are using PanOmics and PanHunter platforms to generate value. And for this, I'm moving to Page 55. I would like to start with our initiative to build proprietary molecular patient database. Molecular patient databases differ from conventional patient databases or biobanks in that they include molecular characterization of key organs, tissues and cells, which play a key role in the disease process. At Evotec, we have been building molecular patient databases for quite some time. And these databases are constantly growing in breadths and depths and are now covering a whole spectrum of indications, which are shown on this slide. I would like to talk in a little bit more detail about our chronic kidney disease molecular patient database, which is a good example for all the other indications. As you can see on Page 56, chronic kidney disease is a highly diverse and, furthermore, only very limited treatment options are available and the disease-causing mechanisms are really not well understood. For example, many chronic kidney diseases are grouped into the CKD of unknown etiology. This group alone makes up 30% of all CKD cases. But even categories that seem more clearly defined, such as the category of glomerular kidney diseases are also far from uniform. This category consists of many different diseases, which clearly are driven by very different disease mechanism. Obviously, it is very important to define and better understand the mechanisms of kidney disease, and only then we will be able to develop better treatments. On the next page, you can see how the CKD molecular patient database grew over the last few years. We started to build it through a collaboration with the Nurture consortium in the U.K. and the Nurture consortium had assembled one of the largest CKD cohorts worldwide with about 4,000 patients. Their database included complete clinical patient data -- patient profiles, including all standard diagnostics and test results as well as treatments. Evotec took the opportunity to carry out the molecular profile of the available patient tissues and samples and thereby add the crucial molecular patient data, which is required to drive precision medicine approaches in CKD. We have continuously expanded this database, which now includes over 10,000 CKD patients. To our knowledge, this constitutes by far the largest CKD patient molecular database worldwide. It is important to note that Evotec has exclusive access to more than 600 billion data points, molecular data points describing kidney disease on the molecular level for the next several years. This places Evotec in a unique position to analyze these data sets long before they become publicly available. This strategy of building and systematically analyzing molecular patient databases is delivering new insight on multiple levels. On the level of genomics -- on the genomics level, patients can be placed more adequately in their respective peer groups from a genome level point of view. One reason why this is important is that the adversities stated by patients on the official forms does not always match what can be deduced from their genome sequence. Quite often, from a genome perspective, patients actually belong into a different peer group than what they state. If not corrected this would, of course, confound any further analysis going forward. Furthermore, molecular patient databases will have a big impact on diagnostics. Quite often the primary diagnosis based on standard diagnosis does not necessarily match with a diagnosis, which is based on molecular profiling. Ultimately, we believe that molecular profiles will be more definitive and more exact when it comes to diagnosis. And finally, molecular patient databases are essential to better understand disease-driving mechanisms and spot opportunities helped interfere with these. Selecting the right target for the right indication is probably the most important decision of any traditional drug discovery project. To underpin the selection of molecular targets with a systematic molecular patient database can establish more definitive links between a potential drug target and a well defined disease. Our strategy to build molecular patient databases has already delivered multiple pharma partnerships, as you can see on this slide. In the last 3 years, we have built 3 partnerships in kidney diseases alone. All of these partnerships are in parts or even completely based on our molecular kidney disease patient database and all partnerships are funded by our partners such as Bayer, Vifor and Novo. The partnerships with Bayer and Novo come with very significant upside for Evotec in terms of success-based development milestones and tiered royalties. This is slightly different in the case of our joint venture with Vifor, which is called Natera. Natera is fully funded by Vifor Pharma, but Evotec still co-owns 50% of all projects. In my last section, I would like to show you another use of our PanOmics and PanHunter platforms, which has the potential to really change the drug discovery paradigm. We previously discussed the importances of molecule disease phenotypes or profiles as a defined and track disease progression on a molecular level. These molecular disease profiles identified in patients can be directly used to screen compound libraries. The reversion of such a patient derived molecular profile in a cell-based disease model essentially ensures disease relevance like no other readout. And at the same time, that allows conclusions about how effective and safe the compound really is. Compounds that revert molecular disease profiles effectively have a much better chance to ultimately deliver the desired effect in clinical studies as they are per definition very disease relevant. But the results of such a screen looks like is shown on the next page. Here, you can see a cellular disease model, which has been screened with a small molecule compound library. In this case, the cellular air sale model fibrosis as defined by a patient-derived molecular profile. This model has then been used to screen a compound library, a small molecule compound library. The primary and only readout used was RNA sequencing conducted through our ScreenSeq platform and supported by the analysis of PanHunter. Every single dot that you see represents the analysis of a complete transcriptome triggered by a single compound. Transcriptomes, which are very similar and profile are also very close to each other. Transcripts, which are dissimilar in profile are very distant from each other in this depiction. The dots in the red circle represent disease cells with a disease transcriptome profile, which model fibrosis. And dots in the blue circle are healthy cells with a healthy transcriptome profile. What we really want to achieve in such a screen is to identify compounds, which move the transcriptome from the disease cells towards the cells with a healthy transcriptome profile. This is indicated by the arrow in between the red and the blue circles turning from red to blue. Most compounds outside the red and blue circles have limited effects regarding the reversion of the disease molecular profile. However, others have very significant effect, as you can see. Some compounds with a strong effect go off into different directions that are clearly undesirable. These are clearly unwanted toxic effects, whereas other compounds are actually quite effective in reverting the disease phenotype, and they do it quite selectively and comprehensively. And these are the compounds that represented by dots that are close to the blue circle, and these are the compounds that we ultimately want to develop. Being able to classify compounds early on in regards to disease relevance and at the same time, being able to discern their selectivity, safety and tox profile is of tremendous benefit if you consider the fact that over 90% of clinical stage compounds failed in late stages of clinical development. Compounds that revert molecular disease phenotypes to the healthy state should have a much higher likelihood of proceeding in clinical trials. Another reason for drug failure is, of course, drug safety or drugs. This Is shown on the next page. Also here, our PanOmics and PanHunter platforms can provide very significant advantage. Drug-induced liver toxicity is a great example as it is difficult -- very difficult to predict and to model. And the fact that DILI is responsible for about 80% of drug withdrawals from the market makes DILI a very important characteristic to determine for any compound. And as you can see on my next slide, we applied our PanOmics and PanHunter platforms to the current gold standard DILI's assay systems. We simply replaced the traditional readout, which is high content imaging with our ScreenSeq tech-driven transcriptome analysis. And by doing this, we were able to achieve vastly superior protection accuracy. The current industry gold standard is -- and DILI has its high content imaging endpoints. This results in daily predictions with about 70% accuracy. Using Evotecs PanOmic and PanHunter platforms, we achieved predictions within 82% accuracy, and we believe that we can even do better than that. Furthermore, beyond this, very significant increase in daily prediction accuracy, our transcriptome analysis not only reveals how likely a compound will lead to but we can also pinpoint the molecular mechanisms that are driving this unwanted effect. The impact of being able to ensure disease relevance by being able to predict tox and safety profiles based on a single readout like transcriptomics predictive power on the discovery -- on the drug discovery and development process cannot be overestimated. It really leads to a paradigm shift in drug discovery and development, and that I will summarize on the next slide. The biggest challenge in the pharmaceutical industry remains to improve productivity. And the reason for this is that 90% of our drugs stay in late-stage of clinical development. So at this point in time of drug development, huge investments have been made and a lot of time has passed by, making the whole process, of course, fairly inefficient. Having a closer look is when you realize that the vast majority of failures over 54% of them are due to inadequate efficacy in Phase III clinical trials. Which means that we either went after the wrong target or the wrong compound. Either way, it was simply is not good enough to achieve efficacy. For this reason, there should be more emphasis than ever before on demonstrating disease relevance of the targets and compounds we select for development. And we should measure disease relevance relative to molecular patient profiles that we know are associated with the disease. This can be done by focusing more on unbiased and comprehensive readouts such as transcriptomics and proteomics which are uniquely suited to measure disease status and disease progression as they capture the more complex picture of a disease. We believe that the shift is ongoing and will only accelerate as the use of Omics technologies to measure and quantify biological processes will continue to grow exponentially. Such a more disease-relevant focused approach will better guide us to find the right drug for the right patient at the right dose and thus help to fulfill the promise of the precision medicine. The following slides in my last slide, just in case, I have been a bit too technical, I would like to leave you with a fairly simple picture about our molecular phenotype driven precision medicine approach. To focus more on comprehensive molecular profiles allows us to see the whole picture of a disease. Furthermore, it allows us to look at the whole picture of the activity of a drug candidate. Only if we see the whole picture of disease and a drug candidate and bring these together, we can be sure that we are on the right path to develop effective and safe drugs. And of course, seeing the whole pictures is a lot prettier than just looking at a subsection. This is where I end, and I will hand over to my colleagues, Craig, Jim and Karen, who will tell you more about our operational excellence, more of how we use machine learning and as well as more about our multi-modality platform. All of these are critical components of Evotec's approach to precision medicine. With this, I'd like to thank you for your attention and after short video my colleague, Craig will take over. [Presentation]

Thank you, and good morning, and good afternoon to everyone from me. As said, for a drug to be successful in the clinic, there has to be both efficacious and safe. Of course, brilliantly discrete some of our technologies, which we expect to transform the translation, the prediction and the precision of those 2 changeable features of our successful efficacy and safety. But in addition, we have to execute the discovery and development process to get to the clinical stage of development with good quality of medicines from any modality, at lower cost, higher speed and with the greater clarity of vision about which patients will benefit from the treatment. A quantum leap, if you like. So in this section with a help of Jim and Karen, together will share with you how Evotec has advanced to the point of very high-performance R&D but is also continuing to invest in enhancing the entire performance of the system through enhanced data exploitation, machine learning and AI, coupled with industry-leading know-how, knowledge and data generation to create the future state of drug discovery and development. On Page 69, by way of brief introduction to my passions, after many years of senior drug discovery leadership and drug discovery roles. And then later mastering how to tailor and apply process improvement and operational excellence to make the inventive process of discovery and development more efficient. I feel the natural next step is to harness machine learning and AI and learning from all the data so that we can invent and develop new essence in a much more cost-effective business model. So on Page 71, perhaps to appreciate the Evotec of today, it's useful to look at what we've been doing in the past few years. In recent times, we've purposefully and very deliberately built the multimodality Autobahn for R&D. We've done this through the acquisition of experienced and seasoned drug discovery talent and technologists. We've made important substantial high quality acquisitions, such as ACTIV and Just and we're focused on enhancing the quality and the executional performance of our processes and cycle times, which support our scientists in their innovation. Cycle time is really crucial for effectiveness. There's no point in us hiring great experienced people and for them to be waiting for results. As a consequence, we've combined and integrated all of these recent enhancements into the -- what we believe is the highest performing flexibly accessible on the engine. In addition, we've identified some of the highest leverage highest impact points for intensive application of computational power and machine learning to improve prediction. Prediction, why is improving production is so important? Well, is because in our industry, the current paradigm of high attrition and high experimentation, the better we can predict, the fewer iterative works that we need to take. The impact of this, as Cord has already said, is potentially enormous and we're already well on this path, as you'll hear later in this section. We believe this path leads to the truly transformational power of medicines discovery in the next decade. If the Autobahn of R&D is today's image, then perhaps the future image will be one in which the R&D Autobahn has a frictionless surface of data on the top, which allows high-speed vectors directly to future medicines, which we know or have a very high confidence will work in the patients for which it's best suited. Maybe it's like comparing Ferrari recent cars today to teleportation vision of the future. On Page 72, that's the concept, if you like, that's the vision, the teleportation of the future. So let's look back closer to home. And look at what's driving our current performance and our business growth. From our position of excellence in the established area of small molecules, we've added additional lanes or modalities beyond the Autobahn. Biologics, cell therapies, gene therapy approaches and indeed more complex modalities of small molecules are all now more possible to run on the Evotec platform. This is a very important advance because it means that we can evaluate new targets and approaches at the very earliest stages of biology. Without being restricted or limited by the biases of technologies or capabilities. Basically, we can follow the science to greater success. And in addition, the best combination of knowledge, experience, computational power and process excellence gives us the best chance to solve problems and create success in the invented space. Failing fast and failing cheap, the mantra for some of the recent yields is unfortunately still failing. Real value comes from creating success rather than failure, which in turn requires problem-solving and inventive step and invention. Our customer return rate of over 90% suggests that our partners feel that we are maximizing their chances of success in this way. And then I think it's worth highlighting that many tech companies have come into health care with AI and machine learning ambitions. But without the platforms to create new data and without the knowledge of which the key problems are to solve in drug discovery and development. I think this is what makes Evotec really unique. The combination of seasoned professional R&D leaders, the platforms to create new data, proprietary data at high-volume and the technology to apply machine learning and AI to the problems to create the best combination of these 3 crucial elements for transforming R&D performance. We'll share with you some examples in design and development readiness in small molecules and antibodies during Karen's and Jim's sections in a few moments. On Page 73. I'm going to follow-on from a lot of concepts and words that I've done so far and move to what we see as the net rolled-up benefit that these advantages are for Evotec and also to our partners in terms [indiscernible] performance effectiveness and speed in drug discovery. The latest industry benchmark data shows that neither the cost nor the timelines of drug discovery have really changed much, typically taking around still 5.5 years to go from a target to a first GLP dose or an IND. Including the cost of attrition, this takes benchmark companies approximately USD 75 million to achieve a single regulatory tox [ style ]. Now at Evotec, we've got very good quality data that shows that in recent years, our combined processes and success rates can achieve the same sort of portfolio delivery to IND at around half the cost of benchmarks and at around 30% less time. This timesaving is really important for us, of course, but especially for our partners. The potential to reach IND, 18 months faster than the competition, generates real added value in a competitive marketplace for innovative breakthroughs. This is especially true in a highly connected world, where many players are pursuing the same scientific concepts and indeed the public disclosures and new science in a straight head-to-head race. On Page 74, particularly in the lower panel, we have a schematic, which indicates how we're achieving this benchmark-busting performance. So first of all, we have multiple ways to start new projects. A very minimal start-up burden and access lanes on the Autobahn, that means we can really quick out the blocks quick to start. Then with intense focus on the core invented process of design, make, test, analyze cycles. This creates the speed to new knowledge and the speed to new insights, and it allows us to front-load the issues, including addressing developability very early in the sequence. And so in turn, this very high integration of technology and disease area knowledge and process excellence under one Evotec roof, removes handovers and white space that reduces the impact of phased transitions really to a minimum. And as I mentioned, creating success where success is possible is really the primary focus of these dedicated teams. And putting resources onto the problem areas flexibly is also part of the advantage of a fully comprehensive Autobahn. But when we find it's not possible to solve a problem, in quick and flexible redeployment of these resources on to other much more high-impact tasks is enabled through the high-capacity and agility of the scientific functions. The net result, benchmark busting performance in the R&D Autobahn, which we have recently called innovation efficiency, and perhaps now we're going to call vroom. Now on my final slide, Page 75. You can see some examples of just as some of our partners for whom this enhances performance has directly contributed to their high-value inflection points, either through asset partnering or through acquisitions. We're really very proud of our contribution to these success stories, and there are many, many more undisclosed examples in the back catalog. This track record and our confidence in the rationale also paves a way to enhance value for Evotec in the co-owned pipeline, our equity strategy and in the future potential of our academic [ bridge ] strategy. So in these past few moments, I tried to bring to life the current performance of the multi-modality on the Autobahn, on which we are laying a new data surface for even higher performance in the future. And now I'll pass on to Karen, who, after a short video, we'll share with you some examples of the underlying computational technologies and the impact that they are having in small molecule projects today. Thank you very much. [Presentation]

Hi, everyone. In the next few slides, beginning on Slide 79, we will be describing our computational drug discovery and development approach focused on small molecules. Our current capabilities include deep learning and computational approaches across the full value chain which rely on Evotec talent having access to or generating the state-of-the-art methodologies and then applying the appropriate methodologies to advance projects. Growth in deep learning and knowledge building is highly dependent on quality data, rigorous analysis, discovery and development expertise and model building. Evotec is positioned to be a leader in this area. The future of small molecule drug discovery needs a breakthrough. That breakthrough needs to capitalize on decades of investment, big data capture, machine learning tools and delivers on the continued unmet medical needs. Evotec's ability to integrate knowledge across the full value chain, drug discovery and development is unique in the industry. This is a distinct competitive advantage. Our partners often come to us with high-priority biological targets for which they need active, selective small molecule modulators to begin the process of creating a drug candidate with a safe and efficacious profile. Depending on the amount of information that's known about the target of interest for related targets, a variety of methods can be applied to achieve the starting points. Evotec is rich with technologies derived from years of characterization, high-quality libraries of compounds to apply to a range of approaches. From high-throughput screening to using fragments in NMR crystallography to diverse virtual screening approaches. We assess what information we have to balance the cost and speed to deliver a successful outcome. For example, one quick statistic in virtual screening, we have a 97% rate in finding actionable active starting points over 5 years working with 82 client programs. Once active starting points are identified, our aim is to get to a drug candidate in the most efficient manner using the application of tools to drive the design and decision-making. On Slide 81, we are showing an example of a telemetry plot for a fairly typical project. In this case, it was for a large pharma client. This plot charts the progress or score from hit to active candidate over the course of 12 months. The score combines multiple objectives, such as potency, solubility and clearance. The Bayesian optimization denoted with a 1 maximizes the information gain at the explore phase of the project. Generative design, denoted by the 2 is an AI approach we used to reinforce structure activity relationship learning in which new molecules are automatically [ donated ]. In this case, we used quantum mechanics to do a protein ligand interaction optimization. And finally, we used machine learning approaches to exploit the learnings for predictive DMPK. While this slide makes it look easy, it's important to point out the curation of high-quality data across important domains of drug discovery and the building of machine learning algorithms with these data has led to a bank of tools that can be applied to projects based on a Gap analysis that is done by our project leaders and our project contributors. Furthermore, Evotec has built a continuum from project ideation to IND-enabled candidates by building capabilities in essential domains for achieving efficacious and safe drug products. Expanding on our computational approaches for TMK designs mentioned in the previous slide, we have extensive human PK and dose prediction capabilities. In our medicinal chemistry toolbox, we have collated over 13,000 transformations that have been successfully used in scale up and manufacturing of drug candidates into a reactions database that can be used to enumerate virtual [ captive ] compounds in our machine learning design methodology. Evotec has extensive computational approaches to efficiently optimize the solid-state form of products covering polymorphisms, crystal structure, soft forms and solubility. And linking back to Cord's presentation on Evotec's development of drug-induced liver injury database and integrated machine learning, we can deliver far safer drug candidates by incorporating insights from using this highly predictive tool. The application of machine learning also provides unique opportunity to closely link successful drug discovery to the identification and development of predictive biomarkers. Using the Evotec Knowledge graph, novel biomarkers are identified in the context of disease and drug mechanism and then cross-validated by testing on high-quality patient samples in an iterative fashion. This allows us to build highly predictive, multivariate signatures with further opportunities to refine signatures based on real-world clinical data. Drugs of the future will be personalized and translatable into patients. And on Slide 84, AI and machine learning approaches have proliferated throughout discovery, development and clinical trial stage of the drug industry. What sets Evotec's small molecule approach apart is the access to high-quality, rigorously generated and carefully curated data and the integration of all of the components of the value chain. For me personally, this is the most exciting aspect of the current paradigm in machine learning in small molecule. Over the next few years, our models will continue to be built and refined, which will lead to higher-quality molecules and experiments. We have deep domain expertise along with full discovery development continuum that allows us to parse the data and train the algorithms to enable an Evotec data surface with end-to-end predictions. Ultimately, our objective is to connect the diseases at a molecular level to molecular interventions. As stated in the beginning of this section on small molecule drugs, the future of discovery needs a breakthrough that capitalizes on decades of investment. Big data capture, machine learning tools and delivers on the continued unmet medical needs. Evotec's ability to integrate knowledge across the full value chain of drug discovery and development is unique in the industry. I will repeat, it is our competitive advantage. The Quantum Leap will occur when we successfully exploit the knowledge built in all of the domains and invent and produce safe and efficacious medicines with biomarkers and diagnostics based on the molecular signatures of disease. And now I will hand over to Jim Thomas to tell you about our approach in Biologics.

So yes, thanks, Karen. I'm really happy to be with all of you today to talk about this great work, I think, that we're doing here. On Page 88 in this part of the presentation, I'll briefly discuss how we're approaching design of our integrated Biologics platform that we call J.Design that's directed particularly toward antibody and antibody-like biotherapeutics. The scope of the platform was from discovery through to manufacturing and facility design. So it's really quite broad. But because we've integrated it from A to Z, it's really quite powerful. In this section, I'm going to briefly touch on how we're using AI and machine learning to create high-quality molecules at discovery and more productive manufacturing processes that can be implemented in a low-cost and flexible manufacturing facilities. There are some powerful concepts associated with the integrated approach of the J.Design platform that are worth mentioning. The first is that we are now capturing all the data from the beginning to the end, and we're using increasingly sophisticated machine learning algorithms to interrogate and learn from the data. This has created a series of feedback loops that continually validate or reject predictions at the various steps of the platform. So we're continuously improving our approaches at each step. And this is why we represent the J.Design as a circle because when we link the various steps together in this way, it creates a large continuous learning [indiscernible]. Executing efficient end-to-end continuous processing in our J.POD facility is really the ultimate validation of picking the right molecule during discovery. So let's go to Slide 89. So now let's open up the loop and examine this somewhat linear approach as we move a molecule all the way from discovery to manufacturing. We start by inputting DNA sequences of hundreds of thousands of antibodies that are in the public domain because it is the least expensive and the most abundant data available. As we know, DNA is a powerful library of information, these sequences contain all the information that we need to efficiently develop and manufacture the molecules that they code for. In this information, we can develop in [ silicone ] structural models and then predict how well they will express themselves and also how best to purify and formulate them. We used a much smaller -- a much more expensive data sets we get from manufacturing to essentially validate our predictions. Let's move to Page 90 or Slide 90. Now I'm going to turn to the J.DESIGN platform. We talked about an exciting approach we're taking to build antibody libraries for discovery. We're using some powerful deep neural network algorithms called GANs, our generative adversarial networks to build very diverse antibody libraries that can be screened against any therapeutic target [indiscernible]. The power of this approach can literally be visualized on thispersondoesnotexist.com website, where GANs are used to generate human faces that are not real people. But they certainly look real enough to fool you and I who have been trained on millions of faces over our lifetime. The way it works is that a discriminator neural network is first trained on a set of neural faces. The generator neural network creates random images and sometimes it fools the discriminator that one of these images is actually a real human face. More human faces are given to the discriminator to make it smarter and more difficult to fool when the generator works harder and gets better at generating these faces that eventually fool the smarter discriminator. So this goes back and forth and you get the picture, right? Or actually, did you get the picture. One of the 4 phases is actually a real -- which of the 4 faces is actually a real person to fill that question out there. And actually it's a trick question because actually, all of these faces have been generated by [indiscernible]. So let's get to page or Slide 91. So now that you know that -- you know how GANs work, you can understand how we're using these powerful algorithms in a similar way for generating very diverse human-like antibodies. The training set for the discriminator are the literally hundreds of thousands of DNA sequences and natural human antibodies we talked about earlier that are in the public domain. So Slide 92, we can use this approach to literally generate billions of unique, diverse antibody molecules that can be screened against any therapeutic target. We have another very sophisticated tool set of algorithms that we call Abacus that we can use to improve known antibody molecules to make them more manufacturable and developable. We're actually using this tool set to improve the training set for the discriminator in the GAN for essentially build into our libraries to transfer learn molecules that have been pre-optimized to be more manufacturable and developable. So we're building high quality, along with high diversity into our antibody libraries, literally building in quality from the very beginning. In the future of our ever-expanding library of team like antibodies, we can screen against internally or externally generated therapeutic targets. Partners who have their own therapeutic target and biology can work directly with us to rapidly select and move a high-quality therapeutic candidate into the clinic. So Slide 93, please. So now it's turned specifically to Abacus. I want to give you a better view of how we use it. Again, starting with this huge training set of natural human antibodies, we can examine and compare the DNA sequence of the lead antibody candidate from the transient mouse, a human being or other sources. And then we can use a computational tool that we've built into Abacus to recommend improvements to the molecule. Improvements might be one or multiple [ assay ] changes that generate [ favor the stability ] in general biophysical characteristics of the molecule. We will then physically make these variants and stream into a battery of robotic high treatment assays to select the best variants for development. While a training protein [indiscernible] has been needed to take the output of Abacus and make the final decision on which variants will be made to improve the native sequence, we're now creating algorithms to replace -- to actually replace the [ still ] protein [indiscernible]. This started actually as a joke from the data scientists to replace their boss, but the tool is actually taking shape. And it's going to be invaluable, especially when we want to screen hundreds or maybe even thousands of potential development cannabis for partners to narrow selection to those with the best [indiscernible] for [ proper research ] , expression purification and formulation. So Slide 94. This slide shows on a couple of antibodies, which -- these are extreme examples, but they do demonstrate the power of Abacus for transforming a very promising and relatively unique therapeutic molecule into an actual drug that could be manufactured and moved to treat patients. The ability to withstand low PH without aggregating it is actually critical for purifying antibodies in the conventional manufacturing process. In these cases, the changes recommended by Abacus resulted in dramatic improvements in stability at low PH. So Slide 95. Shifting now to the -- directly to the process for making these drugs. We've been developing state-of-the-art reagents and methods for highly productive and republished intensifying processes for manufacturing. We're able to reach high expression levels in the period, especially with molecules that we've selected that optimize from our libraries and also using Abacus. We've designed proprietary vectors, host cells, media and culture conditions to routine and reach productivities in the 2 to 4 grams per reactor liter day. So I'm going to repeat that. Actually, we're producing these proteins in the 2 to 4 grams per reactor liter day. So we normally think about this in terms of grams per liter these days and a Fed batch reactor, maybe 5 to 10 grams per reactor. Five to 10 grams per liter over a period of about 12 -- 10 to 12 to 15 days. But in this case, we're producing 2 to 4 grams per reactor every day. So -- and we also use end-to-end processing to continuously purify this material that's being made in the production reactor. This product comes out of the reactor already concentrated and the downstream purification steps further purify and concentrate the product for our clients and partners. On Page 96 or Slide 96. And so there's really no need for large intermediate hold vessels now to store product intermediates and to have central SIP and CIP utilities, because all of our process steps use single-use components. Because our productivities are so high, -- we've reduced our scale of operations dramatically, operating at 500 to 1,000 liters instead of 12,000 to 25,000 liter scale head batch operations. The [ bio rafters ] and downstream processing units are small enough to [indiscernible] autonomous clean rooms, that we call pods. So that's why we call our manufacturing facility J.PODs. The pods are made of extruding airplane-grade aluminum with our own dedicated HVAC units, and they can be moved and reconfigured to run a variety of processes, dealing from a few grams or kilograms to hundreds of kilograms or metric tons. Median buffer vessels are rolled up and [ serially ] connected to pods to support continuous processing. And we can [ go ] in and out of these pods to pass from one ISO control level to another one interim. And the materials surfaces actually make them quite [ cleanable ]. So on Slide 97. The level of complexity of J.POD facilities compared to conventional large-scale stainless-steel facilities is dramatically lower. But that doesn't make these facilities less sophisticated. The complexity has shifted from a conventional manufacturing plant or facility with its miles of stainless-steel piping, fixed unit operations and massive central utilities to a much smaller and self-contained pod configuration, smaller but very sophisticated leaner operations, automation and processing conditions. This reduces cost and speed to build and validate while dramatically improving flexibility and utility. We believe a flexible network of J.POD facilities will create worldwide access to biologics as we work with partners and clients across the globe. Slide 98. So finally, to bring this section to a close, Evotec is creating a multi-modality digital Autobahn for delivering critical industry solutions to partners and clients like enhanced speed to the clinic, better prediction of clinical efficacy and reduced manufacturing costs. It's been a pleasure talking about some of these today with you. Particularly across both the small molecule and large molecule space with Craig and Karen. So I believe after a short video, Werner it's back to you. [Presentation]

Thank you, Craig. Thank you, Karen. Thank you, Jim. When you listen to the technologies that we have built, when you see the co-owned pipeline, how it's evolving, I think you should also appreciate the fact that what we are building is much bigger than Evotec. What we are building is basically platforms and ideas that's changed the way drugs have been made in the past. And with this we allow significantly more people in the future access to efficiently made and also affordable drugs. This principle of the shared economy in drug discovery and development is driving Evotec. Here, precision medicine is paramount. And it is like in many other industries, if you think back 10 or 20 years where you could not have imagined how things will change, but precision medicine is something that is coming. You don't want to have a drug that you take that is toxic and that doesn't work. And this is exactly what Evotec does not stand for. We will make drugs that work. You don't want to have processes that are inefficient. And with this, for us, the integration of machine learning and AI is as logical as using nails and hammers in our industry. But more importantly, it makes us unbiased, and it makes unbiased solutions for the discovery challenges that we want to solve. And we don't want to sell to our partner, what we just said available. We want to create with our partners the best solution for the patients. And with this, this efficiency that we bring to the industry will allow a completely new paradigm of cooperation. Creating a co-owned pipeline is a thought that is massive because what we are doing is we are leveraging the power of all pharmaceutical companies, the power of all biotech financing institutions and the power of all foundations to create with us as a starting platform, a co-owned pipeline. With this, we do not only reduce the cost of a single drug per patient, we also create a business model for Evotec, which is massive and trust at the beginning. This is a short summary of why I want to invite you to a discussion of what we are doing, how we are doing it and where I want you to be invited to go together with us on the Autobahn. The R&D Autobahn for cures. And today, we wanted to give you only a small snapshot of what we are doing. But one thing is clear, vroom is just at the beginning. Thank you so much for listening in. After a short video or a short break, we will answer, of course, all your questions that have been coming in throughout this presentation. And as mentioned already before, don't hesitate to send us any question throughout the day or tomorrow if you have any further questions to this presentation. It's a big pleasure to work with you, for you, and it's a great pleasure to have the team together with me that helped us to bring this together the story. So with this, before we go into the Q&A, let me really wholeheartedly thank Karen, thank Jim, thank Craig, thank Cord and thank also Enno who now did not say too much because he is our CFO, but fully part of the team. And also let me wholeheartedly thank our [Audio Gap] going. With this, let me hand back to the Q&A.

Okay. Yes, of course. Thank you very much. So we are starting right away the Q&A session. I think we keep it in the order that we had presentations. First of all, 2 questions to Werner with regards to the pipeline assets that you described. Want to say it's tough with chikungunya, the biologics here. Can you remind us of the licensing situation with Sanofi regarding the chikungunya biologic, what would be the terms in case they opt in?

So first of all, chikungunya came within the conduction of infectious diseases that we did with Sanofi 2.5 to 3 years ago. This is now a fully owned asset of Evotec that we can develop forward. There is an option, Sanofi to be discussed. These projects with us but commercial terms are not defined. So that's why they would have to go to market [indiscernible]. That's the situation, how we will end [indiscernible].

And one question regarding the growing iPSC portfolio, which was outlined on Page 26. Could you share a bit of a general feedback that you get from your partnering discussions here? What makes you confident to be competitive as large addresses in the market have just recently expressed their willingness to expand in the field of iPSC as well.

Yes, maybe better to hand over this question because is much more directly involved into these discussions to Cord to give you a feel but one thing before I hand over to Cord, is really to say our vision to create this light house of [ ITCs ] is just at the beginning. And the only bottleneck that we currently have is the capacity to build even more, but the demand and the [indiscernible] are just almost endless. With this maybe Cord you can give a bit more flavor here.

Yes. Happy to do so. So the iPSC technology platform, of course, has a number of different applications starting from disease modeling to drug screening and cell therapy. And if you think about it when it comes to disease modeling, it is probably going to be the primary tool in the industry to create cell-based disease models going forward, and there is a tremendous need for that and demand for that. It all depends on having the right -- or accessing on creating the right cell lines here, patient arrived cell lines. And the right conditions for modeling disease, where we have tremendously expanded our platforms in the past and continue to do so. And very often, with many our expansions, actually, our existing partners have opted in on these, and they have been immediately then commercialized. Beyond that, we are, as Werner has alluded, very active in the cell therapy field as well. And here, I think, once again, we are just at the very beginning of this to see that iPSCs, will contribute to cell therapy approaches, not just in diabetes, but in particular, in the oncology field, where CAR-T technologies essentially have shown that in principle, it is possible to create highly effective treatments based on modified T cells and now to really turn these opportunities into products, iPSCs are probably the way to go in the future. And we are investing into the space and we having discussions on many fronts here. But when exactly this will come to pass, I don't want to really comment on this. The -- beyond that, we feel that iPSC will have potential in other areas. There are new modalities coming up in the industry, such as exosomes, for example. And once again, we do feel that the iPSC technologies or technology that uniquely lends itself to the scalable and very robust manufacturing of these kind of modalities going forward. So overall, we have -- we continue to be very excited. We continue to build this up. And I'm convinced that we will be able to announce additional partner just based on this platform.

Thank you, as I have here now on pipeline assets. I think we stick to that and come back to the bigger picture questions for Werner later on. One question was related to the diabetes field. Do we have small molecule programs in that field as well?

Should -- maybe I'll give that a go. So the fact that we are making very mature and highly functional beta cells from iPSC cells for cell therapy purposes does lend itself this approach and also to drug screening and here, even though we are still at the earlier stages, we basically are in the process using these really uniquely suitable human beta cells for drug screening as well, and this also includes small molecules.

All right. And one question related to Beringer assets in respiratory and oncology. When do we see Phase I data?

We have the policy and will continue to that policy that timelines within co-ownership are always aligned with our partners. But most importantly, as also here with the Beringer assets, our partner has an incentive to move these assets forward. There is a diligence clause behind these assets and there is regular contact. And if there would be assets that don't move as outlined in the presentation, assets would fall back to us and would give us repartnering opportunities there. At this stage, very high interest also in novel pain assets going forward, no novel oncology assets going forward out of our Beringer partnership. And that's why we cannot say much more at this point because that would not be aligned with our partner at this stage.

All right. So coming back to the topics of today. I think this is for Cord as well. About the patent protection around [indiscernible] platform? Is there any protection in place?

No, we have not filed any IP on these platforms at this point in time. Also in part because we're not sure how easily enforceable this would be across the industry. And nevertheless, the -- we treat these as the processes, these proprietary processes as trade secrets. And we do feel that these platforms are not sort of -- you don't develop them to a certain point, then never develop them forward anymore. These will be continuously evolving platforms where we will continue to innovate, continue to improve them, moving forward and just trying to stay ahead of the rest of the industry. But there's an aspect to these platforms, which should not be underestimated in terms of intellectual property. And this is through these platforms, you generate huge amount of omics data from patients, from preclinical models, from compounds. And these data sets are Evotec data sets and without -- and these data sets are extremely important as they do serve as reference data sets. For -- to actually interpret future data sets that are being generated. A good example here is, for example, the prediction accuracy that I was discussing regarding drug-induced liver injury. In order to make these kind of predictions, Evotec screened a large set of [ daily ] compounds, known compounds that have a [ daily ] effect, and we generated the omics data. And this is essentially the training data set when we now come up with a new molecule and test it in the same system to actually have as a reference, to make these kind of predictions. And so we feel like it's not just about having a high throughput transcriptomics platform or high proteomics platform. It's being having access to these kind of data sets to build the models to make the predictions. And we certainly -- I have no intention of giving out our data set.

Understood. And a follow-on here on the market environment, which seems to be very crowded in terms of AI and machine learning. Where do you think is this moving to? Are you also considering collaborations with companies such as [indiscernible]? How is data sourced? Is this only in-house in development? Or what is the strategy there going forward?

So also here, many -- maybe to back up on step. So many companies, active biotech companies, particularly active in this space, AI, machine learning space. They usually almost exclusively rely on public domain omics data sets. Unfortunately, from our own experience and hard learnings in the past, we conclude that a large proportion of these publicly available data sets are not a very high quality, are poorly controlled and very not -- and very often not reproducible or have systematic errors in them that make it very -- they don't have an annotation, sufficient annotation to really work with them? So overall, we feel that public domain data sets as of today are there. There's much of them around. But ultimately, we very much rely on data sets that we either generate on our own, and we are investing quite a bit of generating data sets on our own, like I said, in particular for reference data sets. And then, of course, in partnerships with our partners and once again, just I mentioned the BMS collaboration in -- with BMS in the proteomics field. You can never say for sure, but I'm pretty convinced that there's no other company that is generating as much pharmacological proteomics data as Evotec at this point in time.

Maybe [indiscernible] earning aspect, Craig, that you could follow on on that one.

Sure. Happy to pick up from where Cord left off. Cord talked about how fantastic [ power ] we have to generate new data. But as we talked about during the presentation, I think the uniqueness of where Evotec is today in what appears to be quite a crowded environment is that in combination with the AI machine learning tools, we also bring into the mix, the expertise of people who really -- who really know how to do a drug discovery and development, know the kind of problems that we experience and which problems are the real ones that need to be solved. And then in conjunction with that, the AI machine learning and the generation of new data in a high-quality way. And the knowledge of experts in each of the domains, then allows us to bring that combined insight together for really high performance. And the curation of data quality is of pivotal importance. I think Karen actually specifically mentioned that in her presentation. And so the net result is that we feel it in our presentation. [indiscernible] business world and indeed many of the segment or niche AI technology companies. Also come to us because they also see that they need access to the way to experimental work, the high-quality execution and the process excellence that allows them to execute some of their questions in a real-world setting in a wet lab setting with a very high-speed and, of course, high-speed getting to the data is very important if you're doing machine learning because you need the rapid cycles to drive the machine learning, just as in the same way as you would to drive human learning.

All right. So then we've got a number of questions on Alzheimer's and age-related disease. Let me sum it up. First is how close do you think we are to model Alzheimer's, for instance, sufficiently in iPSCs. And maybe that's related to that. Given a large variety of diseases linked to aging, which platform would best recapitulate to cellular aging processes. I think that goes together.