Hi, I'm Garry Pairaudeau, Exscientia's Chief Technology Officer. At Exscientia, our AI platform is ideally suited for solving multi-parameter drug design problems. This allows us to quickly and efficiently discover novel drug candidates with superior properties. We developed a deep pipeline of programs, including our internal projects, which are focused on precision oncology. You've heard us talk before about the hallmark of an Exscientia drug being the combination of precision design and personalized medicine. Today, we're gonna talk a bit more about the precision design element. Before we begin, I'd like to remind you that we may make forward-looking statements during this presentation, which are subject to a number of risks and uncertainties, including those described in the slide and the risk factor sections of our SEC filings. Exscientia has led the field of AI-generated molecule design, including putting the first AI-designed compounds into clinical trials. From when the company was founded to where we are today, we have made significant advances in our technology and AI capabilities. We've developed a comprehensive physics-based platform encompassing molecular dynamics and quantum mechanics, which is combined with our AI generative and active learning capabilities. We have also significantly advanced our engineering and data platforms, enabling both scalability and robustness. Today, we will illustrate how we've used our AI-driven platform to tackle several challenging drug targets, solving complex design challenges to achieve balanced drug molecules that meet the desired target product profiles. Here we are introducing two new molecules from our pipeline, an LSD1 inhibitor, EXS74539, and a MALT1 allosteric protease inhibitor, EXS73565. These two molecules, along with EXS4318, our clinical stage PKC-theta compound that was in-licensed by BMS, are great examples of how our AI platform can solve complex problems such as kinase selectivity in the case of PKC-theta, brain penetration coupled with reversibility in the case of LSD1, and allosteric inhibition in the case of MALT1. I'm now gonna turn it over to Dave Hallett, Exscientia's Chief Scientific Officer, to walk through the data and discovery process for the LSD1 inhibitor and the MALT1 inhibitors that we have developed. Thank you, Garry. First, we'll highlight our precision-designed LSD1 inhibitor, EXS74539 or 539 as we refer to it. While LSD1 is a well-characterized target, there were complex challenges to overcome in designing a molecule that addresses the multiple parameters required to meet our unique target product profile. Based on emerging data from first and second-generation LSD 1 inhibitors, we set out to design what we believe to be the first LSD1 development candidate to combine a reversible mechanism of action with CNS penetrance. Other molecules currently in development either irreversibly deactivate this enzyme, which can contribute to safety risks or are reversible inhibitors, but with little to no CNS penetration, and are therefore unable to address brain metastases prevalent in advanced disease. Our key challenge was to design a molecule to address these deficiencies and provide a predicted human half-life suitable to explore dosing regimens to manage the known on-mechanism safety risks. We believe this will allow us to most effectively target LSD1 in key cancer types where its overexpression plays a critical role in tumor survival. In these patients, the development of brain metastases is common, hence the importance of brain penetration with our molecule. Today, IND-enabling studies for 539 are ongoing, with additional updates expected later this year. Looking now in slightly more detail on LSD1 as an important epigenetic regulator of gene expression, I'm sure some of you are already familiar with its role in hematology and its importance in oncology. A key function of LSD1 is to remove methyl groups from lysine residues on histones, which plays a critical role in regulating the expression of genes which suppress cellular differentiation. Overexpression of LSD1 occurs in several tumor types, including small cell lung cancer and AML, where it drives cancer cell proliferation and survival. Inhibiting LSD1 in these cancer types with five-three-nine has the potential to provide a superior patient solution, targeting both peripheral disease but also enabling the potential treatment of the brain metastases that are common at initial diagnosis and which subsequently develop in approximately half of all small cell lung cancer patients during treatment with current standard of care. After brain metastasis, the five-year survival rate is less than 2%, and this is why CNS penetration was a critical design parameter for Exscientia and for patients. How did we go about designing five-three-nine? At the outset, precise objectives were defined for our LSD1 inhibitor, including potency, a reversible mechanism of action, selectivity, brain penetration, and tight control of duration of inhibition, given the functions of LSD1 outside oncology. LSD1 has an important role in the development of blood cells. Sustained inhibition or degradation of this enzyme will lead to a reduction in platelets, a type of blood cell important for clotting. The enzyme needs controlled drug holidays in order for platelet levels to recover. The problem with irreversible inhibitors is that new LSD1 protein must be resynthesized in the cell for its broad functions to be restored. This process takes more than one day. This also occurs if a protein is degraded. Even with a reversible mechanism of action, it is still important that the half-life in humans is short enough to allow the enzyme to recover. Think dosing once or twice a day rather than once a week. Ultimately, we were seeking a molecule with well-balanced properties. Overall, the highly differentiated molecule we've developed offers a unique profile with favorable ADMET properties that should allow effective management of on-target toxicology. To highlight how different 539 is from other LSD1 inhibitors in development, here we compare 539 against two competitor molecules, specifically looking at factors such as CNS penetration, mechanism of action, and predicted, as in the case of 539, or published clinical dosing regimens. As you can see, only 539 achieves the unique combination of a reversible mechanism, suitable CNS penetration to target brain metastases, and a predicted human half-life aligned with once-a-day dosing. Our molecule also met a long list of other key criteria, such as selectivity against related enzymes, high bioavailability in preclinical species, and in vivo efficacy in relevant models of small cell lung cancer, a potential indication where five-three-nine may have benefit. Garry will now describe the design principles behind five-three-nine. five-three-nine was precision-designed by Exscientia's team of experts through the application of our AI-driven drug design platform. In order to achieve the challenging target product profile, in particular, creating a molecule with good brain penetration, we had to explore new chemical space. Using 3D evolutionary algorithms, which exploit the positive features of each region of the molecule, we were able to create a candidate drug with balanced, high-quality, drug-like properties. Exscientia's generative design algorithms produce populations of molecules meeting specified optimization criteria. Our machine learning models efficiently scored the compounds for CNS penetrance alongside optimizing multiple parameters, including potency and ADME properties. We applied active learning methods to select the most information-rich molecules to make and test at each design cycle. This typically provides a list of 10 best representatives-20 best representatives covering different structures, usually from a much longer list of hundreds of potential molecules. We're designing novel molecules using artificial intelligence. One of the key challenges with this approach is to make predictions outside of the domain of applicability of our machine learning models. We do this with a set of algorithms called active learning. Active learning was used to make sure that we make the most informative compounds in each design cycle and helped us make an early breakthrough through some counter-intuitive choices of molecule to synthesize. The Pareto front, shown here as the dotted arch, identifies the predicted best molecules to make and test. Our active learning approach selects compounds both close to and away from the Pareto front. Some of the selections, which are actually far away from the Pareto front, whilst having suboptimal brain penetration, allowed us to find a new starting point for design, uncovering the chemotype which ultimately led to our candidate, five-three-nine. I'll now hand back to Dave to talk about the data we have generated for the molecule 539. How does this compound perform in vivo? Here we highlight the dose-dependent oral efficacy observed with five-three-nine in a small cell lung cancer xenograft model, together with a well-described blood-based neuroendocrine small cell lung cancer tumor biomarker, progastrin-releasing peptide, or ProGRP for short. On the left, we can see monotherapy efficacy in a small cell lung cancer xenograft when dosed daily up to 3.3 milligrams per kilogram, dosed twice a day. On the right, you can see the corresponding effect on plasma ProGRP levels. At 3.3 milligrams per kilogram, ProGRP levels are below the limit of detection in our assay, and this tracks beautifully with maximal effect on tumor volume. It is important to note that five-three-nine was well-tolerated throughout this 28-day mouse study. Here we showcase the potential safety benefits of having a reversible mechanism coupled with a shorter half-life. This additional in vivo data highlights the activity of five-three-nine compared to an irreversible LSD1 inhibitor currently in clinical development. In the left panel, you see the outcome of a 20-day study in a mouse model. five-three-nine was dosed on a twice-per-day schedule at 6.6 milligrams per kilogram. 16 hours after the final dose on day 20, platelet levels were the same as control. This is in stark contrast to the irreversible inhibitor. Despite only administering this compound once a week at 0.4 milligrams per kilogram, mouse platelets remain substantially depleted on day 20. On the right panel, you see the outcome of a 15-day rat non-GLP study where we looked at platelet levels as a function of both dose and time. Note the higher doses of five-three-nine compared to the efficacy study and the dosing schedule studied here, three days on, four days off. Whilst we initially saw a reduction in rat platelets followed by a rebound off drug, by day 15, platelets were back at control levels. We believe the exquisite control of LSD1 inhibition and the superior management of platelets will be a critical differentiator for five-three-nine in the clinic, particularly in combination with a standard of care that often has negative effects on platelets. The final piece of in vivo data we wish to share is the effect of intermittent dosing on tumor volume. Previous experiments allowed us to build a model which integrates mouse pharmacokinetics with exposures required for different levels of tumor growth inhibition. Using this model, we selected five different dosing regimens, both continuous and intermittent, that we predicted would produce 80%- 100% tumor growth inhibition in a small cell lung cancer xenograft. As can be seen from this plot, and as predicted, all five dosing regimens delivered greater than 80% tumor growth inhibition compared to control. This flexibility to genuinely explore intermittent dosing regimens in the clinic and thus maximize therapeutic window are another reason we believe that five-three-nine is differentiated from current competition. The properties of five-three-nine fully endorsed our decision to continue advancing this compound, and we are currently conducting GLP toxicity and safety pharmacology studies. No unexpected findings were identified in the previously conducted two species non-GLP studies. In summary, five-three-nine is an LSD1 inhibitor designed to have a unique set of properties compared to all known competition. We believe this is a really exciting new candidate that we've added to our internal pipeline that has the potential to bring meaningful benefit to patients. We'll be leveraging model-informed drug development to define the best dose and dosing regimen for future studies. In addition, we have ongoing translational work and look forward to providing further updates this year. Here we show another example of how our team of experts used our AI-driven platform to design another differentiated molecule, the allosteric MALT1 protease inhibitor, EXS73565, or 565 for short. When starting to design a potent and selective inhibitor of MALT1, we looked to allosteric inhibition, as this was known to provide a mechanism of selectivity. For MALT1 specifically, designing an allosteric inhibitor is challenging because the allosteric binding site is hydrophobic in nature, which tends to result in inhibitors which lack good solubility and permeability characteristics. It's also highly mobile, which presents specific computational modeling issues. We ultimately met the design objectives to generate a potent and selective inhibitor, and our resulting molecule, five-six-five, also overcomes a safety flag that we saw when profiling other MALT1 inhibitors currently in development, namely the inhibition of UGT1A1. Inhibiting UGT1A1 causes hyperbilirubinemia and is an indicator for certain drug-drug interactions, which have been shown to drive dose-limiting toxicities in the clinic. We believe that five-six-five's very low activity on this enzyme compared to clinical competition, particularly when contemplating the dosing of five-six-five in combination, could potentially improve patient benefit and expand therapeutic options in hematology indications such as B-cell lymphomas. Five-six-five is in IND-enabling studies with additional updates expected later in 2023. MALT1 is an important oncology target, and here we highlight its critical role in the inhibition of immune cell signaling. MALT1 is a central regulator of NF-kappa B signaling, and it's known to support the uncontrolled proliferation of malignant B and T cells in a number of hematological cancers, specifically B-cell malignancies like the activated B-cell subtype of DLBCL and CLL. By inhibiting MALT1, we have the potential to block the NF-kappa B signaling in immune cells. In the context of B-cell lymphoma, Bruton's tyrosine kinase inhibitors, or BTKs, have revolutionized the treatment landscape. These drugs interfere with the mechanisms underlying malignant B-cell pathophysiology, allowing better drug response as well as lower toxicity. However, these multiple mechanisms have also led to drug resistance, which is compromising treatment solutions and needs to be solved urgently. The potential to combine 565 with BTK inhibitors or even BCL-2 inhibitors is an attractive proposition to overcome resistance and drive improved patient outcomes. When considering an optimal target product profile, the team took into account the likely use of a MALT1 inhibitor in combination therapies. In addition to potency, selectivity, and a balanced set of overall properties, we were mindful of potential drug-drug interactions that we uncovered when benchmarking published MALT1 inhibitors and also from an understanding of the BTK literature. Various clinical-stage BTK inhibitors, such as ibrutinib, are known to cause drug-induced liver injury, likely through a variety of mechanisms, and in designing a compound, we did not want to increase the potential burden on the liver. UGT1A1 is part of a standard panel of in vitro tests which examines the potential of a drug to cause drug-drug interactions and usually includes transporters such as OATPs, BSEP, and multidrug resistance proteins. UGT1A1 has an important role in the metabolism and elimination of bilirubin, which is a yellowish pigment that is made during the natural breakdown of red blood cells. Its yellow color is the visible sign of jaundice. Once its breakdown product of red blood cells has been transported into the liver, UGT1A1 extensively glucuronidates bilirubin, providing a water-soluble metabolite that can be transported into bile and urine. Inhibition of UGT1A1 can cause elevated levels of bilirubin, and abnormally high levels of bilirubin place an additional burden on the liver. Inhibition of UGT1A1 is also known to cause drug-drug interactions. We need to achieve exposures of MALT1 inhibitors high enough to deliver efficacy. This may be difficult if the potential for liver toxicity makes it impossible to deliver high enough doses. In addition, MALT1 inhibitors may well be given in combination with other agents, such as BTK inhibitors, which themselves can occasionally cause liver toxicity. In other words, the cleaner the safety profile of a MALT1 inhibitor, the more likely our ability to dose it high enough to safely deliver efficacy, even in combination with other treatments. How do we match up against our desired target product profile? The chart displayed here compares five-six-five directly with published and patented MALT1 scaffolds from various groups. Five-six-five compares very favorably across all parameters, examining potency, cellular activity, and drug-like properties. Indeed, the only area where you see yellow instead of green was aqueous solubility, which is still favorable compared to other agents in development and has not hampered subsequent development of five-six-five. More importantly, five-six-five has very little activity at UGT1A1 and is highly differentiated in this respect. Indeed, by our calculations, we would predict that many of the other compounds will likely inhibit UGT1A1 to a meaningful degree and thus present challenges in clinical development. Gary will now describe how we designed 565. Thanks, Dave. Exscientia's precision design approach focused on optimizing multiple parameters important to achieving the desired target product profile in parallel. In this case, we were optimizing physical chemical properties alongside MALT1 potency and selectivity. The design of 73565 is one of the first examples where we merged molecular dynamics with artificial intelligence. MD simulations provided additional insights into the critical binding interactions within the allosteric site, which resides in a highly flexible region between the caspase and the IG3 domains of MALT1. As mentioned previously, MALT1 is a great example of why we have incorporated molecular dynamics and physics-based approaches into our technology stack. Using hotspot analysis allowed us to map the allosteric binding pocket, highlighting key interactions needed for design. Molecular dynamics enabled us to understand the flexible motion of that binding pocket and develop a design strategy to improve the potency of the molecule. Using this picture of the binding pocket, combined with knowledge of other allosteric MALT1 inhibitors, our generative design algorithm, Gambit, was used to evolve novel molecules. Five-six-five itself was identified in iterative design cycle 13 and in fewer than 15 months from the commencement of novel design. With that, I'll turn it back to Dave to share the latest data that we have on Five-six-five. The first in vivo data we are showing you was obtained in a B-cell lymphoma cell line that has been shown to be insensitive to the BTK inhibitor ibrutinib, both in vitro and in vivo. The graph shows the effect of 565 in a xenograft model after oral dosing. All doses provided statistically meaningful efficacy as monotherapy. The compound was dosed twice a day and was well-tolerated during the 28 days of dosing. The second piece of in vivo data was obtained in a different B-cell lymphoma cell line, TMD8. This cell line is sensitive to both MALT1 and BTK inhibition in vitro but is poorly responsive to these single agents in vivo. This cell line is a model reflecting a more recalcitrant lymphoma that we are likely to also observe in the clinic and which may benefit from a combination approach. The graphic nicely highlights the synergy seen when our MALT1 inhibitor, five-six-five, is combined with ibrutinib. In this example, monotherapy with either agent was without effect, whereas the combination displayed meaningful antitumor effects. As we have highlighted, we believe we have designed a high-quality allosteric MALT1 inhibitor with a potential key safety differentiator. Five-six-five has excellent pharmacokinetic properties across preclinical species. It was well-tolerated with full tumor growth inhibition observed at doses well below rodent maximum tolerated doses. It has a low predicted human clearance and high oral bioavailability. Human pharmacokinetic predictions support once-daily administration with a low drug-drug interaction risk. Importantly, even for the highest predicted human dose of five-six-five, the risk of inhibiting the UGT1A1 enzyme and several transporters involved in bilirubin disposition is low, resulting in a low hyperbilirubinemia risk and a low burden on the liver. GLP toxicity and safety pharmacology work are progressing as well. We've now highlighted another new exciting program with 73565, an allosteric MALT1 protease inhibitor designed to have a unique set of properties. Importantly, to be able to be used in combination whilst avoiding the potential for additional toxicity from drug-drug interactions. We believe this is another differentiated candidate that we've added to our internal pipeline and another opportunity to bring meaningful benefit to patients. We look forward to providing details on the next steps for this program, including sharing additional work from our translational platform, including in CLL and other subtypes of non-Hodgkin's lymphoma with unmet need. Thank you for joining us to hear about our progress with these new precision design molecules. We look forward to updating you on our progress throughout the year.