 So what is a drug? Drug that affects biological processes and used to prevent diagnosed or treated disease. Drugs can be of natural origin or produced synthetically. The ideal drug should have a specific action to be safe, toxic, without side effects to be chemically and metabolically stable, to be synthetable, to be soluble in therapeutic concentrations, synvoid precipitation in the bloodstream, within lipids as well. In order to lipid membranes and distribute around the body, finally to be a unique molecule. In order to exact the drug with a human body, not molecules in human body, but the reaction to the drug, these actions are considered by two major sciences, pharmacodynamics and pharmacokinetics. The pharmacodynamics focuses on the drug effects on human body. The other name of pharmacodynamics is pharmacology. From the other side, pharmacokinetics focuses on the effects of the human body on the drug when it is available in the body. It's absorption, distribution, metabolism, and expression, the so-called admin processes or admin properties of drugs. The process of drug development consists of two stages, drug discovery, preclinical treatment and clinical trials. Drug discovery starts with the finding of a heat molecule. So this is a molecule that shows some kind of biological activity. Then this molecule is optimized in improving its affinity, selectivity, reducing toxicity, improving water and lipid solubility, improving admin properties in general, and converting the heat molecule into a lead molecule. So the further optimization of the lead molecule delivers the final molecule, which is the drug molecule. Next, they are the preclinical development. So the preclinical studies are focused on clarifying the mode of action of the drug candidate. It's pharmacokinetic behavior in animals like bioavailability, toxic metabolites, if any, roots of excretion, efficacy of animal models. Efficacy means therapeutic effect. Drug formulation is developed in this stage. Drug formulation means how drug will be delivered in the human body, like a tablet, like a solution, like injection, and the stability formulation is performed here. Then they are the clinical trials. So the clinical trials are the longest and expensive stage of the process, consisting of three phases. In the first phase, up to 100 healthy volunteers are involved. One of these phases is to evaluate the state of the drug in human. It's pharmacokinetics in the human body and the immediate side effects, if there are any. In the second phase, the drug is administered to several hundred patients suffering from the target disease. At this phase, the short term state, then is the third phase. In the third phase are involved several thousand patients from several clinical centers around the world. The aim of this phase is to collect sufficient data for the efficacy and safety of the drug. If the drug passed successfully this phase, so it's ready for registration and marketing. But the surveillance of the drug doesn't stop here. The drug continues to be observed for safety and side effects. So this last phase is known as post-marketing surveillance phase. And practically it is endless. It continues until the drug marked. The pharmaceutical industry is one of the most successful businesses in the world. It's neither by financial crisis, political ones. Because, you know, sick people always exist and unfortunately the number even increases during crisis. So if we look on the financial reports of five of the top 10 big pharma companies, I mean Pfizer, JSK, Rosh, Sanofi, and Movartis for the last 10 years, we can see several quite interesting facts. The cost of goods sold is only 23% of the total income the company's seen average. Almost half of the income, 43%, is spent on selling general and administrative, high percent. In order to reduce this expenses, the companies are constantly merging and quizzing each other. 16% of the total income is reinvested in research and development, which is significantly higher than the average value of 7% for other businesses. So definitely drugs are high value added products. The net income is 18%. Well, this puts the pharma industry in the top three most profitable industries in the world. The first one is the software industry. The second is the hardware industry. And the third one is the drug industry. So it's more profitable than the gun industry that we usually think that is very profitable business. But no, drug industry is better. Actually this is quite good news. Obviously people make more drugs than guns or at least sell them more expensive. The cost of developing a drug increases exponentially. It doubles every 10 years. The cost of a new drug for the last 10 years is estimated average in four billion dollars. Especially the cost of the bio drugs. Small, criminal, antibodies, diagnostic products, vaccines, quite a process of drug discovery and development expensive. Because of its low efficiency. From 10,000 precise and tested compounds around 100 show some activity and safety. 10 of them enter the clinical trials and only one is approved for medical use. Recent studies show that the initial number of tested compounds even passes a million. And this process takes up to 12 years between eight and 12 years. But with the anti-COVID vaccines last year, we have seen that the time for drug development can be significantly reduced. The drug discovery and development is a high risk business as well. In average, seven of 10 projects are canceled preliminary because of reasons. So what are these reasons? Why drugs do not succeed? The main reason is the lack of efficacy or effectiveness. So there is a tiny difference between efficacy and effectiveness, but they mean one and the same. Therapeutic effect, lack of efficacy means that the drug is effective on animals and animal models. But when is administered to humans, the therapeutic effect or is negligibly small. The second big reason in the past was the pharmacokinetics of the drug. Low bioavailability, toxic metabolites, short or extremely long half-lives. But during the last 20 years, many in silico tools and models have been developed to assess the army properties of drug candidates during the experimental stage. And attrition rate due to pharmacokinetics reasons decreases from 39% in the past to the current negligible 7%. You see how the in silico methods changed again. Animal toxicity, adverse reactions, commercial and other issues are among the other attrition reasons. So what is the takeaway of all these facts? First of all, drugs are expensive products. They are high value added products. They are kind of luxury goods, non-affordable for most people in this planet. Let's to be clear, most people in this planet have no access to modern drug medication. And unfortunately, this won't change for at least the next 100 years. How the drugs are covered in the past, how they are discovered presently and how this will be happened in the future. So generally, there are four approaches for drug discovery. Though this one is by serendipity. Serendipity means discovery by chance. Trial and terror. A newer one is the biochemical modifications of known drugs or natural products. The other one is by screening of databases. And the most advanced method for drug discovery is the rational drug design. This is the smartest way. This is the cheapest way of drug discovery. And let me show you now some examples of drugs discovered by these four different approaches. There are many examples in the history of pharmacy for drugs discovered by serendipity. Starting with the most popular, the story about penicillin, a drug which saved millions of lives during the Second War and for which Flaming Florian Chain received a Nobel Prize in 1945. This drug is still in use for some conditions. For us tonight is the first diuretic, also was discovered by serendipity. Cycus purin. Cycus purin was testing as an antitubercular antibiotic but became the first immunosuppressive drug that changed the science and practice of organ transplantation. And the latest story is about Sildenafil by Agra which was developing as an antihypertensive drug but is becoming one of the best-selling drug leading to an entirely new pharmacological group in modern pharmacology. Biochemical modification was discovered last year. The acetyl cericillic acid was acetylated the salicylic acid. Salicylic acid is a natural product. The aim of this modification was to increase the stability and to reduce the irritating effect of salicylic acid on the stomach mucosa. Aspirin is a product, it's not a drug. Aspirin is a product. Salicylic acid is the drug. Salicylic acid is the active metabolite of aspirin. Ranitidin is a chemical modification of simitidin with increased half-life. Pindulol originates from propranolol but avoids the first pass-effect in the liver and shows high bioavailability. By random screening was the first sulfonomate drug Prontosil, when a great number of colorants have been screened for antibacterial activity. Screening also is used presently. The screening could be performed in vitro, experimentally, or in silica, virtual screening. The rational drug-dispose-advance approach of drug discovery, it's clear this now. Drug design begins with identification of a biological target. This is a biomolecule which is involved in the disease. Then compound interacting with this macromolecule is discovered, the so-called heat molecule, and then it comes on iterative process of structure optimization until the compound is delivered with optimal selectivity, affinity, non-toxicity, solubility, permeability, et cetera, et cetera properties which are needed for a molecule to become a drug. There are two main approaches in drug design, the ligand-based and structuralist drug design, when the structure of the target macromolecule is unknown. The structure of the ligand is designed and optimized based on the relationship between structure and activity. So this is the ligand-based drug design, focus is on the ligand. But if the 3D structure of the target macromolecule is known, then the ligand is designed to be complementary to the binding site on the macromolecule. Complimentary means steric, electrostatic, and hydrophobic fitting. So this is the structure-based drug design. It's focused on the structure, on the macromolecule. There are several major achievements in the science of drug design that made it the main approach in the current and the future drug discovery. So first of them is the understanding of drug receptor recognition. In the early 1890s, Mille Fisher compared the drug receptor interaction to the key and lock interplay. Consider that both drug and receptor interacted solid bodies and without changing their conformations. Lately, Daniel Koschnan suggested that both molecules undergo conformational changes during the interaction and adopt the most suitable conformation in order to connect each other. So this hypothesis has been proven many times by X-ray structures, by silico simulations. And now it's known that molecules indeed change their conformations during the interaction and adopt conformations that fit optimally each other context surfaces. I mentioned many times target macromolecule, what is a target molecule? So target macromolecule is an internal molecule, endogenous molecule, which is involved in the disease. Affecting the function of target macromolecule could change the etiology of the disease, could change the pathophysiology, or improve the symptoms. Of the 20,000 protein-coding genes, one genome, about 3,000 have been estimated to be part of the so-called draggable genome or draggable proteins. What means draggable proteins? All these proteins are able to bind drug-like molecules, small molecules, which means they have a binding site. Here, inner ring, in the inner ring is given the human protein divided in four different groups into the target development level. Target development level means how much we know about the given protein. T-clinical, this 3%, these are proteins which are related to at least one approved drug. So this group can have 659 proteins currently. Here in the outer ring are the protein families. So 3%, 16% belongs to gamma-protein-coupled receptors, so these 10 receptors. Three are nuclear receptors, 21% are ion genres, 25% are enzymes, 4% are transporter proteins, 9% are different types of penises, and the rest, 22% belong to different other proteins. Many of them are often receptors. T-chemical, this one in the green, the 6%, includes proteins known to bind with high-potency to small molecules, but these small molecules are not yet drugs, but these proteins have a binding site. T-biology, this in pink here, given 53%, includes proteins that have any link to any disease, but they haven't been studied for binding to small molecules. And T-dark, 38%, their binding 38%, contains the unstudied, totally unstudied proteins. So you see, there is a huge field for future discoveries. Since 2014, there has been an initiative funded by the National Institutes of Health aiming to illuminate the drugable genome. All new data discovered in these and similar projects are collected in a specially designed website, which is called FARUS. So FARUS contain all the drugable proteins which are known by now, and any nutrition is collected here. The three destructures of proteins, are is all x-ray crystallography, NMR scopy, and more recently, by the modern cryogenic electron microscopy and collected in the protein data bank, which currently contains more than 180,000 structures. Some of them are single proteins in apoforms without a ligand, but some of them are in complexes with the ligands. And this information is more valuable because it shows where the binding site is on the macromolecule. Another great achievement in drug design has been the invention of automated methods in synthetic chemistry. This automation allows this a great amount of new compounds generated as a combinatorial libraries and the high throughput screening. 1,000 compounds could be tested per day, this way. But the major achievement in drug design is the development of in silica modeling technologies applied for virtual screening, for compounds design, for energy calculations, SAR and QSR analysis, admin modeling, modeling of drug target interactions. But in order to be applied all these advanced technologies, the molecular structure should be encoded numerically. The encoded structures can be analyzed, can be searched, can be visualized, compared to each other. So it's very important first step in the process of drug design. The structures could be encoded by binary strings, could be encoded by smiles strings, could be encoded as 2D graphs or as 3D, 3D structures. The 3D molecular modeling and visualization also are considered among the greatest achievements in drug design technologies. The molecular property should be encoded. Different types of descriptors have been developed over the years. They could be divided into 1D descriptors, 2D descriptors, descriptors and descriptor sets. The 1D descriptors are derived from the structure, ID descriptor, the molecular weight, ID descriptor is element composition, the 2D descriptors are derived from the molecular ground and 2D descriptors are the number of hydrogen bond acceptors, for example, hydrogen bond donors. The distribution coefficients log p, log d, solubility, number of rotatable bonds, molecular fragments, different type of indices, all this information could be extracted from the 2D structure of the molecules. The 3D descriptors are generated from the 3D structures. Molecular volume is a 3D descriptor. Molecular surface area is a 3D, especially the polar surface area, which is very important molecular descriptor. And finally, the descriptor sets, they contain a set of numbers, a set of numbers describing a complex property or a set of numbers describing a set of properties, but they are used as a set, not separately. Once there is a quantitative description of a set of structures and the quantitative measurement of the activities, like IC50s, EC50s, KDs, then different types of quantitative analysis can be applied. And Koley-Hench did this for the first time in 1964, when correlated the antimicrobial of penicillin derivatives with descriptors relating to hydrophobic and electronic properties of molecules. All these type of models where the molecular descriptors are the independent variables and the activities, the dependent variable, is known as quantitative structure, activity, relationship models. All these are the QSR models. And Koley-Hench is considered the father of drug design. Lately, the analysis have been involved methods like principal component analysis, partial squares, chemometrics involved into chemoinformatics, including molecular modeling, chemical information, and tools and algorithms to handle this enormous amount of data coming from the combinatorial chemist page, DC, which were very poor in the 1990s. At the same time, bioinformatics also emerged to organize and analyze the data, the amount of which also increased dramatically, especially after the deciphering of the human genome. Nowadays, the structure activity relationships are analyzed by machine learning methods, like random forest, decision trees, egg gibbons, neural networks, canyons, neighbors, support vector machines. Many authors classify these methods as black box methods because they're not able to distinguish between irrelevant and irrelevant for the activity descriptors, as the regression methods do, for example. But they use all variables at once, but they're very good in predictions and classification, significantly better than the statistical analysis and the regression models. Both type of methods do not replace each other, nor contradict, they are complementary. They should be used in combination. And here comes the artificial intelligence. So there are many definitions of artificial intelligence. I don't think that AI needs a definition, it's a self-explanation, but well, if I have to give one, I would say that AI is a pipeline of machine learning methods, algorithms, and it's able to mimic human intelligence, to mimic, I underline not to substitute, to mimic. Siri Alexa is AI, self-driving cars are AI. Google discovered Netflix recommendations, so all these are AI applications. There are many examples of AI in our everyday life. AI has invaded drug discovery, all fields of this process. AI is not the future, it's the present of drug discovery. In drug design, AI is used to predict the 3D structure of proteins. AI predicts the drug protein interactions, it determines drug activity, construct molecules, then all. In pharmacology, AI is used to design specific molecules as well as multi-target drugs. In chemists, as AI is able to design synthetic route, it's able to predict reaction yield, to clarify reaction mechanisms. AI is quite good to identify new therapeutic targets to all drugs, so-called drug repurposing. Drug repurposing means we find new indications for all drugs. Finally, AI is replace irreplaceable, meaning like predicting toxicity, bioavailability, anti-properties, physico-chemical properties. Here on the right side, I have listed some most popular AI platforms used in drug design, like the Swiss drug design system developed by the Swiss Institute of Bioinformatics, very freely accessible, very good working platform. Of course, AlphaFault, Google's deep mic platform for prediction of 3D structure of proteins. Bio-symmetrics contingent predicts drug target interactions and multi-faction. Merck's Cynthia proposes possible synthesizing routes for a given compound. Cyclicous ligand express finds possible protein targets for a given small molecule. And many others constantly appear in every day. Oops, sorry. Oops, what happened? Attention, I would like to pay to AstraZeneca's artificial platform for the novel design of small molecules named Reinvent. The platform is able to generate small molecules so that satisfy a diverse set of criteria to define a set of criteria and the platform generates a set of molecules corresponding criteria. So the most interesting thing here for us is the brain behind the platform. So this is a TANAS patronus, one of our ex PhD students is now an associate principal scientist in AstraZeneca. He designed and developed this platform. Some of you know him, he has presented several times in our preschools before. I suggested to the organizers next time to invite him to give him a presentation in the next preschool. How the platform works is shown in this presentation at the given link. I recommend you to have a look. It is quite interesting. Well, data visualization also is of great importance for drug design. He also have been several great achievements. All of them were possible because of the invention of molecular mechanics. The molecular mechanics is based on a set of empirical energy functions called a force field. The force field is able to describe the total energy of the system. And it consists of several terms like bond strength energy term, bond angle energy, torsional energy, different types of interactions like van der Waals, electrostatic hydrogen bond formation. All these interactions are considered in the creation of molecular mechanics. Based on molecular mechanics, they are very useful. And widely used techniques for visualization of drug receptor interactions have been developed. These are dynamics, molecular docking, virtual screening. Molecular dynamics is a method for simulating the movements of more than the interactions in different based on the force field. And it provides information so that it could not end by any experimental method for 3D structure resolution like the crystallography, spectroscopy and microscopy. These methods give a static picture of the molecule and the complex of molecules. So they take snapshots. And it makes a movie. It records the movement of molecules. It shows how the molecules connect each other, interact each other. Here's the MD simulation between a human leukocyte protein in red, this given, with a peptide bound into the binding site given in blue. This simulation was generated in our lab for as a part of a study on peptide binding prediction to HLA proteins. Actually, in Bulgaria, we have good traditions in molecular dynamics. There are several all-time schools and experienced scientists in this field. And you will see and hear them over the next few days. Molecular docking predicts the amount of binding between two molecules, the mutual orientation of the molecules, the conformation of each molecule and estimates the energy of the complex. As lower is the energy, as more stable is the complex. Virtual screening is a method for searching a structure with specific elements, so-called pharmacophore. So this is the pharmacophore search method, one of the ligand-based methods in drug design. Virtual screening can also search for a structure able to bind to a specific binding site. So this is the docking-based virtual screening, which is one of the structure-based methods in drug design. And the very top of the automation and artificial intelligence in drug design are these two robots, Adam and Eve, who can by Rosking and Stephen Oliver from the University of Manchester. Adam was constructed to microbiological experiments, analyze the results itself, define hypothesis itself, design experiments to test this hypothesis itself and repeat this cycle anti-validated hypothesis derived. It is a more advanced robot. It works in the field of drug discovery. It is able, or maybe I have to say she, she is able to screen experimentally thousands of compounds per day to discover specific hits, to engineer a specific cell line to test the hits and then to optimize their structures to deliver lead compounds. So what is the takeaway from the history of drug design? The presence and thoughts will be the future of drug design. So it's that any advance in technology finds immediately its application in medicine in pharmacy, in drug discovery and development, the priorities of any new technology. Any investment in drug design is worthwhile because as better is designed a given drug candidate during the experimental stage, the stage of drug design, as less likely is for this drug to fail in the late stages where the tests are more expensive, especially in the clinical trials. The future drug design relies on artificial intelligence, no doubt about that. The ultimate goal of the future drug design is to be able to design and develop a specific, non-toxic, effective and patient-powered drug for a few hours time only. So this is the ultimate goal. So I'm not sure when this goal will be achieved, but I'm sure that this is an achievable goal. So wait and see, or at least those of you who are alive. Well, in the end of my lecture, I'll show you some results from our recent drug design studies. Here's some time ago, we performed a docking by spiritualists screening on a set from Zing database, containing more than 6 million small molecules in order to identify novel hits as acetylcholinesterase inhibitors. Acetylcholinesterase inhibitors are involved in the treatment of neurodegenerative diseases, like Alzheimer's disease. They treat the disease, they improve the symptoms, they do not cure the disease. The top 10 best court hits from the virtual screening were synthesized here, there and tested in vitro for affinity to the enzyme, for neuroduxicity, for intestinal and blood-brain barrier permeability. The affinity to the enzyme was measured by isothermal filtration coloring, calorimetry, nine of the compounds here, there's nine of the 10 compounds are good binders. Significantly better than galantamine. Galantamine is a classical acetylcholinesterase inhibitor. Eight of them are non-toxic in concentrations up to 100 micromoles through in the cell line neuro2A, all of them are able to cross the barrier and the gastrointestinal. These two experiments were performed by parallel artificial membrane permeability say, so-called the PAMBA tests. So in general, by this technique, virtual docking-based virtual screening, we discovered seven new hits as potential acetylcholinesterase inhibitors, non-toxic and permeable through the intestine and through the blood-brain barrier. The other study that I would like to show you is more theoretical by MD simulation. We clarified the mechanism of anti-aggregant activity of curcumin on amyloid better aggregation. The amyloid plugs are one of the hallmark of Alzheimer's disease. The formation starts with aggregation of monomers into a nucleus. The monomers here in the Nea range and form a fibro and this fibro increases in length when several fibro aggregates form the plug. We simulated the interactions of two molecules which are meters of amyloid aggregation with a set of 12 amyloid peptides. So these two molecules were curcumin and ferulic acid. Curcumin is considered as a strong inhibitor of the aggregation with IC50, 0.8 micromoles. Ferulic acid is a weak inhibitor with T5.5. These two molecules are very similar. Curcumin consists of two ferulic acid. We model five systems, one consisting of 12 amyloid monomers, second one consisting of 12 monomers and 12 randomly positioned curcumin molecules. The third one consists of these 12 monomers and 36 randomly positioned curcumin molecules and other two systems containing 12 and 36 ferulic acids. The systems were solvated in isotonic solution and energy minimized, heated to 300 Kelvin degrees, equilibrated and simulated for one microsecond at constant pressure and temperature, classical anti-protocol. On the right side are presented the system at the end of the simulation. So you can see that the curcumin molecules are bound inside the amyloid nucleus while most of the acids are spread around the core. So some of them are bound inside, but most of them are outside the core. We analyze the trajectories of the complexes and these results show how curcumin affects the amyloid aggregation, what it is doing on the amyloids. First of all, it stabilizes the backbone fluctuations here are given their MSF values. In the presence of curcumin, 12 or 36, the fluctuations of the backbone atoms of the amyloid peptides decreases. They are less flexible in the presence of curcumin. The surface, the solvent accessible surface area is as a parameter increases in the presence of curcumin, especially at high concentration, which means that the nucleus is getting bigger, is getting bigger because curcumin bounce inside increases the volume of the nucleus. The number of non-native contacts between its decreases in the presence of curcumin decreases the number of hydrogen bonds between the peptides in the presence of curcumin. So in general, we concluded that curcumin affect the primary nucleation. It's intercalating between the amyloid peptides in the nucleus, increases the site of the core and prevents the peptides to form contacts between them and to form hydrogen bonds. So this is the explanation, the in silico explanation of the mechanism of anti-aggregant activity of curcumin. Well, at the end, I would like to thank the organizations that funding our studies. I thank the Center of Excellence in Attic Center, ICT, the National Center for High Performance and Distributing and the Bulgarian National Science Fund. Thank you for your attention. So we have time for some questions. Thank you, Professor Vikinova for the excellent overview and for the critical assessment. I'm ready to start discovering the development. I think we have a couple of minutes for several questions. So all participants are welcome to have some questions in the chat if they have. In the meantime, there was one question already at the beginning of the lecture. I'm going to read it to you. In the case of the Corona vaccines, were all phases of clinical testing fulfilled? Yes, definitely. Without fulfilling all the stages, they will not be approved. The fact is that all these investigations, all these studies have been made in very short time. But this short time was not because some of the stages were skipped. The shortness comes from decreasing the amount of administrative regulations which take a huge amount, the amount of the time for the clinical trials is spent in administrative regulations. So this was reduced, no doubt about it. All of these vaccines are tested following all the rules of drug design and discovery. Maybe you can see the chat for the next question. If not, and why drugs are tested on animals since animals are different. Some good drugs may be rejected, drugs good for animals, maybe more in case of humans. That is a good question, a very good question. But for the time of developing drug design and development of drugs, animal models are not avoidable. So in silico methods is an alternative to replace the animal models. And there are many in silico models already replacing the animal toxicity tested in animal models. So this is the way. Yet in silico, we will reduce the animal tests. Okay, and maybe the final question before we proceed. Could you comment and a little bit about the protocol for virtuals you ran using docking? For example, what docking approach did you use as to screen the whole thing? I think in the agenda, there is a special lecture on virtual screening and the lack of docking. So the details you can hear and see there. It's a long protocol, there are different protocols. And yeah, you will study how this is doing.