Evaluating BindCraft for Generative Design of High-AffinityPeptides

Introduction Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), remains a major global health concern. It spreads through airborne droplets and has a high mortality rate, particularly without treatment. Drug resistance is rising, with treatments against multidrug-resistant TB (MDR-TB) showing poor treatment success rates. The thick, lipid-rich wall of Mtb and its slow growth reduce antibiotic effectiveness, requiring long treatment courses of 4–6 months. Current therapies often fail against drug-resistant strains, highlighting the urgent need for new, short-course treatment, affordable, and combination-friendly drugs. Areas covered Within this perspective, the authors review and comment on the following topics regarding Mtb resistance emergence and treatment strategies: i) Existing treatment ii) Resistance evolution in Mtb; iii) Key challenges in drug discovery targeting Mtb; iv) emerging strategies and recent advances in Mtb drug discovery, and v) Next-generation approaches. Literature was identified through a search of PubMed, google scholar, and web of science, from January 2010 to March 2025. Expert opinion AI is accelerating the discovery of bioavailable and safe preclinical drug candidates for TB, though data limitations and biological complexity remain challenging. Future progress requires multi-modal models, open-access datasets, and interdisciplinary collaboration.

The 2019 edition "New Challenges in Drug Discovery", one of the biennial meetings under the patronage of the SEQT and EFMC, was held at the Faculty of Pharmacy of the University of the Basque Country in Vitoria-Gasteiz from July 8 to 11, 2019. On this occasion, attendees from Spain as well as from five EU countries and three non-EU countries were spectators of many scientific highlights related with the main research areas for different diseases and with protein-protein interactions, analytical tools and methods in discovery and development of drugs and computer-assisted methods in medicinal chemistry. The interesting scientific program was also complemented by an entertaining social program: a gastronomic opening ceremony held at the Museum of Fine Arts of Alava, visits to the Medieval Old City of Gasteiz and a gala dinner in a typical restaurant nestled in a building from the XVth Century.

Forward genetic screens have revolutionized the field of target discovery, target deconvolution, and target validation, particularly in the oncology field. However, most approaches are limited to the interrogation of non-essential factors and identify gene networks rather than direct target structures. Therefore, the identification of druggable target structures and understanding an active molecules target space remains a challenge in drug discovery and development. To address this, we established a chemical mutagenesis approach in relevant cell types that allows entirely unbiased identification of small molecule targets at amino acid resolution, literally mapping compound-target interaction surfaces. Chemical mutagenesis randomly introduces single nucleotide variants into the entire genome resulting in cell populations which carry substitutions in statistically more than 90% of all amino acids. Challenging such mutagenized cell populations with a drug candidate of interest, followed by next generation sequencing and analysis, reveals direct drug target interactions, drug resistance mechanism, and potential off-targets or concealed efficacies. The alteration of single amino acids thereby allows interrogation of non-essential as well as most essential factors and increases screening resolution to the amino acid level. Several case studies will be shown. Citation Format: Moritz Horn, Georg Michlits, Andreia Rocha, Christoph Stelzer, Josef M. Penninger. High resolution target identification by unbiased forward genetics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 5547.

Simple SummaryDiscovery of anticancer targets with minimal side effects is a major challenge in drug discovery and development. This study developed a fuzzy optimization framework for identifying anticancer targets. The framework was applied to identify not only gene regulator targets but also metabolite- and reaction-centric targets. The computational results show that the combination of a carbon metabolism target and any one-target gene that participates in the sphingolipid, glycerophospholipid, nucleotide, cholesterol biosynthesis, or pentose phosphate pathways is more effective for treatment than one-target inhibition is, and a two-target combination of 5-FU and folate supplement can improve cell viability, reduce metabolic deviation, and reduce side effects of normal cells.The efficient discovery of anticancer targets with minimal side effects is a major challenge in drug discovery and development. Early prediction of side effects is key for reducing development costs, increasing drug efficacy, and increasing drug safety. This study developed a fuzzy optimization framework for Identifying AntiCancer Targets (IACT) using constraint-based models. Four objectives were established to evaluate the mortality of treated cancer cells and to minimize side effects causing toxicity-induced tumorigenesis on normal cells and smaller metabolic perturbations. Fuzzy set theory was applied to evaluate potential side effects and investigate the magnitude of metabolic deviations in perturbed cells compared with their normal counterparts. The framework was applied to identify not only gene regulator targets but also metabolite- and reaction-centric targets. A nested hybrid differential evolution algorithm with a hierarchical fitness function was applied to solve multilevel IACT problems. The results show that the combination of a carbon metabolism target and any one-target gene that participates in the sphingolipid, glycerophospholipid, nucleotide, cholesterol biosynthesis, or pentose phosphate pathways is more effective for treatment than one-target inhibition is. A clinical antimetabolite drug 5-fluorouracil (5-FU) has been used to inhibit synthesis of deoxythymidine-5-triphosphate for treatment of colorectal cancer. The computational results reveal that a two-target combination of 5-FU and a folate supplement can improve cell viability, reduce metabolic deviation, and reduce side effects of normal cells.

Accurately predicting drug sensitivity and understanding what is driving it are major challenges in drug discovery. Graphs are a natural framework for capturing diverse pharmacological data for efficacy predictions, thanks to their ability to integrate multimodal data and represent relationships such as gene-gene or drug-target interactions as edges. They have also had proven success across a range of other drug discovery tasks including repositioning and target identification. In this study, we sought to address the explainability challenges of drug response predictions. Recent developments in the field of Graph AI have led to improvements in interpretability mechanisms that highlight parts of a graph which are driving predictions. We have conducted a comprehensive review of multiple major approaches for tackling drug efficacy prediction using graph methods, benchmarking the performance and interpretability of these algorithms across indications. Methods: We assembled a combined dataset of GDSC1 and GDSC2 drug response data in cell lines, with multiomic cell line data and drug target and chemical structure data. We then applied graph-based approaches for the prediction of binarized IC50 on an indication-by-indication basis. Approach 1 involved the creation of a ‘GDSC knowledge graph’, where drug response and cell line ‘omic information is represented in an unweighted knowledge graph: cell lines are connected to genes expressed in them, drugs are connected to genes they target, and so on. We then used state-of-the-art graph embedding techniques to predict IC50 using paired drug and cell line embeddings. In Approach 2 we used a weighted knowledge graph instead, and generated embeddings using heterogeneous graph neural networks (HGNNs). In Approach 3, we modelled response prediction as a graph classification task, where one single graph captures one drug-cell line interaction. The graph classifier and HGNN models both have in-built interpretability mechanisms, including graph attention, that can signify the genes in the cell line which were most important for the eventual prediction. We can also integrate biomedical prior knowledge with all these models by capturing gene-pathway and gene-gene data in the graphs. Results: Our models outperformed benchmark models including DNNs and GBMs, and identified both established and novel response biomarkers in NSCLC cell lines (AUC = 0.94, Accuracy = 89%). We have also applied our models to Breast Cancer, Pancreatic Cancer, Colorectal Cancer and Haematological malignancies with similar predictive performance and explainability. Conclusions:Our graph analytical framework for response predictions showed better performance than benchmarking models and provided insights from explainability. This framework is easily extendable to response and ‘omic data from any disease model and patient studies. Citation Format: Jake Cohen-Setton, Krishna Bulusu, Jonathan Dry, Ben Sidders. Explainable AI: Graph machine learning for response prediction and biomarker discovery [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 902.

PROTAC technology: A new drug design for chemical biology with many challenges in drug discovery

Chapter 12 - Fundamental approaches of drug discovery

Chapter 25 - New challenges in drug discovery

Cardiovascular disease is a major cause of death worldwide, with high prevalence and morbidity. Recent advances in technology have reported that abnormalities in the gut microbiota are associated with a variety of diseases, including cardiovascular diseases. The gut microbiota is a complex ecosystem that plays an important role in maintaining host health. It has been reported that the imbalance of gut microbiota causes changes in the production of substances derived from gut bacteria, such as short-chain fatty acids, trimethylamine-N-oxide, and lipopolysaccharide, and contributes to the development of cardiovascular diseases. In the drug discovery, it is a promising approach to prevention and therapy of the cardiovascular disease to focus on the relation between gut and heart, such as gut bacteria. However, there are challenges that must be overcome to convert this approach into effective therapy. In this review, we focus on cardiovascular diseases, particularly atherosclerotic disease, heart failure, and atrial fibrillation, and discuss the relationship between gut bacteria and substances derived from gut bacteria in cardiovascular disease. We also discuss the challenges and potential of drug discovery targeting the gut-heart relationship for the treatment and prevention of cardiovascular disease.

Recently, artificial intelligence (AI) techniques have been increasingly used to overcome the challenges in drug discovery. Although traditional AI techniques generally have high accuracy rates, there may be difficulties in explaining the decision process and patterns. This can create difficulties in understanding and making sense of the outputs of algorithms used in drug discovery. Therefore, using explainable AI (XAI) techniques, the causes and consequences of the decision process are better understood. This can help further improve the drug discovery process and make the right decisions. To address this issue, Explainable Artificial Intelligence (XAI) emerged as a process and method that securely captures the results and outputs of machine learning (ML) and deep learning (DL) algorithms. Using techniques such as SHAP (SHApley Additive ExPlanations) and LIME (Locally Interpretable Model-Independent Explanations) has made the drug targeting phase clearer and more understandable. XAI methods are expected to reduce time and cost in future computational drug discovery studies. This review provides a comprehensive overview of XAI-based drug discovery and development prediction. XAI mechanisms to increase confidence in AI and modeling methods. The limitations and future directions of XAI in drug discovery are also discussed.

The second coming of epigenetic drugs: a more strategic and broader research framework could boost the development of new drugs to modify epigenetic factors and gene expression.

Identification of new hits is one of the biggest challenges in drug discovery. Creating a library of well-characterized drug-like compounds is a key step in this process. Our group has developed an in-house chemical library called the Medicinal and Biological Chemistry (MBC) library. This collection has been successfully used to start several medicinal chemistry programs and developed in an accumulation of more than 30 years of experience in drug design and discovery of new drugs for unmet diseases. It contains over 1000 compounds, mainly heterocyclic scaffolds. In this work, analysis of drug-like properties and comparative study with well-known libraries by using different computer software are presented here.

La chimie crée son objet. Cette faculté créatrice, semblable à celle de l’art lui-même, la distingue essentiellement des sciences naturelles et historiques. Marcelin Berthelot, La Synthèse chimique, Alcan, Paris, 1887 This famous quote from Marcelin Berthelot (its English translation is “Chemistry creates its object. This creative faculty, similar to that of art itself, distinguishes it essentially from natural and historical sciences.”) exemplifies like no other the ability of the chemist to create new molecules with novel structures and, following therefrom, novel properties. Because of this creative power, chemistry, and in particular synthetic chemistry, has been assigned multiple enabling roles and several of its sister disciplines have grown “chemical” branches such as “chemical physics”.1 Among these disciplines, “chemical biology” is a younger cousin. While the name was coined at least four decades ago,1 the current understanding of the term was shaped only within the last two decades. The field evolved from bioorganic chemistry, biochemistry, cell biology, and pharmacology, but synthetic organic chemists played a leading role in its inauguration. For instance, Stuart L. Schreiber and K. C. Nicolaou served as Editors of the journal Chemistry & Biology, “the first journal dedicated to the expanding intellectual area in which chemical approaches and biological disciplines overlap.” As they stress in their inaugural Editorial in the first issue: “Both of us started professional life as strict organic chemists, with little knowledge of biology and not much expectation that we would ever need to know any.” [2] Since then, the development and application of organic synthesis methodology to achieve a greater understanding of biology at the molecular level has emerged as one major area of research in chemical biology. For instance, labeling of biomolecules has greatly advanced in the last decades, based largely on the development of novel biomacromolecule synthesis and ligation techniques, which are often rooted in classical organic synthesis methods such as the Huisgen 1,3-dipolar cycloaddition and the Staudinger reaction. Among the various applications of organic synthesis methodology in chemical biology research, it is most likely that the use of small-molecule probes as tools for unraveling and manipulating the inner workings of the cell (chemical genetics) today is commonly associated with the term “chemical biology”. While major studies and efforts have been made during the last decades to fill the chemical toolbox required to meet this daunting task and to equip the chemical biologist with the right “tools” in the struggle to decipher the secrets of nature, this endeavor has only just begun. Notably, although large compound libraries are commercially available these days, their structural complexity and diversity remain fairly limited, and in high-content assays, their performance often leaves room for substantial improvement. Higher structural complexity and incorporation of stereogenic centers often positively correlate with bioactivity, thus calling for the synthesis and application of complex compound classes in chemical biology research that expand the currently accessible tool and probe candidates to novel scaffold classes. This demand can only be met by the continuous introduction of novel synthesis methodology and the development of creative solutions to the problem of making increasingly complex compounds available with higher efficiency and practicability in the formats of compound-collection synthesis. Therefore, the synthesis of structurally and stereochemically complex molecular architectures is at the heart of chemical biology research. Chemical biology needs continuous input from organic synthesis, and organic synthesis may find challenging and unprecedented synthesis targets with an immediate application in the problems faced by chemical biologists: chemical biology and organic synthesis are brothers-in-arms! Ideally, to lend the brotherhood strength and to develop it to maximum impact, both chemical and biological expertise should be established under one roof, that is, within a given research group. As ideal as it would be, such an interdisciplinary team is hard to establish. Limitations arise on one hand from the different cultures of the two sciences and the core expertise of the leading scientists, who usually were trained and started their career in either chemistry or biology. On the other hand, establishing and operating a full chemistry and biology infrastructure is very cost-intensive, and funds on the required scale often are simply not available. Hence, only few groups worldwide can fulfill these requirements. Alternatively, collaboration between different research groups is necessary in chemical biology research, and, indeed, many of the best results obtained in this science represent multiteam efforts. If productive collaborations with mutual appreciation of the partners and their scientific contribution can be established, from a scientific point of view, this brotherhood may actually be the better approach to tackle demanding scientific problems. The combined expertise of the partners in chemistry and biology usually will allow deeper insights to be obtained and high-quality research in both sciences to be performed. This brotherhood may prove vital to yet another sector of science in the near future, that is, to drug discovery. Chemical biology is partly rooted in cell biology and pharmacology, and its repertoire of methods extents into small-molecule synthesis, determination of bioactivity, and identification and validation of small molecule cellular targets. If the small molecules employed in chemical biology research have druglike properties, and modulation of the activity of their cellular targets can be tied to a disease-modifying effect, the link to drug discovery is obvious. In fact, fully fledged chemical biology research programs have the potential to simultaneously produce novel insights into fundamental biological mechanisms, deliver new targets, and supply small-molecule modulators of target activity. Therefore, major challenges in drug discovery may inspire chemical biology research and by extension, organic synthesis endeavors. Conversely, the outcomes of a chemical biology investigation may fuel efforts in drug discovery. This alliance may prove instrumental as a key driver for future research in the pharmaceutical industry. Facing major challenges, pharmaceutical companies very recently have increased collaboration with academic institutions far beyond the occasional support of individual smaller projects (see, for example, reference 3). In so doing, the industry may be well advised to listen to its own opinion leaders. In June 2011, Mark Bunnage (Pfizer) wrote: “This change in model reflects the reality that the vast majority of the initial breakthroughs in target biology research occurs in the academic research environment. It is thus considered essential for pharmaceutical companies and their scientists to become better connected with the external research environment and develop a more extended network of partnerships and genuine collaborations with academia. …︁ It is thus essential for medicinal chemists in industry to increase their awareness of chemical biology approaches and build these into their armamentarium to enable drug discovery.” [4]—He is right! A successful and seminal example of such a fruitful collaboration between academia and industry is the Chemical Genomics Centre (CGC) of the Max Planck Society (Max-Planck Gesellschaft, MPG). The CGC was established in 2005 as a joint initiative of the MPG, Merck KGaA, Schering AG, Bayer CropScience AG, and Organon B.V. Research in the CGC is focused on challenging unsolved problems in chemistry and biology of major relevance to drug discovery, such as stabilization of protein–protein interactions by small molecules and the development of allosteric kinase inhibitors. Both the companies and the MPG funded independent research groups that developed the basic science and transferred it to the companies. If appointment of the group leaders to professorships and integration of the developed technologies into the internal project pipelines of the companies are accepted as stringent criteria for measuring success from both the academic and the industrial point of view, then the establishment of the CGC was a major success. Accordingly, after the first funding period of the CGC (2005–2010) the MPG and Merck KGaA, AstraZeneca AB, Boehringer Ingelheim Pharma GmbH & Co. KG, Bayer Pharma AG, and Bayer CropScience AG have established “CGC II”, and the first research group leaders have been appointed very recently. The success of the CGC and related initiatives suggests that it may be more than advisable to those engaged in drug discovery to take the final verses of Schiller’s poem “The Hostage” to heart (translation by Scott Horton):5 He gazed upon them long in amazement, And then spoke: “You have succeeded, You have turned my heart, In truth, fidelity is no idle delusion, So accept me also as your friend, I would be—grant me this request— The third in your band!”

The major challenge facing drug discovery research today is the lack of productivity as measured by the introduction of new molecular entities (NMEs) into therapy. Only 21 NMEs were approved by the U.S. Food and Drug Administration in 2008, the same level of productivity as the 1950's even though spending on drug discovery research is many times higher. The drug discovery process can take as long as 8-12 yrs between the initial synthesis of a drug candidate and commercialization, and it is costly requiring well over $1 billion on average for every marketed drug that enters clinical practice. We have founded the Pennsylvania Center for Drug Discovery (PCDD) at the Pennsylvania Biotechnology Center (PBC) in Doylestown, Pennsylvania USA. The PBC is a mixed use academic-industrial biotechnology facility with >220 total employees. We have developed programs to advance the mission of accelerating the translation of new basic discoveries into therapies suitable for human clinical evaluation. The PCDD drug discovery capability is built upon a network of non-profit research institutions and small biotechnology companies using industry- standard metrics for the identification of hits, leads and preclinical development compounds, risk analysis and development. The PCDD is also meant to serve as an international think tank to brainstorm ways of improving efficiencies and productivity in early drug discovery. Jobs are already being created in the companies associated with the PCDD, helping to reintegrate senior-level biomedical drug discovery researchers who have been displaced elsewhere due to industrial downsizing into the workforce.

The generation of three-dimensional (3D) molecules based on target structures represents a cutting-edge challenge in drug discovery. Many existing approaches often produce molecules with invalid configurations, unphysical conformations, suboptimal drug-like qualities, limited synthesizability, and require extensive generation times. To address these challenges, we present 3DSMILES-GPT, a fully language-model-driven framework for 3D molecular generation that utilizes tokens exclusively. We treat both two-dimensional (2D) and 3D molecular representations as linguistic expressions, combining them through full-dimensional representations and pre-training the model on a vast dataset encompassing tens of millions of drug-like molecules. This token-only approach enables the model to comprehensively understand the 2D and 3D characteristics of large-scale molecules. Subsequently, we fine-tune the model using pair-wise structural data of protein pockets and molecules, followed by reinforcement learning to further optimize the biophysical and chemical properties of the generated molecules. Experimental results demonstrate that 3DSMILES-GPT generates molecules that comprehensively outperform existing methods in terms of binding affinity, drug-likeness (QED), and synthetic accessibility score (SAS). Notably, it achieves a 33% enhancement in the quantitative estimation of QED, meanwhile the binding affinity estimated by Vina docking maintaining its state-of-the-art performance. The generation speed is remarkably fast, with the average time approximately 0.45 seconds per generation, representing a threefold increase over the fastest existing methods. This innovative 3DSMILES-GPT approach has the potential to positively impact the generation of 3D molecules in drug discovery.