FragRep: A Web Server for Structure-Based Drug Design by Fragment Replacement

Food-derived DPP4 inhibitors: Drug discovery based on high-throughput virtual screening and deep learning

Dipeptidyl peptidase-4 (DPP-4) is a critical target for the treatment of type 2 diabetes. This study outlines the development of six compounds derived from food sources and modified to create promising candidates for the treatment of diabetes. These compounds were identified through a combination of virtual screening, deep learning algorithms, ADMET characterization assessment, and molecular dynamics simulations. Furthermore, a taste prediction model was used to assess the flavor of these DPP-4 inhibiting compounds. After thorough evaluation, we concluded that the six food-derived DPP-4 inhibitors identified have significant potential for therapeutic success. This study has greatly contributed to the discovery of novel dietary supplements for the management of type 2 diabetes.

Fast and Accurate Prediction of Tautomer Ratios in Aqueous Solution via a Siamese Neural Network

Tautomerization plays a critical role in chemical and biological processes, influencing molecular stability, reactivity, biological activity, and ADME-Tox properties. Many drug-like molecules exist in multiple tautomeric states in aqueous solution, complicating the study of protein-ligand interactions. Rapid and accurate prediction of tautomer ratios and identification of predominant species are therefore crucial in computational drug discovery. In this study, we introduce sPhysNet-Taut, a deep learning model fine-tuned on experimental data using a Siamese neural network architecture. This model directly predicts tautomer ratios in aqueous solution based on MMFF94-optimized molecular geometries. On experimental test sets, sPhysNet-Taut achieves state-of-the-art performance with root-mean-square error (RMSE) of 1.9 kcal/mol on the 100-tautomers set and 1.0 kcal/mol on the SAMPL2 challenge, outperforming all other methods. It also provides superior ranking power for tautomer pairs on multiple test sets. Our results demonstrate that fine-tuning on experimental data significantly enhances model performance compared to training from scratch. This work not only offers a valuable deep learning model for predicting tautomer ratios but also presents a protocol for modeling pairwise data. To promote usability, we have developed an accessible tool that predicts stable tautomeric states in aqueous solution by enumerating all possible tautomeric states and ranking them using our model. The source code and web server are freely accessible at https://github.com/xiaolinpan/sPhysNet-Taut and https://yzhang.hpc.nyu.edu/tautomer.

Integrating structure-guided and fragment-based inhibitor design to combat bedaquiline resistant Mycobacterium tuberculosis: a molecular dynamics study

The first FDA approved, MDR-TB inhibitory drug bedaquiline (BDQ), entraps the c-ring of the proton-translocating F0 region of enzyme ATP synthase of Mycobacterium tuberculosis, thus obstructing successive ATP production. Present-day BDQ-resistance has been associated with cardiotoxicity and mutation(s) in the atpE gene encoding the c subunit of ATP synthase (ATPc) generating five distinct ATPc mutants: Ala63→Pro, Ile66→Met, Asp28→Gly, Asp28→Val and Glu61→Asp. We created three discrete libraries, first by repurposing bedaquiline via scaffold hopping approach, second one having natural plant compounds and the third being experimentally derived analogues of BDQ to identify one drug candidate that can inhibit ATPc activity more efficiently with less toxic properties. For this purpose, we adopted techniques like molecular dynamics simulation, virtual screening, PCA, DCCM, binding affinity analysis to gauge structure-function relationship of the L136-ATPc complexes. L136 was found to induce a distinguishable conformational change in the bound ATPc which captivated the c9 rotor ring. L136 displays a binding free energy of -57.294, -59.027, -57.273, -58.726, -55.889 and -58.651 kcal/mol for ATPc_WT and the five respective mutants. The pIC50 value for the L136 ligand for the same proteins was unveiled to be 6.760, 7.285, 6.898, 7.222, 6.987 and 7.687. Moreover, L136 exhibited a strong ADMET profile. Furthermore, we discovered that the change in the hydrophobic platform in ATPc mutants hinders BDQ binding, which is overcome by L136, ensuring efficient binding and providing an assessment of L136's mechanism of ATPc inhibition. L136 provides a scope for in vivo test for future clinical drug trials.

Deep Lead Optimization: Leveraging Generative AI for Structural Modification

The integration of deep learning-based molecular generation models into drug discovery has garnered significant attention for its potential to expedite the development process. Central to this is lead optimization, a critical phase where existing molecules are refined into viable drug candidates. As various methods for deep lead optimization continue to emerge, it is essential to classify these approaches more clearly. We categorize lead optimization methods into two main types: goal-directed and structure-directed. Our focus is on structure-directed optimization, which, while highly relevant to practical applications, is less explored compared to goal-directed methods. Through a systematic review of conventional computational approaches, we identify four tasks specific to structure-directed optimization: fragment replacement, linker design, scaffold hopping, and side-chain decoration. We discuss the motivations, training data construction, and current developments for each of these tasks. Additionally, we use classical optimization taxonomy to classify both goal-directed and structure-directed methods, highlighting their challenges and future development prospects. Finally, we propose a reference protocol for experimental chemists to effectively utilize Generative AI (GenAI)-based tools in structural modification tasks, bridging the gap between methodological advancements and practical applications.

Scaffold hopping approaches for dual-target antitumor drug discovery: opportunities and challenges

INTRODUCTION

Scaffold hopping has emerged as a practical tactic to enrich the synthetic bank of small molecule antitumor agents. Specifically, it enables the chemist to refine the lead compound's pharmacodynamic, pharmacokinetic, and physiochemical properties. Scaffold hopping opens up fresh molecular territory beyond established patented chemical domains.

AREA COVERED

The authors present the scaffold hopping-based drug design strategies for dual inhibitory antitumor structural templates in this review. Minor modifications, structure rigidification and simplification (ring-closing and opening), and complete structural overhauls were the strategies employed by the medicinal chemist to generate a library of bifunctional inhibitors. In addition, the review presents an overview of the computational methods of scaffold hopping (software and programs) and organopalladium catalysis leveraged for the synthesis of templates designed via scaffold hopping.

EXPERT OPINION

The medicinal chemist has demonstrated remarkable prowess in furnishing dual inhibitory antitumor chemical architectures. Scaffold hopping-based drug design strategies have yielded a plethora of pharmacodynamically superior dual modulatory antitumor agents. An integrated approach involving computational advancements, synthetic methodology advancements, and conventional drug design strategies is required to increase the number of scaffold-hopping-assisted drug discovery campaigns.

Chemical analogue based drug design for cancer treatment targeting PI3K: integrating machine learning and molecular modeling

Cancer is a generic term for a group of disorders defined by uncontrolled cell growth and the potential to invade or spread to other parts of the body. Gene and epigenetic alterations disrupt normal cellular control, leading to abnormal cell proliferation, resistance to cell death, blood vessel development, and metastasis (spread to other organs). One of the several routes that play an important role in the development and progression of cancer is the phosphoinositide 3-kinase (PI3K) signaling pathway. Moreover, the gene PIK3CG encodes the catalytic subunit gamma (p110γ) of phosphoinositide 3-kinase (PI3Kγ), a member of the PI3K family. Therefore, in this study, PIK3CG was targeted to inhibit cancer by identifying a novel inhibitor through computational methods. The study screened 1015 chemical fragments against PIK3CG using machine learning-based binding estimation and docking to select the potential compounds. Later, the analogues were generated from the selected hits, and 414 analogues were selected, which were further screened, and as most potential candidates, three compounds were obtained: (a) 84,332, 190,213, and 885,387. The protein-ligand complex's stability and flexibility were then investigated by dynamic modeling. The 100 ns simulation revealed that 885,387 exhibited the steadiest deviation and constant creation of hydrogen bonds. Compared to the other compounds, 885,387 demonstrated a superior binding free energy (ΔG = -18.80 kcal/mol) with the protein when the MM/GBSA technique was used. The study determined that 885,387 showed significant therapeutic potential and justifies further experimental investigation as a possible inhibitor of the PIK3CG target implicated in cancer.

A novel compound to overcome influenza drug resistance in endonuclease inhibitors

Influenza is a seasonal respiratory illness that kills hundreds of thousands of people every year. Currently, neuraminidase inhibitors and endonuclease inhibitors are used in antiviral therapy. However, both drug types have encountered drug-resistant influenza strains in the human body. Fortunately, there is currently no resistance to endonuclease inhibitors in wild strains of influenza. We obtained the molecules with endonuclease inhibitor activity independent of the existing drug-resistant strains through computer-aided drug design, and we hope the obtained results can lay a theoretical foundation for the development of high-activity endonucleases. Combining a traditional fragment-based drug discovery approach with AI-directed fragment growth, we selected and designed a compound that achieved antiviral activity on drug-resistant strains by avoiding mutable residues and drug-resistant residues. We predicted the related properties using an ADMET model. Finally, we obtained a compound similar to baloxavir in terms of binding free energy but not affected by baloxavir resistance.

BioisoIdentifier: an online free tool to investigate local structural replacements from PDB

Within the realm of contemporary medicinal chemistry, bioisosteres are empirically used to enhance potency and selectivity, improve adsorption, distribution, metabolism, excretion and toxicity profiles of drug candidates. It is believed that bioisosteric know-how may help bypass granted patents or generate novel intellectual property for commercialization. Beside the synthetic expertise, the drug discovery process also depends on efficient in silico tools. We hereby present BioisoIdentifier (BII), a web server aiming to uncover bioisosteric information for specific fragment. Using the Protein Data Bank as source, and specific substructures that the user attempt to surrogate as input, BII tries to find suitable fragments that fit well within the local protein active site. BII is a powerful computational tool that offers the ligand design ideas for bioisosteric replacing. For the validation of BII, catechol is conceived as model fragment attempted to be replaced, and many ideas are successfully offered. These outputs are hierarchically grouped according to structural similarity, and clustered based on unsupervised machine learning algorithms. In summary, we constructed a user-friendly interface to enable the viewing of top-ranking molecules for further experimental exploration. This makes BII a highly valuable tool for drug discovery. The BII web server is freely available to researchers and can be accessed at http://www.aifordrugs.cn/index/ . Scientific Contribution: By designing a more optimal computational process for mining bioisosteric replacements from the publicly accessible PDB database, then deployed on a web server for throughly free access for researchers. Additionally, machine learning methods are applied to cluster the bioisosteric replacements searched by the platform, making a scientific contribution to facilitate chemists' selection of appropriate bioisosteric replacements. The number of bioisosteric replacements obtained using BII is significantly larger than the currently available platforms, which expanding the search space for effective local structural replacements.

Structure-based virtual screening and fragment replacement to design novel inhibitors of Coxsackievirus A16 (CVA16)

Numerous studies have shown that hand, foot and mouth disease (HFMD) pathogen Coxsackievirus A16 (CVA16) can also cause severe neurological complications and even death. Currently, there is no effective drugs and vaccines for CVA16. Therefore, developing a drug against CVA16 has become critical. In this study, we conducted two strategies-virtual screening (VS) and fragment replacement to obtain better candidates than the known drug GPP3. Through VS, 37 candidate drugs were screened (exhibiting a lower binding energy than GPP3). After toxicity evaluations, we obtained five candidates, analysed their binding modes and found that four candidates could enter the binding pocket of the GPP3. In another strategy, we analysed the four positions in GPP3 structures by the FragRep webserver and obtained a large number of candidates after replacing different functional groups, we obtained eight candidates (that target the four positions above) with the combined binding score and synthetic accessibility evaluations. AMDock software was uniformly utilized to perform molecular docking evaluation of the candidates with binding activity superior to that of GPP3. Finally, the selected top three molecules (Lapatinib, B001 and C001) and its interaction with CAV16 were validated by molecular dynamics (MD) simulation. The results indicated that all three molecules retain inside the pocket of CAV16 receptor throughout the simulation process, and he binding energy calculated from the MD simulation trajectories also support the strong affinity of the top three molecules towards the CVA16. These results will provide new ideas and technical guidance for designing and applying CVA16 therapeutics.Communicated by Ramaswamy H. Sarma.

BMaps: A Web Application for Fragment-Based Drug Design and Compound Binding Evaluation

Fragment-based drug design uses data about where, and how strongly, small chemical fragments bind to proteins, to assemble new drug molecules. Over the past decade, we have been successfully using fragment data, derived from thermodynamically rigorous Monte Carlo fragment-protein binding simulations, in dozens of preclinical drug programs. However, this approach has not been available to the broader research community because of the cost and complexity of doing simulations and using design tools. We have developed a web application, called BMaps, to make fragment-based drug design widely available with greatly simplified user interfaces. BMaps provides access to a large repository (>550) of proteins with 100s of precomputed fragment maps, druggable hot spots, and high-quality water maps. Users can also employ their own structures or those from the Protein Data Bank and AlphaFold DB. Multigigabyte data sets are searched to find fragments in bondable orientations, ranked by a binding-free energy metric. The designers use this to select modifications that improve affinity and other properties. BMaps is unique in combining conventional tools such as docking and energy minimization with fragment-based design, in a very easy to use and automated web application. The service is available at https://www.boltzmannmaps.com.

Identified Isosteric Replacements of Ligands' Glycosyl Domain by Data Mining

Biologically equivalent replacements of key moieties in molecules rationalize scaffold hopping, patent busting, or R-group enumeration. Yet, this information may depend upon the expert-defined space, and might be subjective and biased toward the chemistries they get used to. Most importantly, these practices are often informatively incomplete since they are often compromised by a try-and-error cycle, and although they depict what kind of substructures are suitable for the replacement occurrence, they fail to explain the driving forces to support such interchanges. The protein data bank (PDB) encodes a receptor-ligand interaction pattern and could be an optional source to mine structural surrogates. However, manual decoding of PDB has become almost impossible and redundant to excavate the bioisosteric know-how. Therefore, a text parsing workflow has been developed to automatically extract the local structural replacement of a specific structure from PDB by finding spatial and steric interaction overlaps between the fragments in endogenous ligands and particular ligand fragments. Taking the glycosyl domain for instance, a total of 49 520 replacements that overlap on nucleotide ribose were identified and categorized based on their SMILE codes. A predominately ring system, such as aliphatic and aromatic rings, was observed; yet, amide and sulfonamide replacements also occur. We believe these findings may enlighten medicinal chemists on the structure design and optimization of ligands using the bioisosteric replacement strategy.

Fragment-Based Approaches Identified Tecovirimat-Competitive Novel Drug Candidate for Targeting the F13 Protein of the Monkeypox Virus

Monkeypox is a serious public health issue in tropical and subtropical areas. Antivirals that target monkeypox proteins might lead to more effective and efficient therapy. The F13 protein is essential for the growth and maturation of the monkeypox virus. F13 inhibition might be a viable therapeutic target for monkeypox. The in silico fragment-based drug discovery method for developing antivirals may provide novel therapeutic options. In this study, we generated 800 compounds based on tecovirimat, an FDA-approved drug that is efficacious at nanomolar quantities against monkeypox. These compounds were evaluated to identify the most promising fragments based on binding affinity and pharmacological characteristics. The top hits from the chemical screening were docked into the active site of the F13 protein. Molecular dynamics simulations were performed on the top two probable new candidates from molecular docking. The ligand-enzyme interaction analysis revealed that the C2 ligand had lower binding free energy than the standard ligand tecovirimat. Water bridges, among other interactions, were shown to stabilize the C2 molecule. Conformational transitions and secondary structure changes in F13 protein upon C2 binding show more native three-dimensional folding of the protein. Prediction of pharmacological properties revealed that compound C2 may be promising as a drug candidate for monkeypox fever. However, additional in vitro and in vivo testing is required for validation.

Exploration of potential inhibitors for SARS-CoV-2 Mpro considering its mutants via structure-based drug design, molecular docking, MD simulations, MM/PBSA, and DFT calculations

The main protease (Mpro) of SARS-COV-2 plays a vital role in the viral life cycle and pathogenicity. Due to its specific attributes, this 3-chymotrypsin like protease can be a reliable target for the drug design to combat COVID-19. Since the advent of COVID-19, Mpro has undergone many mutations. Here, the impact of 10 mutations based on their frequency and five more based on their proximity to the active site was investigated. For comparison purposes, the docking process was also performed against the Mpros of SARS-COV and MERS-COV. Four inhibitors with the highest docking score (11b, α-ketoamide 13b, Nelfinavir, and PF-07321332) were selected for the structure-based ligand design via fragment replacement, and around 2000 new compounds were thus obtained. After the screening of these new compounds, the pharmacokinetic properties of the best ones were predicted. In the last step, comparative molecular dynamics (MD) simulations, molecular mechanics Poisson-Boltzmann surface area calculations (MM/PBSA), and density functional theory calculations were performed. Among the 2000 newly designed compounds, three of them (NE1, NE2, and NE3), which were obtained by modifications of Nelfinavir, showed the highest affinity against all the Mpro targets. Together, NE1 compound is the best candidate for follow-up Mpro inhibition and drug development studies.

Incorporating Target-Specific Pharmacophoric Information into Deep Generative Models for Fragment Elaboration

Despite recent interest in deep generative models for scaffold elaboration, their applicability to fragment-to-lead campaigns has so far been limited. This is primarily due to their inability to account for local protein structure or a user's design hypothesis. We propose a novel method for fragment elaboration, STRIFE, that overcomes these issues. STRIFE takes as input fragment hotspot maps (FHMs) extracted from a protein target and processes them to provide meaningful and interpretable structural information to its generative model, which in turn is able to rapidly generate elaborations with complementary pharmacophores to the protein. In a large-scale evaluation, STRIFE outperforms existing, structure-unaware, fragment elaboration methods in proposing highly ligand-efficient elaborations. In addition to automatically extracting pharmacophoric information from a protein target's FHM, STRIFE optionally allows the user to specify their own design hypotheses.

FragExplorer: GRID-Based Fragment Growing and Replacement

Understanding which chemical modifications can be made to known ligands is a key aspect of structure-based drug design and one that was pioneered by the software GRID. We developed FragExplorer with the explicit aim of showing GRID users which fragments would best match the GRID molecular interaction fields in a protein binding site, given a bound ligand as a starting point. Users can grow ligands or replace existing moieties; the R-Group Exploration mode identifies all potential R-Groups and searches for replacements automatically; the Scaffold Exploration mode does the same for all potential scaffolds. For a ligand with three points of variation, R-Group Exploration will typically explore a chemical space of 10¹⁶ potential molecules; including Scaffold Exploration increases this to 10²². FragExplorer was designed to be integrated within an interactive 3D Editor/Designer; therefore, the speed of computation was an important consideration; a typical fragment search takes 20 seconds. In a fragment reprediction test, FragExplorer demonstrates an overall fragment retrieval rate of 55%, increasing to 69% for smaller fragments. At a 90% substructural match, the retrieval rate increases to ∼80%. We also show how the approach could have been used to hop from olmesartan to azilsartan or to optimize a p38 MAP kinase lead to a compound that bears similarity to a known nanomolar inhibitor.

Efficient Hit-to-Lead Searching of Kinase Inhibitor Chemical Space via Computational Fragment Merging

In early-stage drug discovery, the hit-to-lead optimization (or "hit expansion") stage entails starting from a newly identified active compound and improving its potency or other properties. Traditionally, this process relies on synthesizing and evaluating a series of analogues to build up structure-activity relationships. Here, we describe a computational strategy focused on kinase inhibitors, intended to expedite the process of identifying analogues with improved potency. Our protocol begins from an inhibitor of the target kinase and generalizes the synthetic route used to access it. By searching for commercially available replacements for the individual building blocks used to make the parent inhibitor, we compile an enumerated library of compounds that can be accessed using the same chemical transformations; these huge libraries can exceed many millions─or billions─of compounds. Because the resulting libraries are much too large for explicit virtual screening, we instead consider alternate approaches to identify the top-scoring compounds. We find that contributions from individual substituents are well described by a pairwise additivity approximation, provided that the corresponding fragments position their shared core in precisely the same way relative to the binding site. This key insight allows us to determine which fragments are suitable for merging into single new compounds and which are not. Further, the use of pairwise approximation allows interaction energies to be assigned to each compound in the library without the need for any further structure-based modeling: interaction energies instead can be reliably estimated from the energies of the component fragments, and the reduced computational requirements allow for flexible energy minimizations that allow the kinase to respond to each substitution. We demonstrate this protocol using libraries built from six representative kinase inhibitors drawn from the literature, which target five different kinases: CDK9, CHK1, CDK2, EGFR^T790M, and ACK1. In each example, the enumerated library includes additional analogues reported by the original study to have activity, and these analogues are successfully prioritized within the library. We envision that the insights from this work can facilitate the rapid assembly and screening of increasingly large libraries for focused hit-to-lead optimization. To enable adoption of these methods and to encourage further analyses, we disseminate the computational tools needed to deploy this protocol.

SeamDock: An Interactive and Collaborative Online Docking Resource to Assist Small Compound Molecular Docking

In silico assessment of protein receptor interactions with small ligands is now part of the standard pipeline for drug discovery, and numerous tools and protocols have been developed for this purpose. With the SeamDock web server, we propose a new approach to facilitate access to small molecule docking for nonspecialists, including students. The SeamDock online service integrates different docking tools in a common framework that allows ligand global and/or local docking and a hierarchical approach combining the two for easy interaction site identification. This service does not require advanced computer knowledge, and it works without the installation of any programs with the exception of a common web browser. The use of the Seamless framework linking the RPBS calculation server to the user’s browser allows the user to navigate smoothly and interactively on the SeamDock web page. A major effort has been put into the 3D visualization of ligand, receptor, and docking poses and their interactions with the receptor. The advanced visualization features combined with the seamless library allow a user to share with an unlimited number of collaborators, a docking session, and its full visualization states. As a result, SeamDock can be seen as a free, simple, didactic, evolving online docking resource best suited for education and training.

DeepFrag: a deep convolutional neural network for fragment-based lead optimization

Machine learning has been increasingly applied to the field of computer-aided drug discovery in recent years, leading to notable advances in binding-affinity prediction, virtual screening, and QSAR. Surprisingly, it is less often applied to lead optimization, the process of identifying chemical fragments that might be added to a known ligand to improve its binding affinity. We here describe a deep convolutional neural network that predicts appropriate fragments given the structure of a receptor/ligand complex. In an independent benchmark of known ligands with missing (deleted) fragments, our DeepFrag model selected the known (correct) fragment from a set over 6500 about 58% of the time. Even when the known/correct fragment was not selected, the top fragment was often chemically similar and may well represent a valid substitution. We release our trained DeepFrag model and associated software under the terms of the Apache License, Version 2.0.