Computer-aided Drug Discovery Research Articles

Computer-Aided Drug Discovery Approaches in the Identification of Natural Products against SARS-CoV-2: A Review.

The COVID-19 pandemic is raising a worldwide search for compounds that could act against the disease, mainly due to its mortality. With this objective, many researchers invested in the discovery and development of drugs of natural origin. To assist in this search, the potential of computational tools to reduce the time and cost of the entire process is known. Thus, this review aimed to identify how these tools have helped in the identification of natural products against SARS-CoV-2. For this purpose, a literature review was carried out with scientific articles with this proposal where it was possible to observe that different classes of primary and, mainly, secondary metabolites were evaluated against different molecular targets, mostly being enzymes and spike, using computational techniques, with emphasis on the use of molecular docking. However, it is noted that in silico evaluations still have much to contribute to the identification of an anti-SARS-CoV-2 substance, due to the vast chemical diversity of natural products, identification and use of different molecular targets and computational advancement.

Proteasome and PARP1 dual-target inhibitor for multiple myeloma: Fluzoparib

One of the current mainstream treatments for multiple myeloma (MM) is chemotherapy. However, due to the high clonal heterogeneity and genomic complexity of MM, single-target drugs have limited efficacy and are prone to drug resistance. Therefore, there is an urgent need to develop multi-target drugs against MM. We screened drugs that simultaneously inhibit poly(ADP-ribose) polymerase 1 (PARP1) and 20S proteasome through computer-aided drug discovery (CADD) techniques, and explored the binding mode and dynamic stability of selected inhibitor to proteasome through Molecular biology (MD) simulation method. Thus, the dual-target inhibition effect of fluzoparib was proposed for the first time, and the ability of dual-target inhibition and tumor killing was explored at the enzyme, cell and animal level, respectively. This provides a theoretical and experimental basis for exploring multi-target inhibitory drugs for cancers.

In silico Analysis of Berberine as a Potential Therapeutic Approach for Various Signalling Pathways Linked to Androgenetic Alopecia

Background: Alopecia is defined by a loss of hair density and is often considered a symptom of multiple illnesses, such as infection and inflammation. Objective: The molecular mechanisms underlying the hair-promoting effects include inhibition of 5α-reductase activity, reducing the binding affinity of Dihydrotestosterone (DHT) to androgen receptors, and decreasing/down-regulating TGF-β2 activity, which have a suggestive role in androgenetic alopecia. Finasteride and minoxidil are the approved non-surgical treatment alternatives for hair loss, but they cause side effects in patients. Therefore, bioactive phytoconstituents with multiple targets can be used to find novel, secure, and efficacious hair-promoting medicinal products. Methods: This study has carried out the in silico evaluation of berberine using various software. To find possible interactions between the 5α-reductase enzyme and Transforming Growth Factor-- beta 2 (TGF-β2), a critical protein involved in the human hair development cycle, computer-aided drug discovery was employed. Results: According to in silico studies, berberine has been found to bind well to the 4K7a and 6M2N binding sites. The drug has been found to adhere to Lipinski's rule of five, and its pharmacokinetic characteristics were noteworthy. Drug-likeness and Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) properties showed appreciable results. Furthermore, berberine showed docking scores of -8.4 (5α-reductase) and -7.1 (TGF-β2), which were significantly better than minoxidil (-4.8, -3.2). In general, the drug exhibited improved binding interactions, and the possible toxicity investigations provided very little basis for risk prediction. Conclusion: The current protocol has offered experimental support for berberine's possible therapeutic use in reducing male pattern baldness. Therefore, it can be a possible target for the therapy of androgenetic alopecia through the regulation of TGF-β2 and 5α-reductase activity.

Ligand Binding Affinity Prediction for Membrane Proteins with Alchemical Free Energy Calculation Methods.

Alchemical relative binding free energy (ΔΔG) calculations have shown high accuracy in predicting ligand binding affinity and have been used as important tools in computer-aided drug discovery and design. However, there has been limited research on the application of ΔΔG methods to membrane proteins despite the fact that these proteins represent a significant proportion of drug targets, play crucial roles in biological processes, and are implicated in numerous diseases. In this study, to predict the binding affinity of ligands to G protein-coupled receptors (GPCRs), we employed two ΔΔG calculation methods: thermodynamic integration (TI) with AMBER and the alchemical transfer method (AToM) with OpenMM. We calculated ΔΔG values for 53 transformations involving four class A GPCRs and evaluated the performance of AMBER-TI and AToM-OpenMM. In addition, we conducted tests using different numbers of windows and varying simulation times to achieve reliable ΔΔG results and to optimize resource utilization. Overall, both AMBER-TI and AToM-OpenMM show good agreement with the experimental data. Our results validate the applicability of AMBER-TI and AToM-OpenMM for optimization of lead compounds targeting membrane proteins.

Identifying High-Quality Leads among Screened Anticancerous Compounds Using SMILES Representations.

Cancer is a lethal disease that affects numerous people worldwide. Chemotherapy stands as one of the most effective treatment regimens to combat cancer. Nevertheless, anticancer drugs face a high failure rate due to safety and efficacy issues. Drug failure could be subdued by instigating drug leads with reduced toxicity and enhanced efficacy. Computer-aided drug discovery endorses drug leads in manoeuvring protein and ligand structures or representations. Simplified molecular input line entry system (SMILES) is a linear notation representing the three-dimensional structure of a molecule using symbols and alphanumeric characters. SMILES representation hoards rings and scaffold structures in its depiction. Mining ring and scaffold patterns from molecular SMILES would assist in ascertaining biological properties based on molecular patterns. Moreover, the emergence of artificial intelligence (AI) technologies would accelerate identification of efficient anticancer drug leads. AI algorithms proclaimed for their pattern recognition ability could be employed for identifying molecular patterns from SMILES representation, thereby enabling property prediction. Consequently, we developed a multilayer perceptron (MLP) model for the prediction of anticancer activity using SMILES of NCI-60 cancer growth inhibition data. Furthermore, the top 8 frequent scaffolds were identified on preliminary analysis of cancer growth inhibition data and ChEMBL drugs. The developed MLP model classified anticancer and nonanticancer compounds with a classification accuracy of 0.92. Also, benchmarking of the developed model with machine learning algorithms exhibited better performance of the MLP model.

Insights into the computer-aided drug design and discovery based on anthraquinone scaffold for cancer treatment: A systematic review.

In the search for better anticancer drugs, computer-aided drug design (CADD) techniques play an indispensable role in facilitating the lengthy and costly drug discovery process especially when natural products are involved. Anthraquinone is one of the most widely-recognized natural products with anticancer properties. This review aimed to systematically assess and synthesize evidence on the utilization of CADD techniques centered on the anthraquinone scaffold for cancer treatment. The conduct and reporting of this review were done in accordance to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) 2020 guideline. The protocol was registered in the "International prospective register of systematic reviews" database (PROSPERO: CRD42023432904) and also published recently. The search strategy was designed based on the combination of concept 1 "CADD or virtual screening", concept 2 "anthraquinone" and concept 3 "cancer". The search was executed in PubMed, Scopus, Web of Science and MedRxiv on 30 June 2023. Databases searching retrieved a total of 317 records. After deduplication and applying the eligibility criteria, the final review ended up with 32 articles in which 3 articles were found by citation searching. The CADD methods used in the studies were either structure-based alone (69%) or combined with ligand-based methods via parallel (9%) or sequential (22%) approaches. Molecular docking was performed in all studies, with Glide and AutoDock being the most popular commercial and public software used respectively. Protein data bank was used in most studies to retrieve the crystal structure of the targets of interest while the main ligand databases were PubChem and Zinc. The utilization of in-silico techniques has enabled a deeper dive into the structural, biological and pharmacological properties of anthraquinone derivatives, revealing their remarkable anticancer properties in an all-rounded fashion. By harnessing the power of computational tools and leveraging the natural diversity of anthraquinone compounds, researchers can expedite the development of better drugs to address the unmet medical needs in cancer treatment by improving the treatment outcome for cancer patients.

Computer-Aided Drug Discovery Approaches in the Identification of Anticancer Drugs from Natural Products: A Review.

Natural plant sources are essential in the development of several anticancer drugs, such as vincristine, vinblastine, vinorelbine, docetaxel, paclitaxel, camptothecin, etoposide, and teniposide. However, various chemotherapies fail due to adverse reactions, drug resistance, and target specificity. Researchers are now focusing on developing drugs that use natural compounds to overcome these issues. These drugs can affect multiple targets, have reduced adverse effects, and are effective against several cancer types. Developing a new drug is a highly complex, expensive, and time-consuming process. Traditional drug discovery methods take up to 15 years for a new medicine to enter the market and cost more than one billion USD. However, recent Computer Aided Drug Discovery (CADD) advancements have changed this situation. This paper aims to comprehensively describe the different CADD approaches in identifying anticancer drugs from natural products. Data from various sources, including Science Direct, Elsevier, NCBI, and Web of Science, are used in this review. In-silico techniques and optimization algorithms can provide versatile solutions in drug discovery ventures. The structure-based drug design technique is widely used to understand chemical constituents' molecular-level interactions and identify hit leads. This review will discuss the concept of CADD, in-silico tools, virtual screening in drug discovery, and the concept of natural products as anticancer therapies. Representative examples of molecules identified will also be provided.

Molecular property prediction based on graph structure learning.

Molecular property prediction (MPP) is a fundamental but challenging task in the computer-aided drug discovery process. More and more recent works employ different graph-based models for MPP, which have achieved considerable progress in improving prediction performance. However, current models often ignore relationships between molecules, which could be also helpful for MPP. For this sake, in this article we propose a graph structure learning (GSL) based MPP approach, called GSL-MPP. Specifically, we first apply graph neural network (GNN) over molecular graphs to extract molecular representations. Then, with molecular fingerprints, we construct a molecule similarity graph (MSG). Following that, we conduct GSL on the MSG, i.e. molecule-level GSL, to get the final molecular embeddings, which are the results of fuzing both GNN encoded molecular representations and the relationships among molecules. That is, combining both intra-molecule and inter-molecule information. Finally, we use these molecular embeddings to perform MPP. Extensive experiments on 10 various benchmark datasets show that our method could achieve state-of-the-art performance in most cases, especially on classification tasks. Further visualization studies also demonstrate the good molecular representations of our method. Source code is available at https://github.com/zby961104/GSL-MPP.

Hit discovery of potential CDK8 inhibitors and analysis of amino acid mutations for cancer therapy through computer-aided drug discovery

Cyclin-dependent kinase 8 (CDK8) has emerged as a promising target for inhibiting cancer cell function, intensifying efforts towards the development of CDK8 inhibitors as potential cancer therapeutics. Mutations in CDK8, a protein kinase, are also implicated as a primary factor associated with tumor formation. In this study, we identified potential inhibitors through virtual screening for CDK8 and single amino acid mutations in CDK8, namely D173A (Aspartate 173 mutate to Alanine), D189N (Aspartate 189 mutate to Asparagine), T196A (Threonine 196 mutate to Alanine) and T196D (Threonine 196 mutate to Aspartate). Four databases (CHEMBEL, ZINC, MCULE, and MolPort) containing 65,209,131 molecules have been searched to identify new inhibitors for CDK8 and its single mutations. In the first step, structure-based pharmacophore modeling in the Pharmit server was used to select the compounds to know the inhibitors. Then molecules with better predicted drug-like molecule properties were selected. The final filter used to select more effective inhibitors among the previously selected molecules was molecular docking. Finally, 13 hits for CDK8, 11 hits for D173A, 11 hits for D189N, 15 hits for T196A, and 12 hits for T196D were considered potential inhibitors. A majority of the virtual screening hits exhibited satisfactorily predict pharmacokinetic characteristics and toxicity properties.

GLDM: hit molecule generation with constrained graph latent diffusion model.

Discovering hit molecules with desired biological activity in a directed manner is a promising but profound task in computer-aided drug discovery. Inspired by recent generative AI approaches, particularly Diffusion Models (DM), we propose Graph Latent Diffusion Model (GLDM)-a latent DM that preserves both the effectiveness of autoencoders of compressing complex chemical data and the DM's capabilities of generating novel molecules. Specifically, we first develop an autoencoder to encode the molecular data into low-dimensional latent representations and then train the DM on the latent space to generate molecules inducing targeted biological activity defined by gene expression profiles. Manipulating DM in the latent space rather than the input space avoids complicated operations to map molecule decomposition and reconstruction to diffusion processes, and thus improves training efficiency. Experiments show that GLDM not only achieves outstanding performances on molecular generation benchmarks, but also generates samples with optimal chemical properties and potentials to induce desired biological activity.

Developing an Improved Cycle Architecture for AI-Based Generation of New Structures Aimed at Drug Discovery.

Drug discovery involves a crucial step of optimizing molecules with the desired structural groups. In the domain of computer-aided drug discovery, deep learning has emerged as a prominent technique in molecular modeling. Deep generative models, based on deep learning, play a crucial role in generating novel molecules when optimizing molecules. However, many existing molecular generative models have limitations as they solely process input information in a forward way. To overcome this limitation, we propose an improved generative model called BD-CycleGAN, which incorporates BiLSTM (bidirectional long short-term memory) and Mol-CycleGAN (molecular cycle generative adversarial network) to preserve the information of molecular input. To evaluate the proposed model, we assess its performance by analyzing the structural distribution and evaluation matrices of generated molecules in the process of structural transformation. The results demonstrate that the BD-CycleGAN model achieves a higher success rate and exhibits increased diversity in molecular generation. Furthermore, we demonstrate its application in molecular docking, where it successfully increases the docking score for the generated molecules. The proposed BD-CycleGAN architecture harnesses the power of deep learning to facilitate the generation of molecules with desired structural features, thus offering promising advancements in the field of drug discovery processes.

A dual diffusion model enables 3D molecule generation and lead optimization based on target pockets

Structure-based generative chemistry is essential in computer-aided drug discovery by exploring a vast chemical space to design ligands with high binding affinity for targets. However, traditional in silico methods are limited by computational inefficiency, while machine learning approaches face bottlenecks due to auto-regressive sampling. To address these concerns, we have developed a conditional deep generative model, PMDM, for 3D molecule generation fitting specified targets. PMDM consists of a conditional equivariant diffusion model with both local and global molecular dynamics, enabling PMDM to consider the conditioned protein information to generate molecules efficiently. The comprehensive experiments indicate that PMDM outperforms baseline models across multiple evaluation metrics. To evaluate the applications of PMDM under real drug design scenarios, we conduct lead compound optimization for SARS-CoV-2 main protease (Mpro) and Cyclin-dependent Kinase 2 (CDK2), respectively. The selected lead optimization molecules are synthesized and evaluated for their in-vitro activities against CDK2, displaying improved CDK2 activity.

GxVAEs: Two Joint VAEs Generate Hit Molecules from Gene Expression Profiles

The de novo generation of hit-like molecules that show bioactivity and drug-likeness is an important task in computer-aided drug discovery. Although artificial intelligence can generate molecules with desired chemical properties, most previous studies have ignored the influence of disease-related cellular environments. This study proposes a novel deep generative model called GxVAEs to generate hit-like molecules from gene expression profiles by leveraging two joint variational autoencoders (VAEs). The first VAE, ProfileVAE, extracts latent features from gene expression profiles. The extracted features serve as the conditions that guide the second VAE, which is called MolVAE, in generating hit-like molecules. GxVAEs bridge the gap between molecular generation and the cellular environment in a biological system, and produce molecules that are biologically meaningful in the context of specific diseases. Experiments and case studies on the generation of therapeutic molecules show that GxVAEs outperforms current state-of-the-art baselines and yield hit-like molecules with potential bioactivity and drug-like properties. We were able to successfully generate the potential molecular structures with therapeutic effects for various diseases from patients’ disease profiles.

Clustering of small molecules: new perspectives and their impact on natural product lead discovery

The clustering of small molecules is of considerable importance for computer-aided drug discovery and virtual screening applications. The structure of chemical data in appropriate subspaces of the chemical space is relevant to sample datasets in a representative manner, to generate small libraries with wide or narrow chemical coverage (depending on the specific goals), and to guide the selection of subsets of in silico hits that are submitted for experimental confirmation. In the field of natural products, identifying regions of the chemical space where bioactive compounds congregate and understanding the relationship between biosynthetic gene clusters and the molecular structure of secondary metabolites may have a direct impact on natural product discovery and engineering. Here, we briefly discuss general approximations and available resources for the clustering of small molecules, and how the clustering of small molecules can be boosted by the application of novel clustering approximations, namely subspace clustering and multi-view clustering, which represent opposite philosophies of the clustering paradigm. We present some specific applications of small molecule clustering in the field of natural products, and analyze how a chemogenomic perspective may be particularly embodied in the field of natural products.

Computer-aided drug discovery of c-Abl kinase inhibitors from plant compounds against chronic myeloid leukemia

Chronic myeloid leukemia (CML) is a hematological malignancy characterized by the neoplastic transformation of hematopoietic stem cells, driven by the Philadelphia (Ph) chromosome resulting from a translocation between chromosomes 9 and 22. This Ph chromosome harbors the breakpoint cluster region (BCR) and the Abelson (ABL) oncogene (BCR-ABL1) which have a constitutive tyrosine kinase activity. However, the tyrosine kinase activity of BCR-ABL1 have been identified as a key player in CML initiation and maintenance through c-Abl kinase. Despite advancements in tyrosine kinase inhibitors, challenges such as efficacy, safety concerns, and recurring drug resistance persist. This study aims to discover potential c-Abl kinase inhibitors from plant compounds with anti-leukemic properties, employing drug-likeness assessment, molecular docking, in silico pharmacokinetics (ADMET) screening, density function theory (DFT), and molecular dynamics simulations (MDS). Out of 58 screened compounds for drug-likeness, 44 were docked against c-Abl kinase. The top hit compound (isovitexin) and nilotinib (control drug) were subjected to rigorous analyses, including ADMET profiling, DFT evaluation, and MDS for 100 ns. Isovitexin demonstrated a notable binding affinity (-15.492 kcal/mol), closely comparable to nilotinib (-16.826 kcal/mol), showcasing a similar binding pose and superior structural stability and reactivity. While these findings suggest isovitexin as a potential c-Abl kinase inhibitor, further validation through urgent in vitro and in vivo experiments is imperative. This research holds promise for providing an alternative avenue to address existing CML treatment and management challenges. Communicated by Ramaswamy H. Sarma

Graph Attention Site Prediction (GrASP): Identifying Druggable Binding Sites Using Graph Neural Networks with Attention.

Identifying and discovering druggable protein binding sites is an important early step in computer-aided drug discovery, but it remains a difficult task where most campaigns rely on a priori knowledge of binding sites from experiments. Here, we present a binding site prediction method called Graph Attention Site Prediction (GrASP) and re-evaluate assumptions in nearly every step in the site prediction workflow from data set preparation to model evaluation. GrASP is able to achieve state-of-the-art performance at recovering binding sites in PDB structures while maintaining a high degree of precision which will minimize wasted computation in downstream tasks such as docking and free energy perturbation.

Machine Learning-Based Virtual Screening of Antibacterial Agents against Methicillin-Susceptible and Resistant Staphylococcus aureus.

The application of computer-aided drug discovery (CADD) approaches has enabled the discovery of new antimicrobial therapeutic agents in the past. The high prevalence of methicillin-resistantStaphylococcus aureus(MRSA) strains promoted this pathogen to a high-priority pathogen for drug development. In this sense, modern CADD techniques can be valuable tools for the search for new antimicrobial agents. We employed a combination of a series of machine learning (ML) techniques to select and evaluate potential compounds with antibacterial activity against methicillin-susceptible S. aureus (MSSA) and MRSA strains. In the present study, we describe the antibacterial activity of six compounds against MSSA and MRSA reference (American Type Culture Collection (ATCC)) strains as well as two clinical strains of MRSA. These compounds showed minimal inhibitory concentrations (MIC) in the range from 12.5 to 200 μM against the different bacterial strains evaluated. Our results constitute relevant proven ML-workflow models to distinctively screen for novel MRSA antibiotics.

GraphLambda: Fusion Graph Neural Networks for Binding Affinity Prediction.

Predicting the binding affinity of protein-ligand complexes is crucial for computer-aided drug discovery (CADD) and the identification of potential drug candidates. The deep learning-based scoring functions have emerged as promising predictors of binding constants. Building on recent advancements in graph neural networks, we present graphLambda for protein-ligand binding affinity prediction, which utilizes graph convolutional, attention, and isomorphism blocks to enhance the predictive capabilities. The graphLambda model exhibits superior performance across CASF16 and CSAR HiQ NRC benchmarks and demonstrates robustness with respect to different types of train-validation set partitions. The development of graphLambda underscores the potential of graph neural networks in advancing binding affinity prediction models, contributing to more effective CADD methodologies.

Targeted Affinity Purification and Mechanism of Action of Angiotensin-Converting Enzyme (ACE) Inhibitory Peptides from Sea Cucumber Gonads.

Protein hydrolysates from sea cucumber (Apostichopus japonicus) gonads are rich in active materials with remarkable angiotensin-converting enzyme (ACE) inhibitory activity. Alcalase was used to hydrolyze sea cucumber gonads, and the hydrolysate was separated by the ultrafiltration membrane to produce a low-molecular-weight peptide component (less than 3 kDa) with good ACE inhibitory activity. The peptide component (less than 3 kDa) was isolated and purified using a combination method of ACE gel affinity chromatography and reverse high-performance liquid chromatography. The purified fractions were identified by liquid chromatography-tandem mass spectrometry (LC-MS/MS), and the resulting products were filtered using structure-based virtual screening (SBVS) to obtain 20 peptides. Of those, three noncompetitive inhibitory peptides (DDQIHIF with an IC50 value of 333.5 μmol·L-1, HDWWKER with an IC50 value of 583.6 μmol·L-1, and THDWWKER with an IC50 value of 1291.8 μmol·L-1) were further investigated based on their favorable pharmacochemical properties and ACE inhibitory activity. Molecular docking studies indicated that the three peptides were entirely enclosed within the ACE protein cavity, improving the overall stability of the complex through interaction forces with the ACE active site. The total free binding energies (ΔGtotal) for DDQIHIF, HDWWKER, and THDWWKER were -21.9 Kcal·mol-1, -71.6 Kcal·mol-1, and -69.1 Kcal·mol-1, respectively. Furthermore, a short-term assay of antihypertensive activity in spontaneously hypertensive rats (SHRs) revealed that HDWWKER could significantly decrease the systolic blood pressure (SBP) of SHRs after intravenous administration. The results showed that based on the better antihypertensive activity of the peptide in SHRs, the feasibility of targeted affinity purification and computer-aided drug discovery (CADD) for the efficient screening and preparation of ACE inhibitory peptide was verified, which provided a new idea of modern drug development method for clinical use.

An interdisciplinary course on computer-aided drug discovery to broaden student participation in original scientific research.

We present a new highly interdisciplinary project-based course in computer aided drug discovery (CADD). This course was developed in response to a call for alternative pedagogical approaches during the COVID-19 pandemic, which caused the cancellation of a face-to-face summer research program sponsored by the Louisiana Biomedical Research Network (LBRN). The course integrates guided research and educational experiences for chemistry, biology, and computer science students. We implement research-based methods with publicly available tools in bioinformatics and molecular modeling to identify and prioritize promising antiviral drug candidates for COVID-19. The purpose of this course is three-fold: I. Implement an active learning and inclusive pedagogy that fosters student engagement and research mindset; II. Develop student interdisciplinary research skills that are highly beneficial in a broader scientific context; III. Demonstrate that pedagogical shifts (initially incurred during the COVID-19 pandemic) can furnish longer-term instructional benefits. The course, which has now been successfully taught a total of five times, incorporates four modules, including lectures/discussions, live demos, inquiry-based assignments, and science communication.