Dengue fever, which is a disease caused by the dengue virus (DENV), is a major unsolved issue in many tropical and sub-tropical regions of the world. The absence of treatment that effectively prevent further viral propagation inside the human’s body resulted in a high number of deaths globally each year. Thus, novel anti-dengue therapies are required for effective treatment. Hu-man hexokinase II (HKII), which is the first enzyme in the glycolytic pathway, is an important drug target due to its significant impact on viral replication and survival in host cells. In this study, 23.1 million compounds were computationally-screened against HKII using the Ultrafast Shape Recognition with a CREDO Atom Types (USRCAT) algorithm. In total, 300 compounds with the highest similarity scores relative to three reference molecules, known as Al-pha-D-glucose (GLC), Beta-D-glucose-6-phosphate (BG6), and 2-deoxyglucose (2DG), were aligned. Of these 300 compounds, 165 were chosen for further structure-based screening, based on their similarity scores, ADME analysis, the Lipinski’s Rule of Five, and virtual toxicity test results. The selected analogues were subsequently docked against each domain of the HKII structure (PDB ID: 2NZT) using AutoDock Vina programme. The three top-ranked compounds for each query were then selected from the docking results based on their binding energy, the number of hydrogen bonds formed, and the specific catalytic residues. The best docking results for each analogue were observed for the C-terminus of Chain B. The top-ranked analogues of GLC, compound 10, compound 26, and compound 58, showed predicted binding energies of −7.2, −7.0, and −6.10 kcal/mol and 7, 5, and 2 hydrogen bonds, respectively. The analogues of BG6, compound 30, compound 36, and compound 38, showed predicted binding energies of −7.8, −7.4, and −7.0 kcal/mol and 11, 9, and 5 hydrogen bonds, while the top three analogues of 2DG, known as compound 1, compound 4, and compound 31, showed predicted binding energies of −6.8, −6.3, and −6.3 kcal/mol and 4, 3, and 1 hydrogen bonds, sequentially. The highest-ranked compounds in the docking analysis were then selected for molecular dynamics simulation, where compound 10, compound 30, and compound 1, which are the analogues of GLC, BG6, and 2DG, have shown strong protein-ligand stability with an RMSD value of ±5.0 A° with a 5 H bond, ±4.0 A° with an 8 H bond, and ±0.5 A° with a 2 H bond, respectively, compared to the reference molecules throughout the 20 ns simulation time. Therefore, by using the computational studies, we pro-posed novel compounds, which may act as potential drugs against DENV by inhibiting HKII’s activity.

https://www.mdpi.com/2673-6284/10/1/1/pdf

Virtual screening for potential inhibitors of human hexokinase II for the development of anti-dengue therapeutics

The advancements of information technology and related processing techniques have created a fertile base for progress in many scientific fields and industries. In the fields of drug discovery and development, machine learning techniques have been used for the development of novel drug candidates. The methods for designing drug targets and novel drug discovery now routinely combine machine learning and deep learning algorithms to enhance the efficiency, efficacy, and quality of developed outputs. The generation and incorporation of big data, through technologies such as high-throughput screening and high through-put computational analysis of databases used for both lead and target discovery, has increased the reliability of the machine learning and deep learning incorporated techniques. The use of these virtual screening and encompassing online information has also been highlighted in developing lead synthesis pathways. In this review, machine learning and deep learning algorithms utilized in drug discovery and associated techniques will be discussed. The applications that produce promising results and methods will be reviewed.

/pdf/machine-learning-methods-in-drug-discovery-5ay4j3jh1p.pdf

Machine Learning Methods in Drug Discovery.

Current trends in computer aided drug design and a highlight of drugs discovered via computational techniques: A review

Recently, deep neural network (DNN)-based drug-target interaction (DTI) models are highlighted for their high accuracy with affordable computational costs. Yet, the models' insufficient generalization remains a challenging problem in the practice of in-silico drug discovery. We propose two key strategies to enhance generalization in the DTI model. The first one is to integrate physical models into DNN models. Our model, PIGNet, predicts the atom-atom pairwise interactions via physics-informed equations parameterized with neural networks and provides the total binding affinity of a protein-ligand complex as their sum. We further improved the model generalization by augmenting a wider range of binding poses and ligands to training data. PIGNet achieved a significant improvement in docking success rate, screening enhancement factor, and screening success rate by up to 2.01, 10.78, 14.0 times, respectively, compared to the previous DNN models. The physics-informed model also enables the interpretation of predicted binding affinities by visualizing the energy contribution of ligand substructures, providing insights for ligand optimization. Finally, we devised the uncertainty estimator of our model's prediction to qualify the outcomes and reduce the false positive rates.

PIGNet: A physics-informed deep learning model toward generalized drug-target interaction predictions.

Currently, a worldwide pandemic has been declared in response to the spread of coronavirus disease 2019 (COVID-19), a fatal and fast-spreading viral infection caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The low availability of efficient vaccines and treatment options has resulted in a high mortality rate, bringing the world economy to its knees. Thus, mechanistic investigations of drugs capable of counteracting this disease are in high demand. The main protease (Mpro) expressed by SARS-CoV-2 has been targeted for the development of potential drug candidates due to the crucial role played by Mpro in viral replication and transcription. We generated a phytochemical library containing 1672 phytochemicals derived from 56 plants, which have been reported as having antiviral, antibacterial, and antifungal activity. A molecular docking program was used to screen the top three candidate compounds: epicatechin-3-O-gallate, psi-taraxasterol, and catechin gallate, which had respective binding affinities of −8.4, −8.5, and −8.8 kcal/mol. Several active sites in the targeted protein, including Cys145, His41, Met49, Glu66, and Met165, were found to interact with the top three candidate compounds. The multiple simulation profile, root-mean-square deviation, root-mean-square fluctuation, radius of gyration, and solvent-accessible surface area values supported the inflexible nature of the docked protein–compound complexes. The toxicity and carcinogenicity profiles were assessed, which showed that epicatechin-3-O-gallate, psi-taraxasterol, and catechin gallate had favorable pharmacological properties with no adverse effects. These findings suggest that these compounds could be developed as part of an effective drug development pathway to treat COVID-19.

https://www.mdpi.com/2079-7737/10/7/589/pdf

Plant-Based Phytochemical Screening by Targeting Main Protease of SARS-CoV-2 to Design Effective Potent Inhibitors

The drug development process is a major challenge in the pharmaceutical industry since it takes a substantial amount of time and money to move through all the phases of developing of a new drug. One extensively used method to minimize the cost and time for the drug development process is computer-aided drug design (CADD). CADD allows better focusing on experiments, which can reduce the time and cost involved in researching new drugs. In this context, structure-based virtual screening (SBVS) is robust and useful and is one of the most promising in silico techniques for drug design. SBVS attempts to predict the best interaction mode between two molecules to form a stable complex, and it uses scoring functions to estimate the force of non-covalent interactions between a ligand and molecular target. Thus, scoring functions are the main reason for the success or failure of SBVS software. Many software programs are used to perform SBVS, and since they use different algorithms, it is possible to obtain different results from different software using the same input. In the last decade, a new technique of SBVS called consensus virtual screening (CVS) has been used in some studies to increase the accuracy of SBVS and to reduce the false positives obtained in these experiments. An indispensable condition to be able to utilize SBVS is the availability of a 3D structure of the target protein. Some virtual databases, such as the Protein Data Bank, have been created to store the 3D structures of molecules. However, sometimes it is not possible to experimentally obtain the 3D structure. In this situation, the homology modeling methodology allows the prediction of the 3D structure of a protein from its amino acid sequence. This review presents an overview of the challenges involved in the use of CADD to perform SBVS, the areas where CADD tools support SBVS, a comparison between the most commonly used tools, and the techniques currently used in an attempt to reduce the time and cost in the drug development process. Finally, the final considerations demonstrate the importance of using SBVS in the drug development process.

/pdf/structure-based-virtual-screening-from-classical-to-49xy8rv86g.pdf

Structure-Based Virtual Screening: From Classical to Artificial Intelligence.

DNA is a molecular target for the treatment of several diseases, including cancer, but there are few docking methodologies exploring the interactions between nucleic acids with DNA intercalating agents. Different docking methodologies, such as AutoDock Vina, DOCK 6, and Consensus, implemented into Molecular Architect (MolAr), were evaluated for their ability to analyze those interactions, considering visual inspection, redocking, and ROC curve. Ligands were refined by Parametric Method 7 (PM7), and ligands and decoys were docked into the minor DNA groove (PDB code: 1VZK). As a result, the area under the ROC curve (AUC-ROC) was 0.98, 0.88, and 0.99 for AutoDock Vina, DOCK 6, and Consensus methodologies, respectively. In addition, we proposed a machine learning model to determine the experimental ∆Tm value, which found a 0.84 R2 score. Finally, the selected ligands mono imidazole lexitropsin (42), netropsin (45), and N,N′-(1H-pyrrole-2,5-diyldi-4,1-phenylene)dibenzenecarboximidamide (51) were submitted to Molecular Dynamic Simulations (MD) through NAMD software to evaluate their equilibrium binding pose into the groove. In conclusion, the use of MolAr improves the docking results obtained with other methodologies, is a suitable methodology to use in the DNA system and was proven to be a valuable tool to estimate the ∆Tm experimental values of DNA intercalating agents.

/pdf/evaluation-of-docking-machine-learning-and-molecular-1g7mq7w5.pdf

Evaluation of Docking Machine Learning and Molecular Dynamics Methodologies for DNA-Ligand Systems

Abstract Alzheimer's disease (AD) is a neurodegenerative pathology responsible for 70% of dementia cases worldwide. Despite its relevance, the few drugs available for the treatment of this disease offer only symptomatic relief, with limited efficacy and serious adverse effects. The most accepted hypothesis about the pathogenesis involves the aggregation and deposition of β-amyloid peptides, mainly in the cerebral cortex and hippocampus, through the catalytic action of beta-secretase 1 (BACE-1), making this enzyme a promising target for the development of new drugs. In order to prioritize candidates for BACE-1 inhibitors, a hierarchical virtual screening by pharmacophore model and molecular docking was performed against the 216,833 molecules contained in several databases. Our previously built pharmacophore model was used for the first filtering step, which resulted in the selection of 399 molecules. The remaining molecules were filtered through molecular docking with GOLD 5.4.0. In this step, molecules with scoring values greater than the mean plus standard deviation were evaluated for commercial availability and absence of asymmetric centers. Four molecules were selected and evaluated for mutagenic potential by the AMES test with the help of the pkCSM server. Finally, they were tested against the descriptors on Lipinski and Veber rules, and ZINC01589617 (QFIT = 56.52/Score = 44.95) satisfied all requirements, being subjected to molecular dynamics simulations (t = 100 ns) in order to obtain robust data on the mode of bonding and profile of intermolecular interactions. Those in silico strategies demonstrated that ZINC01589617 is a potential candidate for biological tests. Communicated by Ramaswamy H. Sarma

Tiago Alves de Oliveira

Papers

Structure-Based Virtual Screening: From Classical to Artificial Intelligence.

Evaluation of Docking Machine Learning and Molecular Dynamics Methodologies for DNA-Ligand Systems

Identification of potential human beta-secretase 1 inhibitors by hierarchical virtual screening and molecular dynamics