Prediction method of pharmacokinetic parameters of small molecule drugs based on GCN network model.

CONTEXT: Accurately predicting plasma protein binding rate (PPBR) and oral bioavailability (OBA) helps to better reveal the absorption and distribution of drugs in the human body and subsequent drug design. Although machine learning models have achieved good results in prediction accuracy, they often suffer from insufficient accuracy when dealing with data with irregular topological structures. METHODS: In view of this, this study proposes a pharmacokinetic parameter prediction framework based on graph convolutional networks (GCN), which predicts the PPBR and OBA of small molecule drugs. In the framework, GCN is first used to extract spatial feature information on the topological structure of drug molecules, in order to better learn node features and association information between nodes. Then, based on the principle of drug similarity, this study calculates the similarity between small molecule drugs, selects different thresholds to construct datasets, and establishes a prediction model centered on the GCN algorithm. The experimental results show that compared with traditional machine learning prediction models, the prediction model constructed based on the GCN method performs best on PPBR and OBA datasets with an inter-molecular similarity threshold of 0.25, with MAE of 0.155 and 0.167, respectively. In addition, in order to further improve the accuracy of the prediction model, GCN is combined with other algorithms. Compared to using a single GCN method, the distribution of the predicted values obtained by the combined model is highly consistent with the true values. In summary, this work provides a new method for improving the rate of early drug screening in the future.

Chemoinformatic approaches for navigating large chemical spaces.

INTRODUCTION: Large chemical spaces (CSs) include traditional large compound collections, combinatorial libraries covering billions to trillions of molecules, DNA-encoded chemical libraries comprising complete combinatorial CSs in a single mixture, and virtual CSs explored by generative models. The diverse nature of these types of CSs require different chemoinformatic approaches for navigation. AREAS COVERED: An overview of different types of large CSs is provided. Molecular representations and similarity metrics suitable for large CS exploration are discussed. A summary of navigation of CSs in generative models is provided. Methods for characterizing and comparing CSs are discussed. EXPERT OPINION: The size of large CSs might restrict navigation to specialized algorithms and limit it to considering neighborhoods of structurally similar molecules. Efficient navigation of large CSs not only requires methods that scale with size but also requires smart approaches that focus on better but not necessarily larger molecule selections. Deep generative models aim to provide such approaches by implicitly learning features relevant for targeted biological properties. It is unclear whether these models can fulfill this ideal as validation is difficult as long as the covered CSs remain mainly virtual without experimental verification.

Isolation and in silico investigation of cannflavins from Cannabis sativa leaves as potential anti-SARS-CoV-2 agents targeting the Papain-Like Protease.

This study aimed to isolate and identify three prenylflavonoids (cannflavin A, B, and C) from Cannabis sativa leaves using different chromatographic techniques. The potential of the isolated compounds against SARS-CoV-2 was suggested through several in silico analysis. Structural similarity studies against nine co-crystallized ligands of SARS-CoV-2's proteins indicated the similarities of the isolated cannflavins with the SARS-CoV-2 Papain-Like Protease (PLP) ligand, Y95. Then, flexible allignment study confirmed this similarity. Docking experiments showed successful binding of all cannflavins within the active pocket of PLP, with energies comparable to Y95. Among them, cannflavin A demonstrated the most similar binding mode, while cannflavin C exhibited the best energy. Molecular dynamics (MD) simulations and MM-GPSA confirmed the accurate binding of cannflavin A to the PLP. In silico ADMET studies indicated favourable drug-like properties for all three compounds, suggesting their potential as anti-SARS-CoV-2 agents. Further In vitro and In vivo investigations are necessary to validate these findings and establish their efficacy and safety profiles.

Classification Maps: A New Mathematical Tool Supporting the Diagnosis of Age-Related Macular Degeneration.

OBJECTIVE: A new diagnostic graphical tool-classification maps-supporting the detection of Age-Related Macular Degeneration (AMD) has been constructed. METHODS: The classification maps are constructed using the ordinal regression model. In the ordinal regression model, the ordinal variable (the dependent variable) is the degree of the advancement of AMD. The other variables, such as CRT (Central Retinal Thickness), GCC (Ganglion Cell Complex), MPOD (Macular Pigment Optical Density), ETDRS (Early Treatment Diabetic Retinopathy Study), Snellen and Age have also been used in the analysis and are represented on the axes of the maps. RESULTS: Here, 132 eyes were examined and classified to the AMD advancement level according to the four-point Age-Related Eye Disease Scale (AREDS): AREDS 1, AREDS 2, AREDS 3 and AREDS 4. These data were used for the creation of two-dimensional classification maps for each of the four stages of AMD. CONCLUSIONS: The maps allow us to perform the classification of the patient's eyes to particular stages of AMD. The pairs of the variables represented on the axes of the maps can be treated as diagnostic identifiers necessary for the classification to particular stages of AMD.

Discovery of novel TRPV1 modulators through machine learning-based molecular docking and molecular similarity searching.

The transient receptor potential vanilloid 1 (TRPV1) channel belongs to the transient receptor potential channel superfamily and participates in many physiological processes. TRPV1 modulators (both agonists and antagonists) can effectively inhibit pain caused by various factors and have curative effects in various diseases, such as itch, cancer, and cardiovascular diseases. Therefore, the development of TRPV1 channel modulators is of great importance. In this study, the structure-based virtual screening and ligand-based virtual screening methods were used to screen compound databases respectively. In the structure-based virtual screening route, a full-length human TRPV1 protein was first constructed, three molecular docking methods with different precisions were performed based on the hTRPV1 structure, and a machine learning-based rescoring model by the XGBoost algorithm was constructed to enrich active compounds. In the ligand-based virtual screening route, the ROCS program was used for 3D shape similarity searching and the EON program was used for electrostatic similarity searching. Final 77 compounds were selected from two routes for in vitro assays. The results showed that 8 of them were identified as active compounds, including three hits with IC50 values close to capsazepine. In addition, one hit is a partial agonist with both agonistic and antagonistic activity. The mechanisms of some active compounds were investigated by molecular dynamics simulation, which explained their agonism or antagonism.

Influence of Molecular Structure and Physicochemical Properties of Immunosuppressive Drugs on Micelle Formulation Characteristics and Cutaneous Delivery.

The aim of this study was to investigate whether subtle differences in molecular properties affected polymeric micelle characteristics and their ability to deliver poorly water-soluble drugs into the skin. D-α-tocopherol-polyethylene glycol 1000 was used to prepare micelles containing ascomycin-derived immunosuppressants-sirolimus (SIR), pimecrolimus (PIM) and tacrolimus (TAC)-which have similar structures and physicochemical properties and have dermatological applications. Micelle formulations were prepared by thin-film hydration and extensively characterized. Cutaneous delivery and biodistribution were determined and compared. Sub-10 nm micelles were obtained for the three immunosuppressants with incorporation efficiencies >85%. However, differences were observed for drug loading, stability (at the highest concentration), and their in vitro release kinetics. These were attributed to differences in drug aqueous solubility and lipophilicity. Differences between the cutaneous biodistribution profiles and drug deposition in the different skin compartments pointed to the impact of differences in thermodynamic activity. Therefore, despite their structural similarities, SIR, TAC and PIM did not demonstrate the same behaviour either in the micelles or when applied to the skin. These outcomes indicate that polymeric micelles should be optimized even for closely related drug molecules and support the hypothesis that drugs are released from micelles prior to skin penetration.

2-[(2-Amino-6-methylpyrimidin-4-yl)sulfanyl]-N-arylacetamides: Discovery of a new class of anti-tubercular agents and prospects for their further structural modification.

The synthesis of 2-[(2-amino-6-methylpyrimidin-4-yl)sulfanyl]-N-arylacetamides 6a-j was encouraged by their antibacterial activity and drug-likeness predictions. Of the compounds, two bearing 4­isopropylphenyl 6c and 2,5­dichlorophenyl 6i moieties were found to be threefold more potent than the first-line tuberculosis drug ethambutol. A molecular docking study revealed that compound 6c may selectively bind to cyclopropane mycolic acid synthase 1, an enzyme essential for the construction of the tuberculosis bacteria cell wall. Keeping this in mind, a recently developed ligand-based virtual screening strategy combining the molecular similarity search and docking approaches was adopted to identify more potent analogs of the parent compound. As a result, a series of new ligands 18p-w with phenyl-substituted azinyl amide groups were in silico discovered. Due to their high binding affinities to the enzyme and improved toxicity profiles, the ligands are undoubtedly worth future synthetic efforts.

Visualizing chemical space networks with RDKit and NetworkX.

This article demonstrates how to create Chemical Space Networks (CSNs) using a Python RDKit and NetworkX workflow. CSNs are a type of network visualization that depict compounds as nodes connected by edges, defined as a pairwise relationship such as a 2D fingerprint similarity value. A step by step approach is presented for creating two different CSNs in this manuscript, one based on RDKit 2D fingerprint Tanimoto similarity values, and another based on maximum common substructure similarity values. Several different CSN visualization features are included in the tutorial including methods to represent nodes with color based on bioactivity attribute value, edges with different line styles based on similarity value, as well as replacing the circle nodes with 2D structure depictions. Finally, some common network property and analysis calculations are presented including the clustering coefficient, degree assortativity, and modularity. All code is provided in the form of Jupyter Notebooks and is available on GitHub with a permissive BSD-3 open-source license: https://github.com/vfscalfani/CSN_tutorial.

The Computational Preventive Potential of the Rare Flavonoid, Patuletin, Isolated from Tagetes patula, against SARS-CoV-2.

The rare flavonoid, patuletin, was isolated from the flowers of Tagetes patula growing in Egypt. The rarity of the isolated compound inspired us to scrutinize its preventive effect against COVID-19 utilizing a multi-step computational approach. Firstly, a structural similarity study was carried out against nine ligands of nine SARS-CoV-2 proteins. The results showed a large structural similarity between patuletin and F86, the ligand of SARS-CoV-2 RNA-dependent RNA polymerase (RdRp). Then, a 3D-Flexible alignment study of patuletin and F86 verified the proposed similarity. To determine the binding opportunity, patuletin was docked against the RdRp showing a correct binding inside its active pocket with an energy of -20 kcal/mol that was comparable to that of F86 (-23 kcal/mol). Following, several MD simulations as well as MM-PBSA studies authenticated the accurate binding of patuletin in the RdRp via the correct dynamic and energetic behaviors over 100 ns. Additionally, in silico ADMET studies showed the general safety and drug-likeness of patuletin.

Molecular Similarity Perception Based on Machine-Learning Models.

Molecular similarity is an impressively broad topic with many implications in several areas of chemistry. Its roots lie in the paradigm that 'similar molecules have similar properties'. For this reason, methods for determining molecular similarity find wide application in pharmaceutical companies, e.g., in the context of structure-activity relationships. The similarity evaluation is also used in the field of chemical legislation, specifically in the procedure to judge if a new molecule can obtain the status of orphan drug with the consequent financial benefits. For this procedure, the European Medicines Agency uses experts' judgments. It is clear that the perception of the similarity depends on the observer, so the development of models to reproduce the human perception is useful. In this paper, we built models using both 2D fingerprints and 3D descriptors, i.e., molecular shape and pharmacophore descriptors. The proposed models were also evaluated by constructing a dataset of pairs of molecules which was submitted to a group of experts for the similarity judgment. The proposed machine-learning models can be useful to reduce or assist human efforts in future evaluations. For this reason, the new molecules dataset and an online tool for molecular similarity estimation have been made freely available.

Reference dose prediction by using CDK molecular descriptors: A non-experimental method.

Reference dose (RfD) is an estimate of a daily dose that individual can be exposed chronically without obvious deleterious effects during a lifetime. In the area of toxicology, researchers always use the traditional approach by employing NOAEL/LOAEL or the benchmark dose (BMD) and other dose-response approaches to estimate RfD. These methods have, despite their typicalness, certain limitations. In this study, we present a novel method of the estimation of reference dose without experiments. The information of the organic chemicals is available from the Integrated Risk Information System (IRIS) of USEPA. Molecular descriptors for each molecular structure were calculated by an integrated platform, and the chemicals were classified into four categories based on molecular similarity: 128 contained benzene rings, 47 were heteroaromatics, 104 contained halogen substituents and 44 were halogenated aliphatic hydrocarbons. The predictive model of RfD was constructed by the multiple linear stepwise regression (MLR) method. Approximately 95% and 82% of the data points differ by less than 10-fold and 5-fold between the predicted values and the true values respectively. The non-experimental method improves the estimation efficiency and has a certain reference value to predict.

Feature Reduction for Molecular Similarity Searching Based on Autoencoder Deep Learning.

The concept of molecular similarity has been commonly used in rational drug design, where structurally similar molecules are examined in molecular databases to retrieve functionally similar molecules. The most used conventional similarity methods used two-dimensional (2D) fingerprints to evaluate the similarity of molecules towards a target query. However, these descriptors include redundant and irrelevant features that might impact the performance of similarity searching methods. Thus, this study proposed a new approach for identifying the important features of molecules in chemical datasets based on the representation of the molecular features using Autoencoder (AE), with the aim of removing irrelevant and redundant features. The proposed approach experimented using the MDL Data Drug Report standard dataset (MDDR). Based on experimental findings, the proposed approach performed better than several existing benchmark similarity methods such as Tanimoto Similarity Method (TAN), Adapted Similarity Measure of Text Processing (ASMTP), and Quantum-Based Similarity Method (SQB). The results demonstrated that the performance achieved by the proposed approach has proven to be superior, particularly with the use of structurally heterogeneous datasets, where it yielded improved results compared to other previously used methods with the similar goal of improving molecular similarity searching.

Isolation and In Silico SARS-CoV-2 Main Protease Inhibition Potential of Jusan Coumarin, a New Dicoumarin from Artemisia glauca.

A new dicoumarin, jusan coumarin, (1), has been isolated from Artemisia glauca aerial parts. The chemical structure of jusan coumarin was estimated, by 1D, 2D NMR as well as HR-Ms spectroscopic methods, to be 7-hydroxy-6-methoxy-3-[(2-oxo-2H-chromen-6-yl)oxy]-2H-chromen-2-one. As the first time to be introduced in nature, its potential against SARS-CoV-2 has been estimated using various in silico methods. Molecular similarity and fingerprints experiments have been utilized for 1 against nine co-crystallized ligands of COVID-19 vital proteins. The results declared a great similarity between Jusan Coumarin and X77, the ligand of COVID-19 main protease (PDB ID: 6W63), Mpro. To authenticate the obtained outputs, a DFT experiment was achieved to confirm the similarity of X77 and 1. Consequently, 1 was docked against Mpro. The results clarified that 1 bonded in a correct way inside Mpro active site, with a binding energy of -18.45 kcal/mol. Furthermore, the ADMET and toxicity profiles of 1 were evaluated and showed the safety of 1 and its likeness to be a drug. Finally, to confirm the binding and understand the thermodynamic characters between 1 and Mpro, several molecular dynamics (MD) simulations studies have been administered. Additionally, the known coumarin derivative, 7-isopentenyloxycoumarin (2), has been isolated as well as ß-sitosterol (3).

Jusanin, a New Flavonoid from Artemisia commutata with an In Silico Inhibitory Potential against the SARS-CoV-2 Main Protease.

A new flavonoid, Jusanin, (1) has been isolated from the aerial parts of Artemisia commutata. The chemical structure of Jusanin has been elucidated using 1D, 2D NMR, and HR-Ms spectroscopic methods to be 5,2',4'-trihydroxy-6,7,5'-trimethoxyflavone. Being new in nature, the inhibition potential of 1 has been estimated against SARS-CoV-2 using different in silico techniques. Firstly, molecular similarity and fingerprint studies have been conducted for Jusanin against co-crystallized ligands of eight different SARS-CoV-2 essential proteins. The studies indicated the similarity between 1 and X77, the co-crystallized ligand SARS-CoV-2 main protease (PDB ID: 6W63). To confirm the obtained results, a DFT study was carried out and indicated the similarity of (total energy, HOMO, LUMO, gap energy, and dipole moment) between 1 and X77. Accordingly, molecular docking studies of 1 against the target enzyme have been achieved and showed that 1 bonded correctly in the protein's active site with a binding energy of -19.54 Kcal/mol. Additionally, in silico ADMET in addition to the toxicity evaluation of Jusanin against seven models have been preceded and indicated the general safety and the likeness of Jusanin to be a drug. Finally, molecular dynamics simulation studies were applied to investigate the dynamic behavior of the Mpro-Jusanin complex and confirmed the correct binding at 100 ns. In addition to 1, three other metabolites have been isolated and identified to be сapillartemisin A (2), methyl-3-[S-hydroxyprenyl]-cumarate (3), and ß-sitosterol (4).

Identification of Corosolic and Oleanolic Acids as Molecules Antagonizing the Human RORγT Nuclear Receptor Using the Calculated Fingerprints of the Molecular Similarity.

RORγT is a protein product of the RORC gene belonging to the nuclear receptor subfamily of retinoic-acid-receptor-related orphan receptors (RORs). RORγT is preferentially expressed in Th17 lymphocytes and drives their differentiation from naive CD4+ cells and is involved in the regulation of the expression of numerous Th17-specific cytokines, such as IL-17. Because Th17 cells are implicated in the pathology of autoimmune diseases (e.g., psoriasis, inflammatory bowel disease, multiple sclerosis), RORγT, whose activity is regulated by ligands, has been recognized as a drug target in potential therapies against these diseases. The identification of such ligands is time-consuming and usually requires the screening of chemical libraries. Herein, using a Tanimoto similarity search, we found corosolic acid and other pentacyclic tritepenes in the library we previously screened as compounds highly similar to the RORγT inverse agonist ursolic acid. Furthermore, using gene reporter assays and Th17 lymphocytes, we distinguished compounds that exert stronger biological effects (ursolic, corosolic, and oleanolic acid) from those that are ineffective (asiatic and maslinic acids), providing evidence that such combinatorial methodology (in silico and experimental) might help wet screenings to achieve more accurate results, eliminating false negatives.

Molecular Similarity for Drug Discovery, Target Prediction and Chemical Space Visualization.

Similar drug molecules often have similar properties and activities. Therefore, quantifying molecular similarity is central to drug discovery and optimization. Here I review computational methods using molecular similarity measures developed in my group within the interdisciplinary network NCCR TransCure investigating the physiology, structural biology and pharmacology of ion channels and membrane transporters. We designed a 3D molecular shape and pharmacophore comparison algorithm to optimize weak and unselective inhibitors by scaffold hopping and discovered potent and selective inhibitors of the ion channels TRPV6 and TRPM4, of endocannabinoid membrane transport, and of the divalent metal transporters DMT1 and ZIP8. We predicted off-target effects by combining molecular similarity searches from different molecular fingerprints against target annotated compounds from the ChEMBL database. Finally, we created interactive chemical space maps reflecting molecular similarities to facilitate the selection of screening compounds and the analysis of screening results. These different tools are available online at https://gdb.unibe.ch/tools/.

Building 2D classification models and 3D CoMSIA models on small-molecule inhibitors of both wild-type and T790M/L858R double-mutant EGFR.

Epidermal growth factor receptor (EGFR) has received widespread attention because it is an important target for anticancer drug design. Mutations in the EGFR, especially the T790M/L858R double mutation, have made cancer treatment more difficult. We herein built the structure-activity relationship models of small-molecule inhibitors on wild-type and T790M/L858R double-mutant EGFR with a whole dataset of 379 compounds. For 2D classification models, we used ECFP4 fingerprints to build support vector machine and random forest models and used SMILES to build self-attention recurrent neural network models. Each of all six models resulted in an accuracy of above 0.87 and the Matthews correlation coefficient value of above 0.76 on the test set, respectively. We concluded that inhibitors containing anilinoquinoline and methoxy or fluoro phenyl are highly active against wild EGFR. Substructures such as anilinopyrimidine, acrylamide, amino phenyl, methoxy phenyl, and thienopyrimidinyl amide appeared more in highly active inhibitors against double-mutant EGFR. We also used self-organizing map to cluster the inhibitors into six subsets based on ECFP4 fingerprints and analyzed the activity characteristics of different scaffolds in each subset. Among them, three datasets, which are based on pteridin, anilinopyrimidine, and anilinoquinoline scaffold, were selected to build 3D comparative molecular similarity analysis models individually. Models with the leave-one-out coefficient of determination (q2) above 0.65 were selected, and five descriptor types (steric, electrostatic, hydrophobic, donor, and acceptor) were used to study the effects of side chains of inhibitors on the activity against wild-type and mutant-type EGFR.

Discovery of Octahydroisoindolone as a Scaffold for the Selective Inhibition of Chitinase B1 from Aspergillus fumigatus: In Silico Drug Design Studies.

Chitinases represent an alternative therapeutic target for opportunistic invasive mycosis since they are necessary for fungal cell wall remodeling. This study presents the design of new chitinase inhibitors from a known hydrolysis intermediate. Firstly, a bioinformatic analysis of Aspergillus fumigatus chitinase B1 (AfChiB1) and chitotriosidase (CHIT1) by length and conservation was done to obtain consensus sequences, and molecular homology models of fungi and human chitinases were built to determine their structural differences. We explored the octahydroisoindolone scaffold as a potential new antifungal series by means of its structural and electronic features. Therefore, we evaluated several synthesis-safe octahydroisoindolone derivatives by molecular docking and evaluated their AfChiB1 interaction profile. Additionally, compounds with the best interaction profile (1-5) were docked within the CHIT1 catalytic site to evaluate their selectivity over AfChiB1. Furthermore, we considered the interaction energy (MolDock score) and a lipophilic parameter (aLogP) for the selection of the best candidates. Based on these descriptors, we constructed a mathematical model for the IC50 prediction of our candidates (60-200 µM), using experimental known inhibitors of AfChiB1. As a final step, ADME characteristics were obtained for all the candidates, showing that 5 is our best designed hit, which possesses the best pharmacodynamic and pharmacokinetic character.

Ligand-Based Virtual Screening Based on the Graph Edit Distance.

Chemical compounds can be represented as attributed graphs. An attributed graph is a mathematical model of an object composed of two types of representations: nodes and edges. Nodes are individual components, and edges are relations between these components. In this case, pharmacophore-type node descriptions are represented by nodes and chemical bounds by edges. If we want to obtain the bioactivity dissimilarity between two chemical compounds, a distance between attributed graphs can be used. The Graph Edit Distance allows computing this distance, and it is defined as the cost of transforming one graph into another. Nevertheless, to define this dissimilarity, the transformation cost must be properly tuned. The aim of this paper is to analyse the structural-based screening methods to verify the quality of the Harper transformation costs proposal and to present an algorithm to learn these transformation costs such that the bioactivity dissimilarity is properly defined in a ligand-based virtual screening application. The goodness of the dissimilarity is represented by the classification accuracy. Six publicly available datasets-CAPST, DUD-E, GLL&GDD, NRLiSt-BDB, MUV and ULS-UDS-have been used to validate our methodology and show that with our learned costs, we obtain the highest ratios in identifying the bioactivity similarity in a structurally diverse group of molecules.

Dissimilar Ligands Bind in a Similar Fashion: A Guide to Ligand Binding-Mode Prediction with Application to CELPP Studies.

The molecular similarity principle has achieved great successes in the field of drug design/discovery. Existing studies have focused on similar ligands, while the behaviors of dissimilar ligands remain unknown. In this study, we developed an intercomparison strategy in order to compare the binding modes of ligands with different molecular structures. A systematic analysis of a newly constructed protein-ligand complex structure dataset showed that ligands with similar structures tended to share a similar binding mode, which is consistent with the Molecular Similarity Principle. More importantly, the results revealed that dissimilar ligands can also bind in a similar fashion. This finding may open another avenue for drug discovery. Furthermore, a template-guiding method was introduced for predicting protein-ligand complex structures. With the use of dissimilar ligands as templates, our method significantly outperformed the traditional molecular docking methods. The newly developed template-guiding method was further applied to recent CELPP studies.