Quantitative structure–activity relationship analysis using deep learning based on a novel molecular image input technique

Quantitative structure–activity relationship (QSAR) analysis uses structural, quantum chemical, and physicochemical features calculated from molecular geometry as explanatory variables predicting physiological activity. Recently, deep learning based on advanced artificial neural networks has demonstrated excellent performance in the discipline of QSAR research. While it has properties of feature representation learning that directly calculate feature values from molecular structure, the use of this potential function is limited in QSAR modeling. The present study applied this function of feature representation learning to QSAR analysis by incorporating 360° images of molecular conformations into deep learning. Accordingly, I successfully constructed a highly versatile identification model for chemical compounds that induce mitochondrial membrane potential disruption with the external validation area under the receiver operating characteristic curve of ≥0.9.     

Combined 3D QSAR and molecular docking studies to reveal novel cannabinoid ligands with optimum binding activity

The combination of NMR spectroscopy and molecular modeling studies provided the putative bioactive conformation for the analgesic cannabinoid (CB) ligand (−)-2-(6a,7,10,10a-tetrahydro-6,6,9-trimethylhydroxy-6H-dibenzo[b,d]pyranyl)-2-hexyl 1,3-dithiolane which served as a template in reported three-dimensional quantitative structure–activity relationship (3D QSAR) studies [Durdagi et al., J. Med. Chem. 2007, 50, 2875]. The reported 3D models of the CB1 receptor allowed us to construct a new 3D QSAR model based on theoretical calculations and molecular docking studies. Statistical comparison of the constructed two 3D QSAR studies showed the improvement of the new model. In addition, the new model can explain more effectively the experimental data and thus it can serve more efficiently in the rational drug design of pharmacologically optimized CB analogues.     

Discovery of molecular mechanism of a clinical herbal formula upregulating serum HDL-c levels in treatment of metabolic syndrome by in vivo and computational studies

Decreased HDL cholesterol (HDL-c) is considered as an independent risk factor of cardiovascular disease in metabolic syndrome (Mets). Wendan decoction (WDD), a famous clinical traditional Chinese medicine formula in Mets in China, which can obviously up-regulate serum HDL-c levels in Mets. However, till now, the molecular mechanism of up-regulation still remained unclear. In this study, an integrated approach that combined serum ABCA1 in vivo assay, QSAR modeling and molecular docking was developed to explore the molecular mechanism and chemical substance basis of WDD upregulating HDL-c levels. Compared with Mets model group, serum ABCA1 and HDL-c levels intervened by two different doses of WDD for two weeks were significantly up-regulated. Then, kohonen and LDA were applied to develop QSAR models for ABCA1 up-regulators based flavonoids. The derived QSAR model produced the overall accuracy of 100%, a very powerful tool for screening ABCA1 up-regulators. The QSAR model prediction revealed 67 flavonoids in WDD were ABCA1 up-regulators. Finally, they were subjected to the molecular docking to understand their roles in up-regulating ABCA1 expression, which led to discovery of 23 ABCA1 up-regulators targeting LXR beta. Overall, QSAR modeling and docking studies well accounted for the observed in vivo activities of ABCA1 affected by WDD.     

In silico study of M18 aspartyl amino peptidase (M18AAP) of Plasmodium vivax as an antimalarial drug target

Plasmodium vivax (Pv) is the second most malaria causing pathogen among Plasmodium species. M18 aspartic aminopeptidase (M18AAP) protein is a single gene copy present in Plasmodium. This protein is functional at the terminal stage of hemoglobin degradation of host and completes the hydrolysis process which makes it an important target for new chemotherapeutics. No experimental and structural study on M18AAP protein of P. vivax is reported till today. This paper advocates the application of multiple computational approaches like protein model prediction, ligand-based 3D QSAR study, pharmacophore, structure-based virtual screening and molecular docking simulation for identification of potent lead molecules against the enzyme. The 3D QSAR model was developed using known bioactive compounds against the PvM18AAP protein which statistically signify the k-NN model with q^2 = 0.7654. The study reports a lead molecule from ligand-centric approach with good binding affinity and possessing lowest docking score. The findings will be helpful for in-vivo and in-vitro validations and development of potent anti-malarial molecules against the drug resistant strains of malaria parasite.     

3D-QSAR studies on fluroquinolones derivatives as inhibitors for tuberculosis

A quantitative structure activity relationship (QSAR) study was performed on the fluroquinolones known to have anti-tuberculosis activity. The 3D-QSAR models were generated using stepwise variable selection of the four methods - multiple regression (MR), partial least square regression (PLSR), principal component regression (PCR) and artificial neural networks (kNN-MFA). The statistical result showed a significant correlation coefficient q(2) (90%) for MR model and an external test set of (pred_r(2)) -1.7535, though the external predictivity showed to improve using kNN-MFA method with pred_r(2) of -0.4644. Contour maps showed that steric effects dominantly determine the binding affinities. The QSAR models may lead to a better understanding of the structural requirements of anti-tuberculosis compounds and also help in the design of novel molecules.     

QSAR and complex network study of the chiral HMGR inhibitor structural diversity

Efficient drugs such as statins or mevinic acids are inhibitors of the rate-limiting enzyme of cholesterol biosynthesis, 3-hydroxy-3-methyl-glutaryl coenzyme A reductase (HMGR), an enzyme responsible for the double reduction of 3-hydroxy-3-methyl-glutaryl coenzyme A into mevalonic acid. These compounds promoted the synthesis and evaluation of new inhibitors for HMGR, named HMGRIs. The high number of possible candidates creates the necessity of Quantitative Structure–Activity Relationship models in order to guide the HMGRI synthesis. There are two main problems of the reported QSAR models: the homogeneous series of the compounds and the chirality of many candidates. In this work, we propose for the first time a QSAR model for a very large and heterogeneous series of HMGRIs. The model is based on the Topological Indices (TIs) of molecular structures. Using the predictions of this model as input, we construct the first complex network that describes the drug–drug similarity relationships for more than 1600 experimentally non-explored chiral HMGRIs isomers. We also presented a reduced version of this network (Giant Component) that contains the most representative set of chiral HMGRI candidates. The work suggests a new mixed application in the QSAR study of relevant aspects of structural diversity by using chiral/non-chiral TIs, combined with complex networks.     

Binding affinity prediction of novel estrogen receptor ligands using receptor-based 3-D QSAR methods

We have recently reported the development of a 3-D QSAR model for estrogen receptor ligands showing a significant correlation between calculated molecular interaction fields and experimentally measured binding affinity. The ligand alignment obtained from docking simulations was taken as basis for a comparative field analysis applying the GRID/GOLPE program. Using the interaction field derived with a water probe and applying the smart region definition (SRD) variable selection procedure, a significant and robust model was obtained (q²_LOO=0.921, SDEP=0.345). To further analyze the robustness and the predictivity of the established model several recently developed estrogen receptor ligands were selected as external test set. An excellent agreement between predicted and experimental binding data was obtained indicated by an external SDEP of 0.531. Two other traditionally used prediction techniques were applied in order to check the performance of the receptor-based 3-D QSAR procedure. The interaction energies calculated on the basis of receptor–ligand complexes were correlated with experimentally observed affinities. Also ligand-based 3-D QSAR models were generated using program FlexS. The interaction energy-based model, as well as the ligand-based 3-D QSAR models yielded models with lower predictivity. The comparison with the interaction energy-based model and with the ligand-based 3-D QSAR models, respectively, indicates that the combination of receptor-based and 3-D QSAR methods is able to improve the quality of prediction.     

A model of autocatalytic replication

Summary The catalytic effects that existing polymer chains have on the formation of new chains are modeled using ideas from spin glasses and neural networks. Computer simulation shows that isolated groups of chains in this model are capable of accurately replicating a wide variety of complex structures without templating. Replication in the model arises spontaneously and rapidly, leading to an extremely simple realization of a system exhibiting Darwinian evolution.     

QSAR studies and pharmacophore identification for arylsubstituted cycloalkenecarboxylic acid methyl esters with affinity for the human dopamine transporter

Data from a series of 29 monoamine transport inhibitors were used to generate 2D and 3D QSAR models for their binding affinity to the human dopamine transporter (hDAT). Among the inhibitors were many non-nitrogen containing compounds. The 2D QSAR analysis resulted in the equation -logK(i)=4.00-3.93E(LUMO)-0.67E(HOMO)-3.24sigma(p), which predicted the importance of electron withdrawing groups in the aromatic moiety. However, the model failed to predict the observed poor binding of nitro-substituted compounds. In contrast, a derived 3D QSAR model was capable of predicting these more correctly.     

Analysis of metabolic network based on conservation of molecular structure

Recent work has revealed much about chemical reactions inside hundreds of organisms as well as universal characteristics of metabolic networks, which shed light on the evolution of the networks. However, characteristics of individual metabolites have been neglected. For example, some carbohydrates have structures that are decomposed into small molecules by metabolic reactions, but coenzymes such as ATP are mostly preserved. Such differences in metabolite characteristics are important for understanding the universal characteristics of metabolic networks. To quantify the structure conservation of metabolites, we defined the "structure conservation index" (SCI) for each metabolite as the fraction of metabolite atoms restored to their original positions through metabolic reactions. As expected, coenzymes and coenzyme-like metabolites that have reaction loops in the network show a higher SCI. Using the index, we found that the sum of metabolic fluxes is negatively correlated with the structure preservation of metabolite. Also, we found that each reaction path around high SCI metabolites changes independently, while changes in reaction paths involving low SCI metabolites coincide through evolution processes. These correlations may provide a clue to universal properties of metabolic networks.     

Developing potential Helicobacter pylori urease inhibitors from novel oxoindoline derivatives: Synthesis,biological evaluation and in silico study

By recruiting the important moiety from Shikonin, a series of novel oxoindoline derivatives S1–S20 have been synthesized for inhibiting H. pylori urease. The most potent compound S18 displayed better activity (IC₅₀?=?0.71?μM; MIC?=?0.48?μM) than the positive controls AHA (IC₅₀?=?17.2?μM) and Metronidazole (MIC?=?31.3?μM). With low cytotoxicity, it showed considerable potential for further development. Docking simulation revealed the possible binding pattern of this series. 3D QSAR model was built to discuss SAR and give useful hints for future modification.     

Why fluorescent probes for endoplasmic reticulum are selective: an experimental and QSAR-modelling study

The basis of the selectivity of fluorochromes routinely used to visualize the endoplasmic reticulum (ER) in live cells remains obscure. To clarify this, interactions of living cells with fluorochromes of varied physicochemical properties were analyzed experimentally and numerically using a quantitative structure activity relationship analysis (QSAR). Routine selective ER probes were found to be amphipathic, lipophilic cations with moderate-sized conjugated systems. The moderately lipophilic character permits probe uptake by passive diffusion without nonspecific accumulation in biomembranes. The moderately amphipathic character favors uptake into the ER, perhaps owing to its high concentration of zwitterionic lipid head-groups. The QSAR model rationalizes the impractical character of some ER probes mentioned in the literature, and could permit design of novel ER probes with different emission colors. The possibility of using the QSAR model as a tool to predict the accumulation of xenobiotics in the ER of living cells is illustrated by the localization of certain antipsychotic drugs in cultured cells.