Prediction of 5-hydroxytryptamine transporter inhibitors based on machine learning

In patients with depression, the use of 5-HT reuptake inhibitors can improve the condition. Machine learning methods can be used in ligand-based activity prediction processes. In order to predict SERT inhibitors, the SERT inhibitor data from the ChEMBL database was screened and pre-processed. Then 4 machine learning methods (LR, SVM, RF, and KNN) and 4 molecular fingerprints (CDK, Graph, MACCS, and PubChem) were used to build 16 prediction models. The top 5 models of accuracy (Q) in the cross-validation of training set were used to build three different ensemble learning models. In the test1 set, the VOT_CLF3 model had the largest SP (0.871), Q (0.869), AUC (0.919), and MCC (0.728). In the unbalanced test2 set, VOT_CLF3 had the largest SE (0.857), SP (0.867), Q (0.865) and MCC (0.639). VOT_CLF3 was recommended for the virtual screening process of SERT inhibitors. In addition, 12 molecular structural alerts that frequently appear in SERT inhibitors were found (P < 0.05), which provided important reference value for the design work of SERT inhibitors.     

How to Separate Kinase Inhibition from Undesired Monoamine Oxidase A Inhibition—The Development of the DYRK1A Inhibitor AnnH75 from the Alkaloid Harmine

The β-carboline alkaloid harmine is a potent DYRK1A inhibitor, but suffers from undesired potent inhibition of MAO-A, which strongly limits its application. We synthesized more than 60 analogues of harmine, either by direct modification of the alkaloid or by de novo synthesis of β-carboline and related scaffolds aimed at learning about structure–activity relationships for inhibition of both DYRK1A and MAO-A, with the ultimate goal of separating desired DYRK1A inhibition from undesired MAO-A inhibition. Based on evidence from published crystal structures of harmine bound to each of these enzymes, we performed systematic structure modifications of harmine yielding DYRK1A-selective inhibitors characterized by small polar substituents at N-9 (which preserve DYRK1A inhibition and eliminate MAO-A inhibition) and beneficial residues at C-1 (methyl or chlorine). The top compound AnnH75 remains a potent DYRK1A inhibitor, and it is devoid of MAO-A inhibition. Its binding mode to DYRK1A was elucidated by crystal structure analysis, and docking experiments provided additional insights for this attractive series of DYRK1A and MAO-A inhibitors.     

Recent research and development of DYRK1A inhibitors

Dual specificity tyrosine phosphorylation regulated kinase 1 A(DYRK1 A) is an evolutionarily conserved protein kinase belonging to the CMGC kinase family, which is closely related to Down syndrome(DS)and Alzheimer’s disease(AD). In recent years, not only the treatment of diabetes, but also the treatment of cancer gradually focuses on targeting DYRK1 A. Therefore, a series of DYRK1 A inhibitors have been developed to treat relevant diseases and clarify their treatment mechanism furtherly. DYRK1 A...     

Design of (quinolin-4-ylthio)carboxylic acids as new Escherichia coli DNA gyrase B inhibitors: machine learning studies,molecular docking,synthesis and biological testing

Spread of multidrug‐resistant Escherichia coli clinical isolates is a main problem in the treatment of infectious diseases. Therefore, the modern scientific approaches in decision this problem require not only a prevention strategy, but also the development of new effective inhibitory compounds with selective molecular mechanism of action and low toxicity. The goal of this work is to identify more potent molecules active against E. coli strains by using machine learning, docking studies, synthesis and biological evaluation. A set of predictive QSAR models was built with two publicly available structurally diverse data sets, including recent data deposited in PubChem. The predictive ability of these models tested by a 5-fold cross-validation, resulted in balanced accuracies (BA) of 59–98% for the binary classifiers. Test sets validation showed that the models could be instrumental in predicting the antimicrobial activity with an accuracy (with BA = 60–99 %) within the applicability domain. The models were applied to screen a virtual chemical library, which was designed to have activity against resistant E. coli strains. The eight most promising compounds were identified, synthesized and tested. All of them showed the different levels of anti-E. coli activity and acute toxicity. The docking results have shown that all studied compounds are potential DNA gyrase inhibitors through the estimated interactions with amino acid residues and magnesium ion in the enzyme active center The synthesized compounds could be used as an interesting starting point for further development of drugs with low toxicity and selective molecular action mechanism against resistant E. coli strains. The developed QSAR models are freely available online at OCHEM http://ochem.eu/article/112525 and can be used to virtual screening of potential compounds with anti-E. coli activity.     

Synthesis of functionalized amides of 2-(arylamino)pyrimidine series

New amides of 2-(arylamino)pyrimidine series were synthesized with pharmacophoric fragments of tyrosine kinase and histone deacetylase inhibitors and functional groups providing chemosorption of compounds on nanocarriers.     

Using random forest to classify T-cell epitopes based on amino acid properties and molecular features

T-lymphocyte (T-cell) is a very important component in human immune system. T-cell epitopes can be used for the accurately monitoring the immune responses which activation by major histocompatibility complex (MHC), and rationally designing vaccines. Therefore, accurate prediction of T-cell epitopes is crucial for vaccine development and clinical immunology. In current study, two types peptide features, i.e., amino acid properties and chemical molecular features were used for the T-cell epitopes peptide representation. Based on these features, random forest (RF) algorithm, a powerful machine learning algorithm, was used to classify T-cell epitopes and non-T-cell epitopes. The classification accuracy, sensitivity, specificity, Matthews correlation coefficient (MCC), and area under the curve (AUC) values for proposed method are 97.54%, 97.22%, 97.60%, 0.9193, and 0.9868, respectively. These results indicate that current method based on the combined features and RF is effective for T-cell epitopes prediction.     

Novel and Potential Small Molecule Scaffolds as DYRK1A Inhibitors by Integrated Molecular Docking-Based Virtual Screening and Dynamics Simulation Study

The dual-specificity tyrosine phosphorylation-regulated kinase 1A (DYRK1A) is a novel, promising and emerging biological target for therapeutic intervention in neurodegenerative diseases, especially in Alzheimer’s disease (AD). The molMall database, comprising rare, diverse and unique compounds, was explored for molecular docking-based virtual screening against the DYRK1A protein, in order to find out potential inhibitors. Ligands exhibiting hydrogen bond interactions with key amino acid residues such as Ile165, Lys188 (catalytic), Glu239 (gk+1), Leu241 (gk+3), Ser242, Asn244, and Asp307, of the target protein, were considered potential ligands. Hydrogen bond interactions with Leu241 (gk+3) were considered key determinants for the selection. High scoring structures were also docked by Glide XP docking in the active sites of twelve DYRK1A related protein kinases, viz. DYRK1B, DYRK2, CDK5/p25, CK1, CLK1, CLK3, GSK3β, MAPK2, MAPK10, PIM1, PKA, and PKCα, in order to find selective DYRK1A inhibitors. MM/GBSA binding free energies of selected ligand–protein complexes were also calculated in order to remove false positive hits. Physicochemical and pharmacokinetic properties of the selected six hit ligands were also computed and related with the proposed limits for orally active CNS drugs. The computational toxicity webserver ProTox-II was used to predict the toxicity profile of selected six hits (molmall IDs 9539, 11352, 15938, 19037, 21830 and 21878). The selected six docked ligand–protein systems were exposed to 100 ns molecular dynamics (MD) simulations to validate their mechanism of interactions and stability in the ATP pocket of human DYRK1A kinase. All six ligands were found to be stable in the ATP binding pocket of DYRK1A kinase.     

Application of machine learning algorithms to screen potential biomarkers under cadmium exposure based on human urine metabolic profiles

Exposure to environmental cadmium increases the health risk of residents. Early urine metabolic detection using high-resolution mass spectrometry and machine learning algorithms would be advantageous to predict the adverse health effects. Here, we conducted machine learning approaches to screen potential biomarkers under cadmium exposure in 403 urine samples. In positive and negative ionization mode, 4207 and 3558 features were extracted, respectively. We compared seven machine learning algorithms and found that the extreme gradient boosting (XGBoost) and random forest (RF) classifiers showed better accuracy and predictive performance than others. Following 5-fold cross-validation, the value of area under curve (AUC) was both 0.93 for positive and negative ionization modes in XGBoost classifier. In the RF classifier, AUC were 0.80 and 0.84 for positive and negative ionization modes, respectively. We then identified a biomarker panel based on XGBoost and RF classifiers. The incorporation of machine learning models into urine analysis using high-resolution mass spectrometry could allow a convenient assessment of cadmium exposure.