Combinatorial Chemistry & High Throughput Screening - Volume 19, Issue 2, 2016

Lung cancer, characterized by uncontrolled cell growth in tissues of the lung, is one of the leading causes of cancer mortality worldwide. Many etiologic factors for lung cancer tumorigenesis have been identified to date, such as smoking and exposure to radon, cooking fumes and asbestos. Atmospheric pollution has become increasingly heavy in China in recent years. Accordingly, greater numbers of people are paying attention to the air quality around them. PM2.5 (particulate matter with a diameter of 2.5 micrometers or less), which is one of the most important indicators for measuring air quality, can penetrate and be retained in lung tissue. It is believed that PM2.5 may represent a new type of etiological factor for lung cancer. This study constitutes the analysis of the association between PM2.5 and lung cancer. Genes related to small/nonsmall cell lung cancer were evaluated by assigning scores to measure the impact caused by PM2.5. Analyses of small/nonsmall cell lung cancer genes with high scores revealed that it is theoretically possible that PM2.5 is an etiologic factor for lung cancer. Our results provided new insights of the relationship between lung cancers and air pollution.

Identification of potential druggable targets utilizing protein-protein interactions network (PPIN) has been emerging as a hotspot in drug discovery and development research. However, it remains unclear whether the currently used PPIN topological properties are enough to discriminate the drug targets from non-drug targets. In this study, three-step classification models using different network topological properties were designed and implemented using support vector machine (SVM) to compare the enrichment of known drug targets from non-targets. Surprisingly, none of the models was able to identify more than 75% of the true targets in the test set. It appears that the currently used simple PPIN topological properties are not likely robust enough for prediction of potential drug targets with high confidence, which also echoes similarly unsatisfying prediction data reported previously. However, we proposed that quality and quantity improvement of the protein-protein interactions (PPI) data for model training will help increasing the prediction accuracy.

It is crucial to identify the molecular targets of a compound during the course of the new drug discovery and drug development. Due to the complexity of biological systems, finding drug targets by biological experiments is very tedious and expensive. In the paper, we used chemicalchemical interactions in the STITCH database to construct a network of drug-drug association. Based on the network, a learning method keeping local and global consistency was presented to infer drug targets. We achieved an accuracy of 57.75% in the first order prediction using leave-one-out cross validation, which was higher than the accuracy of 53.77% achieved by the local neighbor model. We manually validated 27 absent drug targets in the crossvalidation using drug-target interactions from other databases. Applying the presented method to large-scale prediction of unknown targets, we manually confirmed 14 pairs of drug-target interactions among the newly predicted drug targets. These results suggested that the presented method was a promising tool for large-scale identification of drug targets.

Drug-target interaction is an important topic in drug discovery and drug repositioning. KEGG database offers a drug annotation and classification using a target-based classification system. In this study, we gave an investigation on five target-based classes: (I) G protein-coupled receptors; (II) Nuclear receptors; (III) Ion channels; (IV) Enzymes; (V) Pathogens, using molecular descriptors to represent each drug compound. Two popular feature selection methods, maximum relevance minimum redundancy and incremental feature selection, were adopted to extract the important descriptors. Meanwhile, an optimal prediction model based on nearest neighbor algorithm was constructed, which got the best result in identifying drug target-based classes. Finally, some key descriptors were discussed to uncover their important roles in the identification of drug-target classes.

A metabolic pathway is a series of biological processes providing necessary molecules and energies for an organism, which could be essential to the lives of the living organisms. Most metabolic pathways require the involvement of compounds and given a compound it is helpful to know what types of metabolic pathways the compound participates in. In this study, compounds are first represented by molecular fragments which are then delivered to a prediction engine called Sequential Minimal Optimization (SMO) for predictions. Maximum relevance and minimum redundancy (mRMR) and incremental feature selection are adopted to extract key features based on which an optimal prediction engine is established. The proposed method is effective comparing to the random forest, Dagging and a popular method that integrating chemical-chemical interactions and chemical-chemical similarities. We also make predictions using some compounds with unknown metabolic pathways and choose 17 compounds for analysis. The results indicate that the method proposed may become a useful tool in predicting and analyzing metabolic pathways.

Most essential functions are associated with various protein–protein interactions, particularly the cytokine–receptor interaction. Knowledge of the heterogeneous network of cytokine– receptor interactions provides insights into various human physiological functions. However, only a few studies are focused on the computational prediction of these interactions. In this study, we propose a novel machine-learning-based method for predicting cytokine–receptor interactions. A protein sequence is first transformed by incorporating the sequence evolutional information and then formulated with the following three aspects: (1) the k-skip-n-gram model, (2) physicochemical properties, and (3) local pseudo position-specific score matrix (local PsePSSM). The random forest classifier is subsequently employed to predict potential cytokine–receptor interactions. Experimental results on a dataset of Homo sapiens show that the proposed method exhibits improved performance, with 3.4% higher overall prediction accuracy, than existing methods.

In the present study, molecular descriptors and physicochemical properties were used to encode drug molecules. Based on this molecular representation method, Random forest was applied to construct a drug-drug combination network. After feature selection, an optimal features subset was built, which described the main factors of drugs in our prediction. As a result, the selected features can be clustered into three categories: elemental analysis, chemistry, and geometric features. And all of the three types features are essential elements of the drug-drug combination network. The final prediction model achieved a Matthew's correlation coefficient (MCC) of 0.5335 and an overall prediction accuracy of 88.79% for the 10-fold cross-validation test.

The functional screening of compounds is an important topic in chemistry and biomedicine that can uncover the essential properties of compounds and provide information concerning their correct use. In this study, we investigated the bioactive compounds reported in Selleckchem, which were assigned to 22 pathways. A computational method was proposed to identify the pathways of the bioactive compounds. Unlike most existing methods that only consider compound structural information, the proposed method adopted both the structural and interaction information from the compounds. The total accuracy achieved by our method was 61.79% based on jackknife analysis of a dataset of 1,832 bioactive compounds. Its performance was quite good compared with that of other machine learning algorithms (with total accuracies less than 46%). Finally, some of the false positives obtained by the method were analyzed to investigate the likelihood of compounds being annotated to new pathways.

The order of reaction, especially in 1,3-dipolar cycloadditions directly affects the products' stereo selectivity. Due to this fact that a wide range of heterocyclic rings of natural products and biologically active molecules are synthesizing via this valuable procedure, understanding about the order of this reaction is so useful in designing the synthesis of different types of heterocyclic species. Therefore, the order of 1, 3-dipolar reaction has been carefully studied by many researchers but it seems that this question is still open despite many valuable answers. Considering this, in the present work, it is attempted to pursue this subject by theoretical investigation of any possible pathway of 1, 3-dipolar reaction of tetra amino ethylene as a highly electron rich dipolarophile and trifluoro methyl azide as an electron poor 1,3-dipole. During the calculations, one, two, and three step mechanism(s) have been found to be possible for the present 1, 3-dipolar reaction.