In this paper, we focus on generating realistic images from text descriptions. Current methods first generate an initial image with rough shape and color, and then refine the initial image to a high-resolution one. Most existing text-to-image synthesis methods have two main problems.(1) These methods depend heavily on the quality of the initial images. If the initial image is not well initialized, the following processes can hardly refine the image to a satisfactory quality. (2) Each word contributes a different level of importance when depicting different image contents, however, unchanged text representation is used in existing image refinement processes. In this paper, we propose the Dynamic Memory Generative Adversarial Network (DM-GAN) to generate high-quality images. The proposed method introduces a dynamic memory module to refine fuzzy image contents, when the initial images are not well generated. A memory writing gate is designed to select the important text information based on the initial image content, which enables our method to accurately generate images from the text description. We also utilize a response gate to adaptively fuse the information read from the memories and the image features. We evaluate the DM-GAN model on the Caltech-UCSD Birds 200 dataset and the Microsoft Common Objects in Context dataset. Experimental results demonstrate that our DM-GAN model performs favorably against the state-of-the-art approaches.

The protr package is freely available from CRAN: http://cran.r-project.org/package=protr, ProtrWeb, is freely available at http://protrweb.scbdd.com/.

Drug-target interactions (DTIs) are central to current drug discovery processes and public health fields. Analyzing the DTI profiling of the drugs helps to infer drug indications, adverse drug reactions, drug-drug interactions, and drug mode of actions. Therefore, it is of high importance to reliably and fast predict DTI profiling of the drugs on a genome-scale level. Here, we develop the TargetNet server, which can make real-time DTI predictions based only on molecular structures, following the spirit of multi-target SAR methodology. Naïve Bayes models together with various molecular fingerprints were employed to construct prediction models. Ensemble learning from these fingerprints was also provided to improve the prediction ability. When the user submits a molecule, the server will predict the activity of the user's molecule across 623 human proteins by the established high quality SAR model, thus generating a DTI profiling that can be used as a feature vector of chemicals for wide applications. The 623 SAR models related to 623 human proteins were strictly evaluated and validated by several model validation strategies, resulting in the AUC scores of 75-100 %. We applied the generated DTI profiling to successfully predict potential targets, toxicity classification, drug-drug interactions, and drug mode of action, which sufficiently demonstrated the wide application value of the potential DTI profiling. The TargetNet webserver is designed based on the Django framework in Python, and is freely accessible at http://targetnet.scbdd.com .

The Caco-2 cell monolayer model is a popular surrogate in predicting the in vitro human intestinal permeability of a drug due to its morphological and functional similarity with human enterocytes. A quantitative structure-property relationship (QSPR) study was carried out to predict Caco-2 cell permeability of a large data set consisting of 1272 compounds. Four different methods including multivariate linear regression (MLR), partial least-squares (PLS), support vector machine (SVM) regression and Boosting were employed to build prediction models with 30 molecular descriptors selected by nondominated sorting genetic algorithm-II (NSGA-II). The best Boosting model was obtained finally with R(2) = 0.97, RMSEF = 0.12, Q(2) = 0.83, RMSECV = 0.31 for the training set and RT(2) = 0.81, RMSET = 0.31 for the test set. A series of validation methods were used to assess the robustness and predictive ability of our model according to the OECD principles and then define its applicability domain. Compared with the reported QSAR/QSPR models about Caco-2 cell permeability, our model exhibits certain advantage in database size and prediction accuracy to some extent. Finally, we found that the polar volume, the hydrogen bond donor, the surface area and some other descriptors can influence the Caco-2 permeability to some extent. These results suggest that the proposed model is a good tool for predicting the permeability of drug candidates and to perform virtual screening in the early stage of drug development.

With the increase of complexity and risk in drug discovery processes, human intestinal absorption (HIA) prediction has become more and more important. Up to now, some predictive models have been constructed to estimate HIA of new drug-like compounds with acceptable accuracies, but there are still some issues to be explored including the limited and unbalanced HIA data, the performance of different types of descriptors and the application domain issues of published models. To address these problems, in this study, we collected a relatively large dataset consisting of 970 compounds, and 9 different types of descriptors were calculated for further modeling. For all the modeling processes, a parameter named samplesize in the random forest (RF) method was applied to balance the dataset. And then, classification models were established based on different training sets and different combinations of descriptors. published models, our model exhibits some advantages in data size, model accuracy and model practicability to some extent. This structure-activity relationship model is necessary and useful for HIA prediction and it could be a convenient tool for virtual screening in the early stage of drug development.

Urban data is massive, heterogeneous, and spatio-temporal, posing a substantial challenge for visualization and analysis. In this paper, we design and implement a novel visual analytics approach, Visual Analyzer for Urban Data (VAUD), that supports the visualization, querying, and exploration of urban data. Our approach allows for cross-domain correlation from multiple data sources by leveraging spatial-temporal and social inter-connectedness features. Through our approach, the analyst is able to select, filter, aggregate across multiple data sources and extract information that would be hidden to a single data subset. To illustrate the effectiveness of our approach, we provide case studies on a real urban dataset that contains the cyber-, physical-, and social- information of 14 million citizens over 22 days.

Although a wide variety of machine learning (ML) algorithms have been utilized to learn quantitative structure–activity relationships (QSARs), there is no agreed single best algorithm for QSAR learning. Therefore, a comprehensive understanding of the performance characteristics of popular ML algorithms used in QSAR learning is highly desirable. In this study, five linear algorithms [linear function Gaussian process regression (linear-GPR), linear function support vector machine (linear-SVM), partial least squares regression (PLSR), multiple linear regression (MLR) and principal component regression (PCR)], three analogizers [radial basis function support vector machine (rbf-SVM), K-nearest neighbor (KNN) and radial basis function Gaussian process regression (rbf-GPR)], six symbolists [extreme gradient boosting (XGBoost), Cubist, random forest (RF), multiple adaptive regression splines (MARS), gradient boosting machine (GBM), and classification and regression tree (CART)] and two connectionists [principal component analysis artificial neural network (pca-ANN) and deep neural network (DNN)] were employed to learn the regression-based QSAR models for 14 public data sets comprising nine physicochemical properties and five toxicity endpoints. The results show that rbf-SVM, rbf-GPR, XGBoost and DNN generally illustrate better performances than the other algorithms. The overall performances of different algorithms can be ranked from the best to the worst as follows: rbf-SVM > XGBoost > rbf-GPR > Cubist > GBM > DNN > RF > pca-ANN > MARS > linear-GPR ≈ KNN > linear-SVM ≈ PLSR > CART ≈ PCR ≈ MLR. In terms of prediction accuracy and computational efficiency, SVM and XGBoost are recommended to the regression learning for small data sets, and XGBoost is an excellent choice for large data sets. We then investigated the performances of the ensemble models by integrating the predictions of multiple ML algorithms. The results illustrate that the ensembles of two or three algorithms in different categories can indeed improve the predictions of the best individual ML algorithms.

Lipophilicity, evaluated by either n-octanol/water partition coefficient or n-octanol/buffer solution distribution coefficient, is of high importance in pharmacology, toxicology, and medicinal chemistry. A quantitative structureproperty relationship study was carried out to predict distribution coefficients at pH 7.4 (logD 7.4 ) of a large data set consisting of 1130 organic compounds. Partial least squares and support vector machine (SVM) regressions were employed to build prediction models with 30 molecular descriptors selected by genetic algorithm. The obtained results demonstrated that the SVM model is more reliable and has a better prediction performance than the partial least squares model. The square correlation coefficients of fitting, cross validation, and prediction are 0.92, 0.90, and 0.89, respectively. The corresponding root mean square errors are 0.52, 0.59, and 0.56, respectively. The robustness, reliability, and generalization ability of the model were assessed by Y-randomization test and applicability domain. When compared with logD 7.4 values calculated by five existing methods from Discovery Studio and ChemAxon, our SVM model shows superiority over them. The results indicated that our model could give a reliable and robust prediction of logD 7.4 .