Search

Accurate identification of DNA-binding proteins (DBPs) is critical for both understanding protein function and drug design. DBPs also play essential roles in different kinds of biological activities such as DNA replication, repair, transcription, and splicing. As experimental identification of DBPs is time-consuming and sometimes biased toward prediction, constructing an effective DBP model represents an urgent need, and computational methods that can accurately predict potential DBPs based on sequence information are highly desirable. In this paper, a novel predictor called DeepDNAbP has been developed to accurately predict DBPs from sequences using a convolutional neural network (CNN) model. First, we perform three feature extraction methods, namely position-specific scoring matrix (PSSM), pseudo-amino acid composition (PseAAC) and tripeptide composition (TPC), to represent protein sequence patterns. Secondly, SHapley Additive exPlanations (SHAP) are employed to remove the redundant and irrelevant features for predicting DBPs. Finally, the best features are provided to the CNN classifier to construct the DeepDNAbP model for identifying DBPs. The final DeepDNAbP predictor achieves superior prediction performance in K-fold cross-validation tests and outperforms other existing predictors of DNA-protein binding methods. DeepDNAbP is poised to be a powerful computational resource for the prediction of DBPs. The web application and curated datasets in this study are freely available at: http://deepdbp.sblog360.blog/., (Copyright © 2022. Published by Elsevier Ltd.)

Leveraging on the underlying low-dimensional structure of data, low-rank and sparse modeling approaches have achieved great success in a wide range of applications. However, in many applications the data can display structures beyond simply being low-rank or sparse. Fully extracting and exploiting hidden structure information in the data is always desirable and favorable. To reveal more underlying effective manifold structure, in this paper, we explicitly model the data relation. Specifically, we propose a structure learning framework that retains the pairwise similarities between the data points. Rather than just trying to reconstruct the original data based on self-expression, we also manage to reconstruct the kernel matrix, which functions as similarity preserving. Consequently, this technique is particularly suitable for the class of learning problems that are sensitive to sample similarity, e.g., clustering and semisupervised classification. To take advantage of representation power of deep neural network, a deep auto-encoder architecture is further designed to implement our model. Extensive experiments on benchmark data sets demonstrate that our proposed framework can consistently and significantly improve performance on both evaluation tasks. We conclude that the quality of structure learning can be enhanced if similarity information is incorporated., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2020 Elsevier Ltd. All rights reserved.)

Background: This study intends to establish a combined prediction model that integrates the clinical symptoms,the lung lesion volume, and the radiomics features of patients with COVID-19, resulting in a new model to predict the severity of COVID-19. Methods: The clinical data of 386 patients with COVID-19 at several hospitals, as well as images of certain patients during their hospitalization, were collected retrospectively to create a database of patients with COVID-19 pneumonia. The contour of lungs and lesion locations may be retrieved from CT scans using a CT-image-based quantitative discrimination and trend analysis method for COVID-19 and the Mask R-CNN deep neural network model to create 3D data of lung lesions. The quantitative COVID-19 factors were then determined, on which the diagnosis of the development of the patients' symptoms could be established. Then, using an artificial neural network, a prediction model of the severity of COVID-19 was constructed by combining characteristic imaging features on CT slices with clinical factors. ANN neural network was used for training, and tenfold cross-validation was used to verify the prediction model. The diagnostic performance of this model is verified by the receiver operating characteristic (ROC) curve. Results: CT radiomics features extraction and analysis based on a deep neural network can detect COVID-19 patients with an 86% sensitivity and an 85% specificity. According to the ROC curve, the constructed severity prediction model indicates that the AUC of patients with severe COVID-19 is 0.761, with sensitivity and specificity of 79.1% and 73.1%, respectively. Conclusions: The combined prediction model for severe COVID-19 pneumonia, which is based on deep learning and integrates clinical aspects, pulmonary lesion volume, and radiomics features of patients, has a remarkable differential ability for predicting the course of disease in COVID-19 patients. This may assist in the early prevention of severe COVID-19 symptoms. [ABSTRACT FROM AUTHOR]

Graph clustering has received significant attention in recent years due to the breakthrough of graph neural networks (GNNs). However, GNNs frequently assume strong data homophily, which is not true in many real-world applications. Furthermore, practical graphs are typically noisy and sparse, which inevitably degrades the clustering performance. To this end, we propose a novel Contrastive Graph Clustering (CGC) method with adaptive filter framework. We first design an adaptive filter that can automatically learn a suitable filter for different data, mining holistic information beyond low-frequency components and encoding topology structure information into features. Afterward, we learn a refined graph based on a graph-level contrastive mechanism, which further boosts graph discriminability. Extensive experiments show that the proposed CGC method achieves significant improvement over state-of-the-art methods on several benchmark datasets. In particular, our simple method, which does not employ neural networks, outperforms many deep learning approaches. • High-frequency information is considered for graph clustering. • Contrastive regularizer is applied to refine graph structure. • Homophilic and heterophilic graphs are evaluated. [ABSTRACT FROM AUTHOR]

• A new deep learning framework is proposed in this paper, which can leverage the object label inter-dependent for weakly supervised learning. • We creatively introduce a novel initialization method of label embeddings for inter-relationships learning. • Ablation studies of our model are provided in detail. • Our models can be applied to object classification and weakly supervised pointwise object localization. In computer vision, the research community has been looking to how to benefit from weakly supervised learning that utilizes easily obtained image-level labels to train neural network models. The existing deep convolutional neural networks for weakly supervised learning, however, generally do not fully exploit the label dependencies in an image. To make full use of this information, in this paper, we propose a new framework for weakly supervised learning of deep convolutional neural networks, introducing graph convolutional networks to capture the semantic label co-occurrence in an image. Moreover, we propose a novel initialization method for label embedding in graph convolutional networks, which enables a smoother optimization for interrelationships learning. Extensive experiments and comparisons on four public benchmark datasets (PASCAL VOC 2007, PASCAL VOC 2012, Microsoft COCO, and NUS-WIDE) show the superior performance of our approach in both image classification and weakly supervised pointwise object localization. These results lead us to conclude that the label dependencies in the input image can provide valuable evidence for learning strongly localized features. [ABSTRACT FROM AUTHOR]

Recommender systems aim to maximize the overall accuracy for long-term recommendations. However, most of the existing recommendation models adopt a static view, and ignore the fact that recommendation is a dynamic sequential decision-making process. As a result, they fail to adapt to new situations and suffer from the cold-start problem. Although sequential recommendation methods have been gaining attention recently, the objective of long-term recommendation still has not been explicitly addressed since these methods are developed for short-term prediction situations. To overcome these problems, we propose a novel top-N interactive recommender system based on deep reinforcement learning. In our model, the processes of recommendation are viewed as Markov decision processes (MDP), wherein the interactions between agent (recommender system) and environment (user) are simulated by the recurrent neural network (RNN). In addition, reinforcement learning is employed to optimize the proposed model for the purpose of maximizing long-term recommendation accuracy. Experimental results based on several benchmarks show that our model significantly outperforms previous top-N methods in terms of Hit-Rate and NDCG for the long-term recommendation, and can be applied to both cold-start and warm-start scenarios. • A novel top-N interactive recommender system based on deep reinforcement learning is proposed. • The interactions between recommender system and users are simulated by recurrent neural networks. • The proposed model can deal with both cold-start and warm-start scenarios. • Reinforcement learning and supervised learning are employed to optimize the proposed model for long-term recommendation accuracy. • Experiments and comparisons are conducted to show the merits of the proposed model. [ABSTRACT FROM AUTHOR]