Abstract Image forgery localization identifies tampered regions within an image by extracting distinctive forgery features. Current methods mainly use convolutional neural networks (CNNs) to extract features. However, CNNs’ limited receptive field emphasizes local features, impeding the global modeling of crucial lower-level features like edges and textures, leading to decreased precision. Moreover, prior methods use pyramid networks for multi-scale feature extraction but show deficiencies in multi-scale and interlayer modeling, leading to inadequate multi-scale information representation and limiting flexibility to tampered regions of varying sizes. To address these issues, this paper proposes a Transformer-based model integrating multi-scale and boundary features. The model employs a Pyramid Vision Transformer as the encoder, using self-attention over convolution to enhance global context modeling. Building on this, the model incorporates a multi-scale feature enhancement module that enriches forgery features by paralleling various convolutional layers. Features at various encoder stages are integrated through a cross-stage interaction module, enabling multi-level feature correlation for a strong feature representation. Furthermore, the model includes a forgery boundary information-guided branch, which focuses precisely on tampered region structures without introducing irrelevant noise. Experiments demonstrate that our model surpasses previous methods in localization accuracy, with F1 and AUC improving by 8.5% and 2.2% in pre-training, respectively.

Network pharmacology (NP) provides a new methodological perspective for understanding traditional medicine from a holistic perspective, giving rise to frontiers such as traditional Chinese medicine network pharmacology (TCM-NP). With the development of artificial intelligence (AI) technology, it is key for NP to develop network-based AI methods to reveal the treatment mechanism of complex diseases from massive omics data. In this review, focusing on the TCM-NP, we summarize involved AI methods into three categories: network relationship mining, network target positioning and network target navigating, and present the typical application of TCM-NP in uncovering biological basis and clinical value of Cold/Hot syndromes. Collectively, our review provides researchers with an innovative overview of the methodological progress of NP and its application in TCM from the AI perspective.

MOTIVATION: Drug repositioning (DR), identifying novel indications for approved drugs, is a cost-effective strategy in drug discovery. Despite numerous proposed DR models, integrating network-based features, differential gene expression, and chemical structures for high-performance DR remains challenging. RESULTS: We propose a comprehensive deep pretraining and fine-tuning framework for DR, termed DrugRepPT. Initially, we design a graph pretraining module employing model-augmented contrastive learning on a vast drug-disease heterogeneous graph to capture nuanced interactions and expression perturbations after intervention. Subsequently, we introduce a fine-tuning module leveraging a graph residual-like convolution network to elucidate intricate interactions between diseases and drugs. Moreover, a Bayesian multiloss approach is introduced to balance the existence and effectiveness of drug treatment effectively. Extensive experiments showcase the efficacy of our framework, with DrugRepPT exhibiting remarkable performance improvements compared to SOTA (state of the arts) baseline methods (improvement 106.13% on Hit@1 and 54.45% on mean reciprocal rank). The reliability of predicted results is further validated through two case studies, i.e. gastritis and fatty liver, via literature validation, network medicine analysis, and docking screening. AVAILABILITY AND IMPLEMENTATION: The code and results are available at https://github.com/2020MEAI/DrugRepPT.

Understanding the mechanisms of candidate drugs play an important role in drug discovery. The activating/inhibiting mechanisms between drugs and targets are major types of mechanisms of drugs. Owing to the complexity of drug-target (DT) mechanisms and data scarcity, modelling this problem based on deep learning methods to accurately predict DT activating/inhibiting mechanisms remains a considerable challenge. Here, by considering network pharmacology, we propose a multi-view deep learning model, DrugAI, which combines four modules, i.e. a graph neural network for drugs, a convolutional neural network for targets, a network embedding module for drugs and targets and a deep neural network for predicting activating/inhibiting mechanisms between drugs and targets. Computational experiments show that DrugAI performs better than state-of-the-art methods and has good robustness and generalization. To demonstrate the reliability of the predictive results of DrugAI, bioassay experiments are conducted to validate two drugs (notopterol and alpha-asarone) predicted to activate TRPV1. Moreover, external validation bears out 61 pairs of mechanism relationships between natural products and their targets predicted by DrugAI based on independent literatures and PubChem bioassays. DrugAI, for the first time, provides a powerful multi-view deep learning framework for robust prediction of DT activating/inhibiting mechanisms.

Understanding the biological functions of molecules in specific human tissues or cell types is crucial for gaining insights into human physiology and disease. To address this issue, it is essential to systematically uncover associations among multilevel elements consisting of disease phenotypes, tissues, cell types and molecules, which could pose a challenge because of their heterogeneity and incompleteness. To address this challenge, we describe a new methodological framework, called Graph Local InfoMax (GLIM), based on a human multilevel network (HMLN) that we established by introducing multiple tissues and cell types on top of molecular networks. GLIM can systematically mine the potential relationships between multilevel elements by embedding the features of the HMLN through contrastive learning. Our simulation results demonstrated that GLIM consistently outperforms other state-of-the-art algorithms in disease gene prediction. Moreover, GLIM was also successfully used to infer cell markers and rewire intercellular and molecular interactions in the context of specific tissues or diseases. As a typical case, the tissue-cell-molecule network underlying gastritis and gastric cancer was first uncovered by GLIM, providing systematic insights into the mechanism underlying the occurrence and development of gastric cancer. Overall, our constructed methodological framework has the potential to systematically uncover complex disease mechanisms and mine high-quality relationships among phenotypical, tissue, cellular and molecular elements.

Inference of disease-gene associations helps unravel the pathogenesis of diseases and contributes to the treatment. Although many machine learning-based methods have been developed to predict causative genes, accurate association inference remains challenging. One major reason is the inaccurate feature selection and accumulation of error brought by commonly used multi-stage training architecture. In addition, the existing methods do not incorporate cell-type-specific information, thus fail to study gene functions at a higher resolution. Therefore, we introduce single-cell transcriptome data and construct a context-aware network to unbiasedly integrate all data sources. Then we develop a graph convolution-based approach named CIPHER-SC to realize a complete end-to-end learning architecture. Our approach outperforms four state-of-the-art approaches in five-fold cross-validations on three distinct test sets with the best AUC of 0.9501, demonstrating its stable ability either to predict the novel genes or to predict with genetic basis. The ablation study shows that our complete end-to-end design and unbiased data integration boost the performance from 0.8727 to 0.9443 in AUC. The addition of single-cell data further improves the prediction accuracy and makes our results be enriched for cell-type-specific genes. These results confirm the ability of CIPHER-SC to discover reliable disease genes. Our implementation is available at http://github.com/YidingZhang117/CIPHER-SC.

SUMMARY: Although many quantitative structure-activity relationship (QSAR) models are trained and evaluated for their predictive merits, understanding what models have been learning is of critical importance. However, the interpretation and visualization of QSAR model results remain challenging, especially for 'black box' models such as deep neural network (DNN). Here, we take a step forward to interpret the learned chemical features from DNN QSAR models, and present VISAR, an interactive tool for visualizing the structure-activity relationship. VISAR first provides functions to construct and train DNN models. Then VISAR builds the activity landscapes based on a series of compounds using the trained model, showing the correlation between the chemical feature space and the experimental activity space after model training, and allowing for knowledge mining from a global perspective. VISAR also maps the gradients of the chemical features to the corresponding compounds as contribution weights for each atom, and visualizes the positive and negative contributor substructures suggested by the models from a local perspective. Using the web application of VISAR, users could interactively explore the activity landscape and the color-coded atom contributions. We propose that VISAR could serve as a helpful tool for training and interactive analysis of the DNN QSAR model, providing insights for drug design, and an additional level of model validation. AVAILABILITY AND IMPLEMENTATION: The source code and usage instructions for VISAR are available on github https://github.com/qid12/visar. CONTACT: [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Gaining insight into chemogenomic drug-target interactions, such as those involving the substructures of synthetic drugs and protein domains, is important in fragment-based drug discovery and drug repositioning. Previous studies evaluated the interactions locally, thereby ignoring the competitive effects of different substructures or domains, but this could lead to high false-positive estimation, calling for a computational method that presents more predictive power. RESULTS: A statistical model, termed Global optimization-based InFerence of chemogenomic features from drug-Target interactions, or GIFT, is proposed herein to evaluate substructure-domain interactions globally such that all substructure-domain contributions to drug-target interaction are analyzed simultaneously. Combinations of different chemical substructures were included since they may function as one unit. When compared to previous methods, GIFT showed better interpretive performance, and performance for the recovery of drug-target interactions was good. Among 53 known drug-domain interactions, 81% were accurately predicted by GIFT. Eighteen of the top 100 predicted combined substructure-domain interactions had corresponding drug-target structures in the Protein Data Bank database, and 15 out of the 18 had been proved. GIFT was then implemented to predict substructure-domain interactions based on drug repositioning. For example, the anticancer activities of tazarotene, adapalene, acitretin and raloxifene were identified. In summary, GIFT is a global chemogenomic inference approach and offers fresh insight into drug-target interactions.

A typical characteristic of traditional Chinese medicine (TCM) is that TCM practitioners always prescribe different herbal formulae according to the condition of individual patients, making TCM as a kind of personalized medicine. In this study, we conducted a network analysis to evaluate the prescription rules underlying a set of anti-rheumatoid arthritis (RA) prescriptions collected from a TCM master in China, Mr. Ji-Ren Li. A herb network-compound network-disease network workflow was established for the TCM prescription analysis. Then the core herb network was constructed from 871 anti-RA prescriptions. We further demonstrated that the potential biological activities exerted by the herbs from the core network are significantly associated with the RA genes in terms of GO biological processes. The results give evidence for the effectiveness and reasonableness of herb formulae prescribed by Mr. Li.

In the state of the art hybrid video coding structure, intra prediction and transform coding are the most important two parts for intra picture coding. Generally, the prediction process and the transform process are separated. The current block is first predicted from the neighboring reconstructed pixels, the residue block is then transformed. In our approach, the prediction process and transform process are combined, therefore more information can be used for prediction and more accurate prediction can be achieved.

MOTIVATION: Tumor necrosis factor-alpha (TNF-α), a major inflammatory cytokine, is closely related to several cardiovascular pathological processes. However, its effects on the cell cycle of vascular endothelial cells (VECs) have been the subject of some controversy. To investigate the molecular mechanism underlying this process, we constructed time-course protein-protein interaction (PPI) networks of TNF-α induced regulation of cell cycle in VECs using microarray datasets and genome-wide PPI datasets. Then, we analyzed the topological properties of the responsive PPI networks and calculated the node degree and node betweenness centralization of each gene in the networks. We found that p21, p27 and cyclinD1, key genes of the G1/S checkpoint, are in the center of responsive PPI networks and their roles in PPI networks are significantly altered with induction of TNF-α. According to the following biological experiments, we proved that TNF-α can promote G(1)/S transition of cell cycle in VECs and facilitate the cell cycle activation induced by vascular endothelial growth factor. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Understanding how drugs and diseases are associated in the molecular level is of critical importance to unveil disease mechanisms and treatments. Until recently, few studies attempt end to discover important gene modules shared by both drugs and diseases. RESULTS: Here, we propose a novel presentation of drug-gene-disease relationship, a 'co-module', which is characterized by closely related drugs, diseases and genes. We first define a network-based gene closeness profile to relate drug to disease. Then, we develop a Bayesian partition method to identify drug-gene-disease co-modules underlying the gene closeness data. Genes share similar notable patterns with respect not only to the drugs but also the diseases within a co-module. Simulations show that our method, comCIPHER, achieves a better performance compared with a popular co-module detection method, PPA. We apply comCIPHER to a set consisting of 723 drugs, 275 diseases and 1442 genes and demonstrate that our co-module approach is able to identify new drug-disease associations and highlight their molecular basis. Disease co-morbidity emerges as well. Three co-modules are further illustrated in which new drug applications, including the anti-cancer metastasis activity of an anti-asthma drug Pranlukast, and a cardiovascular stress-testing agent Arbutamine for obesity, as well as potential side-effects, e.g. hypotension for Triamterene, are computationally identified. AVAILABILITY: The compiled version of comCIPHER can be found at http://bioinfo.au.tsinghua.edu.cn/comCIPHER/. The 86 co-modules can be downloaded from http://bioinfo.au.tsinghua.edu.cn/comCIPHER/Co_Module_Results.zip. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

BACKGROUND: Pathways in biological system often cooperate with each other to function. Changes of interactions among pathways tightly associate with alterations in the properties and functions of the cell and hence alterations in the phenotype. So, the pathway interactions and especially their changes over time corresponding to specific phenotype are critical to understanding cell functions and phenotypic plasticity. METHODS: With prior-defined pathways and incorporated protein-protein interaction (PPI) data, we counted PPIs between corresponding gene sets of each pair of distinct pathways to construct a comprehensive pathway network. Then we proposed a novel concept, characteristic sub pathway network (CSPN), to realize the phenotype-specific pathway interactions. By adding gene expression data regarding a given phenotype, angiogenesis, active PPIs corresponding to stimulation of interleukin-1 (IL-1) and tumor necrosis factor alpha (TNF-alpha) on human umbilical vein endothelial cells (HUVECs) respectively were derived. Two kinds of CSPN, namely the static or the dynamic CSPN, were detected by counting active PPIs. RESULTS: A comprehensive pathway network containing 37 signalling pathways as nodes and 263 pathway interactions were obtained. Two phenotype-specific CSPNs for angiogenesis, corresponding to stimulation of IL-1 and TNF-alpha on HUVEC respectively, were addressed. From phenotype-specific CSPNs, a static CSPN involving interactions among B cell receptor, T cell receptor, Toll-like receptor, MAPK, VEGF, and ErbB signalling pathways, and a dynamic CSPN involving interactions among TGF-beta, Wnt, p53 signalling pathways and cell cycle pathway, were detected for angiogenesis on HUVEC after stimulation of IL-1 and TNF-alpha respectively. We inferred that, in certain case, the static CSPN maintains related basic functions of the cells, whereas the dynamic CSPN manifests the cells' plastic responses to stimulus and therefore reflects the cells' phenotypic plasticity. CONCLUSION: The comprehensive pathway network helps us realize the cooperative behaviours among pathways. Moreover, two kinds of potential CSPNs found in this work, the static CSPN and the dynamic CSPN, are helpful to deeply understand the specific function of HUVEC and its phenotypic plasticity in regard to angiogenesis.

BACKGROUND: Traditional Chinese Medicine (TCM) is characterized by the wide use of herbal formulae, which are capable of systematically treating diseases determined by interactions among various herbs. However, the combination rule of TCM herbal formulae remains a mystery due to the lack of appropriate methods. METHODS: From a network perspective, we established a method called Distance-based Mutual Information Model (DMIM) to identify useful relationships among herbs in numerous herbal formulae. DMIM combines mutual information entropy and "between-herb-distance" to score herb interactions and construct herb network. To evaluate the efficacy of the DMIM-extracted herb network, we conducted in vitro assays to measure the activities of strongly connected herbs and herb pairs. Moreover, using the networked Liu-wei-di-huang (LWDH) formula as an example, we proposed a novel concept of "co-module" across herb-biomolecule-disease multilayer networks to explore the potential combination mechanism of herbal formulae. RESULTS: DMIM, when used for retrieving herb pairs, achieves a good balance among the herb's frequency, independence, and distance in herbal formulae. A herb network constructed by DMIM from 3865 Collaterals-related herbal formulae can not only nicely recover traditionally-defined herb pairs and formulae, but also generate novel anti-angiogenic herb ingredients (e.g. Vitexicarpin with IC50=3.2 μM, and Timosaponin A-III with IC50=3.4 μM) as well as herb pairs with synergistic or antagonistic effects. Based on gene and phenotype information associated with both LWDH herbs and LWDH-treated diseases, we found that LWDH-treated diseases show high phenotype similarity and identified certain "co-modules" enriched in cancer pathways and neuro-endocrine-immune pathways, which may be responsible for the action of treating different diseases by the same LWDH formula. CONCLUSIONS: DMIM is a powerful method to identify the combination rule of herbal formulae and lead to new discoveries. We also provide the first evidence that the co-module across multilayer networks may underlie the combination mechanism of herbal formulae and demonstrate the potential of network biology approaches in the studies of TCM.

The neuro-endocrine-immune (NEI) system plays a critical regulatory role in modulating host homeostasis and optimizing health. We created the dbNEI 2 years ago to collect NEI molecules and interactions. For transferring the conceptual NEI to the systematic NEI network and uncovering the NEI's medical function, we updated the dbNEI 2.0 in three ways: (i) extended NEI molecules to 2242 genes and 7657 chemical compounds by using gene ontology-based (GO-based) data mining strategy, (ii) added multilayer interactions of NEI molecules including KEGG signal transduction and metabolic pathways, HPRD protein-protein interactions (PPI), transcription factor and microRNA regulations and (iii) connected 611 drugs and 823 diseases through multilayer NEI interactions. The reconstructed drug-NEI-disease network will facilitate the systematic study of NEI system.

MOTIVATION: Network reconstruction of biological entities is very important for understanding biological processes and the organizational principles of biological systems. This work focuses on integrating both the literatures and microarray gene-expression data, and a combined literature mining and microarray analysis (LMMA) approach is developed to construct gene networks of a specific biological system. RESULTS: In the LMMA approach, a global network is first constructed using the literature-based co-occurrence method. It is then refined using microarray data through a multivariate selection procedure. An application of LMMA to the angiogenesis is presented. Our result shows that the LMMA-based network is more reliable than the co-occurrence-based network in dealing with multiple levels of KEGG gene, KEGG Orthology and pathway. AVAILABILITY: The LMMA program is available upon request.