Graph-pMHC: graph neural network approach to MHC class II peptide presentation and antibody immunogenicity.

Antigen presentation on MHC class II (pMHCII presentation) plays an essential role in the adaptive immune response to extracellular pathogens and cancerous cells. But it can also reduce the efficacy of large-molecule drugs by triggering an anti-drug response. Significant progress has been made in pMHCII presentation modeling due to the collection of large-scale pMHC mass spectrometry datasets (ligandomes) and advances in machine learning. Here, we develop graph-pMHC, a graph neural network approach to predict pMHCII presentation. We derive adjacency matrices for pMHCII using Alphafold2-multimer and address the peptide-MHC binding groove alignment problem with a simple graph enumeration strategy. We demonstrate that graph-pMHC dramatically outperforms methods with suboptimal inductive biases, such as the multilayer-perceptron-based NetMHCIIpan-4.0 (+20.17% absolute average precision). Finally, we create an antibody drug immunogenicity dataset from clinical trial data and develop a method for measuring anti-antibody immunogenicity risk using pMHCII presentation models. Our model increases receiver operating characteristic curve (ROC)-area under the ROC curve (AUC) by 2.57% compared to just filtering peptides by hits in OASis alone for predicting antibody drug immunogenicity.

MS-BACL: enhancing metabolic stability prediction through bond graph augmentation and contrastive learning.

MOTIVATION: Accurately predicting molecular metabolic stability is of great significance to drug research and development, ensuring drug safety and effectiveness. Existing deep learning methods, especially graph neural networks, can reveal the molecular structure of drugs and thus efficiently predict the metabolic stability of molecules. However, most of these methods focus on the message passing between adjacent atoms in the molecular graph, ignoring the relationship between bonds. This makes it difficult for these methods to estimate accurate molecular representations, thereby being limited in molecular metabolic stability prediction tasks. RESULTS: We propose the MS-BACL model based on bond graph augmentation technology and contrastive learning strategy, which can efficiently and reliably predict the metabolic stability of molecules. To our knowledge, this is the first time that bond-to-bond relationships in molecular graph structures have been considered in the task of metabolic stability prediction. We build a bond graph based on 'atom-bond-atom', and the model can simultaneously capture the information of atoms and bonds during the message propagation process. This enhances the model's ability to reveal the internal structure of the molecule, thereby improving the structural representation of the molecule. Furthermore, we perform contrastive learning training based on the molecular graph and its bond graph to learn the final molecular representation. Multiple sets of experimental results on public datasets show that the proposed MS-BACL model outperforms the state-of-the-art model. AVAILABILITY AND IMPLEMENTATION: The code and data are publicly available at https://github.com/taowang11/MS.

GTE: a graph learning framework for prediction of T-cell receptors and epitopes binding specificity.

The interaction between T-cell receptors (TCRs) and peptides (epitopes) presented by major histocompatibility complex molecules (MHC) is fundamental to the immune response. Accurate prediction of TCR-epitope interactions is crucial for advancing the understanding of various diseases and their prevention and treatment. Existing methods primarily rely on sequence-based approaches, overlooking the inherent topology structure of TCR-epitope interaction networks. In this study, we present $GTE$, a novel heterogeneous Graph neural network model based on inductive learning to capture the topological structure between TCRs and Epitopes. Furthermore, we address the challenge of constructing negative samples within the graph by proposing a dynamic edge update strategy, enhancing model learning with the nonbinding TCR-epitope pairs. Additionally, to overcome data imbalance, we adapt the Deep AUC Maximization strategy to the graph domain. Extensive experiments are conducted on four public datasets to demonstrate the superiority of exploring underlying topological structures in predicting TCR-epitope interactions, illustrating the benefits of delving into complex molecular networks. The implementation code and data are available at https://github.com/uta-smile/GTE.

GMFGRN: a matrix factorization and graph neural network approach for gene regulatory network inference.

The recent advances of single-cell RNA sequencing (scRNA-seq) have enabled reliable profiling of gene expression at the single-cell level, providing opportunities for accurate inference of gene regulatory networks (GRNs) on scRNA-seq data. Most methods for inferring GRNs suffer from the inability to eliminate transitive interactions or necessitate expensive computational resources. To address these, we present a novel method, termed GMFGRN, for accurate graph neural network (GNN)-based GRN inference from scRNA-seq data. GMFGRN employs GNN for matrix factorization and learns representative embeddings for genes. For transcription factor-gene pairs, it utilizes the learned embeddings to determine whether they interact with each other. The extensive suite of benchmarking experiments encompassing eight static scRNA-seq datasets alongside several state-of-the-art methods demonstrated mean improvements of 1.9 and 2.5% over the runner-up in area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC). In addition, across four time-series datasets, maximum enhancements of 2.4 and 1.3% in AUROC and AUPRC were observed in comparison to the runner-up. Moreover, GMFGRN requires significantly less training time and memory consumption, with time and memory consumed <10% compared to the second-best method. These findings underscore the substantial potential of GMFGRN in the inference of GRNs. It is publicly available at https://github.com/Lishuoyy/GMFGRN.

Biological knowledge graph-guided investigation of immune therapy response in cancer with graph neural network.

The determination of transcriptome profiles that mediate immune therapy in cancer remains a major clinical and biological challenge. Despite responses induced by immune-check points inhibitors (ICIs) in diverse tumor types and all the big breakthroughs in cancer immunotherapy, most patients with solid tumors do not respond to ICI therapies. It still remains a big challenge to predict the ICI treatment response. Here, we propose a framework with multiple prior knowledge networks guided for immune checkpoints inhibitors prediction-DeepOmix-ICI (or ICInet for short). ICInet can predict the immune therapy response by leveraging geometric deep learning and prior biological knowledge graphs of gene-gene interactions. Here, we demonstrate more than 600 ICI-treated patients with ICI response data and gene expression profile to apply on ICInet. ICInet was used for ICI therapy responses prediciton across different cancer types-melanoma, gastric cancer and bladder cancer, which includes 7 cohorts from different data sources. ICInet is able to robustly generalize into multiple cancer types. Moreover, the performance of ICInet in those cancer types can outperform other ICI biomarkers in the clinic. Our model [area under the curve (AUC = 0.85)] generally outperformed other measures, including tumor mutational burden (AUC = 0.62) and programmed cell death ligand-1 score (AUC = 0.74). Therefore, our study presents a prior-knowledge guided deep learning method to effectively select immunotherapy-response-associated biomarkers, thereby improving the prediction of immunotherapy response for precision oncology.

Identification of vital chemical information via visualization of graph neural networks.

Qualitative or quantitative prediction models of structure-activity relationships based on graph neural networks (GNNs) are prevalent in drug discovery applications and commonly have excellently predictive power. However, the network information flows of GNNs are highly complex and accompanied by poor interpretability. Unfortunately, there are relatively less studies on GNN attributions, and their developments in drug research are still at the early stages. In this work, we adopted several advanced attribution techniques for different GNN frameworks and applied them to explain multiple drug molecule property prediction tasks, enabling the identification and visualization of vital chemical information in the networks. Additionally, we evaluated them quantitatively with attribution metrics such as accuracy, sparsity, fidelity and infidelity, stability and sensitivity; discussed their applicability and limitations; and provided an open-source benchmark platform for researchers. The results showed that all attribution techniques were effective, while those directly related to the predicted labels, such as integrated gradient, preferred to have better attribution performance. These attribution techniques we have implemented could be directly used for the vast majority of chemical GNN interpretation tasks.

Prediction of multi-relational drug-gene interaction via Dynamic hyperGraph Contrastive Learning.

Drug-gene interaction prediction occupies a crucial position in various areas of drug discovery, such as drug repurposing, lead discovery and off-target detection. Previous studies show good performance, but they are limited to exploring the binding interactions and ignoring the other interaction relationships. Graph neural networks have emerged as promising approaches owing to their powerful capability of modeling correlations under drug-gene bipartite graphs. Despite the widespread adoption of graph neural network-based methods, many of them experience performance degradation in situations where high-quality and sufficient training data are unavailable. Unfortunately, in practical drug discovery scenarios, interaction data are often sparse and noisy, which may lead to unsatisfactory results. To undertake the above challenges, we propose a novel Dynamic hyperGraph Contrastive Learning (DGCL) framework that exploits local and global relationships between drugs and genes. Specifically, graph convolutions are adopted to extract explicit local relations among drugs and genes. Meanwhile, the cooperation of dynamic hypergraph structure learning and hypergraph message passing enables the model to aggregate information in a global region. With flexible global-level messages, a self-augmented contrastive learning component is designed to constrain hypergraph structure learning and enhance the discrimination of drug/gene representations. Experiments conducted on three datasets show that DGCL is superior to eight state-of-the-art methods and notably gains a 7.6% performance improvement on the DGIdb dataset. Further analyses verify the robustness of DGCL for alleviating data sparsity and over-smoothing issues.

HimGNN: a novel hierarchical molecular graph representation learning framework for property prediction.

Accurate prediction of molecular properties is an important topic in drug discovery. Recent works have developed various representation schemes for molecular structures to capture different chemical information in molecules. The atom and motif can be viewed as hierarchical molecular structures that are widely used for learning molecular representations to predict chemical properties. Previous works have attempted to exploit both atom and motif to address the problem of information loss in single representation learning for various tasks. To further fuse such hierarchical information, the correspondence between learned chemical features from different molecular structures should be considered. Herein, we propose a novel framework for molecular property prediction, called hierarchical molecular graph neural networks (HimGNN). HimGNN learns hierarchical topology representations by applying graph neural networks on atom- and motif-based graphs. In order to boost the representational power of the motif feature, we design a Transformer-based local augmentation module to enrich motif features by introducing heterogeneous atom information in motif representation learning. Besides, we focus on the molecular hierarchical relationship and propose a simple yet effective rescaling module, called contextual self-rescaling, that adaptively recalibrates molecular representations by explicitly modelling interdependencies between atom and motif features. Extensive computational experiments demonstrate that HimGNN can achieve promising performances over state-of-the-art baselines on both classification and regression tasks in molecular property prediction.

Predicting molecular properties based on the interpretable graph neural network with multistep focus mechanism.

Graph neural networks based on deep learning methods have been extensively applied to the molecular property prediction because of its powerful feature learning ability and good performance. However, most of them are black boxes and cannot give the reasonable explanation about the underlying prediction mechanisms, which seriously reduce people's trust on the neural network-based prediction models. Here we proposed a novel graph neural network named iteratively focused graph network (IFGN), which can gradually identify the key atoms/groups in the molecule that are closely related to the predicted properties by the multistep focus mechanism. At the same time, the combination of the multistep focus mechanism with visualization can also generate multistep interpretations, thus allowing us to gain a deep understanding of the predictive behaviors of the model. For all studied eight datasets, the IFGN model achieved good prediction performance, indicating that the proposed multistep focus mechanism also can improve the performance of the model obviously besides increasing the interpretability of built model. For researchers to use conveniently, the corresponding website (http://graphadmet.cn/works/IFGN) was also developed and can be used free of charge.

Chemical structure-aware molecular image representation learning.

Current methods of molecular image-based drug discovery face two major challenges: (1) work effectively in absence of labels, and (2) capture chemical structure from implicitly encoded images. Given that chemical structures are explicitly encoded by molecular graphs (such as nitrogen, benzene rings and double bonds), we leverage self-supervised contrastive learning to transfer chemical knowledge from graphs to images. Specifically, we propose a novel Contrastive Graph-Image Pre-training (CGIP) framework for molecular representation learning, which learns explicit information in graphs and implicit information in images from large-scale unlabeled molecules via carefully designed intra- and inter-modal contrastive learning. We evaluate the performance of CGIP on multiple experimental settings (molecular property prediction, cross-modal retrieval and distribution similarity), and the results show that CGIP can achieve state-of-the-art performance on all 12 benchmark datasets and demonstrate that CGIP transfers chemical knowledge in graphs to molecular images, enabling image encoder to perceive chemical structures in images. We hope this simple and effective framework will inspire people to think about the value of image for molecular representation learning.

Cellograph: a semi-supervised approach to analyzing multi-condition single-cell RNA-sequencing data using graph neural networks.

With the growing number of single-cell datasets collected under more complex experimental conditions, there is an opportunity to leverage single-cell variability to reveal deeper insights into how cells respond to perturbations. Many existing approaches rely on discretizing the data into clusters for differential gene expression (DGE), effectively ironing out any information unveiled by the single-cell variability across cell-types. In addition, DGE often assumes a statistical distribution that, if erroneous, can lead to false positive differentially expressed genes. Here, we present Cellograph: a semi-supervised framework that uses graph neural networks to quantify the effects of perturbations at single-cell granularity. Cellograph not only measures how prototypical cells are of each condition but also learns a latent space that is amenable to interpretable data visualization and clustering. The learned gene weight matrix from training reveals pertinent genes driving the differences between conditions. We demonstrate the utility of our approach on publicly-available datasets including cancer drug therapy, stem cell reprogramming, and organoid differentiation. Cellograph outperforms existing methods for quantifying the effects of experimental perturbations and offers a novel framework to analyze single-cell data using deep learning.

A deep learning framework for predicting disease-gene associations with functional modules and graph augmentation.

BACKGROUND: The exploration of gene-disease associations is crucial for understanding the mechanisms underlying disease onset and progression, with significant implications for prevention and treatment strategies. Advances in high-throughput biotechnology have generated a wealth of data linking diseases to specific genes. While graph representation learning has recently introduced groundbreaking approaches for predicting novel associations, existing studies always overlooked the cumulative impact of functional modules such as protein complexes and the incompletion of some important data such as protein interactions, which limits the detection performance. RESULTS: Addressing these limitations, here we introduce a deep learning framework called ModulePred for predicting disease-gene associations. ModulePred performs graph augmentation on the protein interaction network using L3 link prediction algorithms. It builds a heterogeneous module network by integrating disease-gene associations, protein complexes and augmented protein interactions, and develops a novel graph embedding for the heterogeneous module network. Subsequently, a graph neural network is constructed to learn node representations by collectively aggregating information from topological structure, and gene prioritization is carried out by the disease and gene embeddings obtained from the graph neural network. Experimental results underscore the superiority of ModulePred, showcasing the effectiveness of incorporating functional modules and graph augmentation in predicting disease-gene associations. This research introduces innovative ideas and directions, enhancing the understanding and prediction of gene-disease relationships.

Fusing graph transformer with multi-aggregate GCN for enhanced drug-disease associations prediction.

BACKGROUND: Identification of potential drug-disease associations is important for both the discovery of new indications for drugs and for the reduction of unknown adverse drug reactions. Exploring the potential links between drugs and diseases is crucial for advancing biomedical research and improving healthcare. While advanced computational techniques play a vital role in revealing the connections between drugs and diseases, current research still faces challenges in the process of mining potential relationships between drugs and diseases using heterogeneous network data. RESULTS: In this study, we propose a learning framework for fusing Graph Transformer Networks and multi-aggregate graph convolutional network to learn efficient heterogenous information graph representations for drug-disease association prediction, termed WMAGT. This method extensively harnesses the capabilities of a robust graph transformer, effectively modeling the local and global interactions of nodes by integrating a graph convolutional network and a graph transformer with self-attention mechanisms in its encoder. We first integrate drug-drug, drug-disease, and disease-disease networks to construct heterogeneous information graph. Multi-aggregate graph convolutional network and graph transformer are then used in conjunction with neural collaborative filtering module to integrate information from different domains into highly effective feature representation. CONCLUSIONS: Rigorous cross-validation, ablation studies examined the robustness and effectiveness of the proposed method. Experimental results demonstrate that WMAGT outperforms other state-of-the-art methods in accurate drug-disease association prediction, which is beneficial for drug repositioning and drug safety research.

Predicting the functional state of protein kinases using interpretable graph neural networks from sequence and structural data.

Protein kinases are central to cellular activities and are actively pursued as drug targets for several conditions including cancer and autoimmune diseases. Despite the availability of a large structural database for kinases, methodologies to elucidate the structure-function relationship of these proteins (without manual intervention) are lacking. Such techniques are essential in structural biology and to accelerate drug discovery efforts. Here, we implement an interpretable graph neural network (GNN) framework for classifying the functionally active and inactive states of a large set of protein kinases by only using their tertiary structure and amino acid sequence. We show that the GNN models can classify kinase structures with high accuracy (>97%). We implement the Gradient-weighted Class Activation Mapping for graphs (Graph Grad-CAM) to automatically identify structurally important residues and residue-residue contacts of the kinases without any a priori input. We show that the motifs identified through the Graph Grad-CAM methodology are functionally critical, consistent with the existing kinase literature. Notably, the highly conserved DFG and HRD motifs of the well-known hydrophobic spine are identified by the interpretable framework in addition to some of the lesser known motifs. Further, using Grad-CAM maps as the vector embedding of the protein structures, we identify the subtle differences in the crystal structures among different sub-classes of kinases in the Protein Data Bank (PDB). Frameworks such as the one implemented here, for high-throughput identification of protein structure-function relationships are essential in designing targeted small molecules therapies as well as in engineering new proteins for novel applications.

De novo prediction of RNA-protein interactions with graph neural networks.

RNA-binding proteins (RBPs) are key co- and post-transcriptional regulators of gene expression, playing a crucial role in many biological processes. Experimental methods like CLIP-seq have enabled the identification of transcriptome-wide RNA-protein interactions for select proteins; however, the time- and resource-intensive nature of these technologies call for the development of computational methods to complement their predictions. Here, we leverage recent, large-scale CLIP-seq experiments to construct a de novo predictor of RNA-protein interactions based on graph neural networks (GNN). We show that the GNN method allows us not only to predict missing links in an RNA-protein network, but to predict the entire complement of targets of previously unassayed proteins, and even to reconstruct the entire network of RNA-protein interactions in different conditions based on minimal information. Our results demonstrate the potential of modern machine learning methods to extract useful information on post-transcriptional regulation from large data sets.

A heterogeneous network-based method with attentive meta-path extraction for predicting drug-target interactions.

Predicting drug-target interactions (DTIs) is crucial at many phases of drug discovery and repositioning. Many computational methods based on heterogeneous networks (HNs) have proved their potential to predict DTIs by capturing extensive biological knowledge and semantic information from meta-paths. However, existing methods manually customize meta-paths, which is overly dependent on some specific expertise. Such strategy heavily limits the scalability and flexibility of these models, and even affects their predictive performance. To alleviate this limitation, we propose a novel HN-based method with attentive meta-path extraction for DTI prediction, named HampDTI, which is capable of automatically extracting useful meta-paths through a learnable attention mechanism instead of pre-definition based on domain knowledge. Specifically, by scoring multi-hop connections across various relations in the HN with each relation assigned an attention weight, HampDTI constructs a new trainable graph structure, called meta-path graph. Such meta-path graph implicitly measures the importance of every possible meta-path between drugs and targets. To enable HampDTI to extract more diverse meta-paths, we adopt a multi-channel mechanism to generate multiple meta-path graphs. Then, a graph neural network is deployed on the generated meta-path graphs to yield the multi-channel embeddings of drugs and targets. Finally, HampDTI fuses all embeddings from different channels for predicting DTIs. The meta-path graphs are optimized along with the model training such that HampDTI can adaptively extract valuable meta-paths for DTI prediction. The experiments on benchmark datasets not only show the superiority of HampDTI in DTI prediction over several baseline methods, but also, more importantly, demonstrate the effectiveness of the model discovering important meta-paths.

Spatial transcriptomics prediction from histology jointly through Transformer and graph neural networks.

The rapid development of spatial transcriptomics allows the measurement of RNA abundance at a high spatial resolution, making it possible to simultaneously profile gene expression, spatial locations of cells or spots, and the corresponding hematoxylin and eosin-stained histology images. It turns promising to predict gene expression from histology images that are relatively easy and cheap to obtain. For this purpose, several methods are devised, but they have not fully captured the internal relations of the 2D vision features or spatial dependency between spots. Here, we developed Hist2ST, a deep learning-based model to predict RNA-seq expression from histology images. Around each sequenced spot, the corresponding histology image is cropped into an image patch and fed into a convolutional module to extract 2D vision features. Meanwhile, the spatial relations with the whole image and neighbored patches are captured through Transformer and graph neural network modules, respectively. These learned features are then used to predict the gene expression by following the zero-inflated negative binomial distribution. To alleviate the impact by the small spatial transcriptomics data, a self-distillation mechanism is employed for efficient learning of the model. By comprehensive tests on cancer and normal datasets, Hist2ST was shown to outperform existing methods in terms of both gene expression prediction and spatial region identification. Further pathway analyses indicated that our model could reserve biological information. Thus, Hist2ST enables generating spatial transcriptomics data from histology images for elucidating molecular signatures of tissues.

How much can deep learning improve prediction of the responses to drugs in cancer cell lines?

The drug response prediction problem arises from personalized medicine and drug discovery. Deep neural networks have been applied to the multi-omics data being available for over 1000 cancer cell lines and tissues for better drug response prediction. We summarize and examine state-of-the-art deep learning methods that have been published recently. Although significant progresses have been made in deep learning approach in drug response prediction, deep learning methods show their weakness for predicting the response of a drug that does not appear in the training dataset. In particular, all the five evaluated deep learning methods performed worst than the similarity-regularized matrix factorization (SRMF) method in our drug blind test. We outline the challenges in applying deep learning approach to drug response prediction and suggest unique opportunities for deep learning integrated with established bioinformatics analyses to overcome some of these challenges.

Multi-sample dual-decoder graph autoencoder.

Self-supervised learning has shown superior performance on graph-related tasks in recent years. The most advanced methods are based on contrast learning, which severely limited by structured data augmentation techniques and complex training methods. Generative self-supervised learning, especially graph autoencoders (GAEs), can prevent the above dependence and has been demonstrated as an effective approach. In addition, most previous works only reconstruct the graph topological structure or node features. Few works consider both and combine them together to obtain their complementary information. To overcome these problems, we propose a generative self-supervised graph representation learning methodology named Multi-View Dual-decoder Graph Autoencoder (MDGA). Specifically, we first design a multi-sample graph learning strategy which benefits the generalization of the dual-decoder graph autoencoder. Moreover, the proposed model reconstructs the graph topological structure with a traditional GAE and extracts node attributes by masked feature reconstruction. Experimental results on five public benchmark datasets demonstrate that MDGA outperforms state-of-the-art methods in both node classification and link prediction tasks.

Single-cell RNA-seq data analysis based on directed graph neural network.

Single-cell RNA sequencing (scRNA-seq) data scale surges with high-throughput sequencing technology development. However, although single-cell data analysis is a powerful tool, various issues have been reported, such as sequencing sparsity and complex differential patterns in gene expression. Statistical or traditional machine learning methods are inefficient, and the accuracy needs to be improved. The methods based on deep learning can not directly process non-Euclidean spatial data, such as cell diagrams. In this study, we have developed graph autoencoders and graph attention network for scRNA-seq analysis based on a directed graph neural network named scDGAE. Directed graph neural networks cannot only retain the connection properties of the directed graph but also expand the receptive field of the convolution operation. Cosine similarity, median L1 distance, and root-mean-squared error are used to measure the gene imputation performance of different methods with scDGAE. Furthermore, adjusted mutual information, normalized mutual information, completeness score, and Silhouette coefficient score are used to measure the cell clustering performance of different methods with scDGAE. Experiment results show that the scDGAE model achieves promising performance in gene imputation and cell clustering prediction on four scRNA-seq data sets with gold-standard cell labels. Furthermore, it is a robust framework that can be applied to general scRNA-Seq analyses.