Protein-ligand binding affinity (PLA) prediction plays a pivotal role in drug discovery. Although previous in-silico approaches have presented promising performance, they predominantly focus on a single perspective, often failing to simultaneously capture the intricate mutual effects among molecules and the distinct mechanisms of action inherent within these complexes. In this study, we propose DualBind, a novel dual-module framework that integrates the strengths of both interaction-free and interaction-based paradigms. DualBind simultaneously models individual characteristics of ligands, pockets, and proteins, as well as their mutual effects. Specifically, the interaction-free module employs a Structure-aware Transformer to capture both local and global relationships among molecules. Meanwhile, the interaction-based module leverages an adaptive graph neural network to model intra- and inter-molecular interactions. This dual-module architecture facilitates the extraction of both coarse sequence-level and fine-grained graph-level features, enhancing representation learning and PLA prediction accuracy. Comprehensive experiments on three datasets demonstrated that DualBind consistently outperformed state-of-the-art methods, achieving a 4.3% reduction in RMSE, while visualization tests on HDAC families further confirmed its biological relevance and reliability. The code is available at https://github.com/Arirhenium/DualBind.

Abstract Synthetic lethality (SL) is a type of genetic interaction that occurs when defects in two genes cause cell death, while a defect in a single gene does not. Targeting an SL partner of a gene mutated in cancer can selectively kill tumor cells. Traditional wet-lab experiments for SL screening are resource-intensive. Hence, many computational methods have been developed for virtual screening of SL gene pairs. This study benchmarks recent machine learning methods for SL prediction, including three matrix factorization and eight deep learning models. We scrutinize model performance using various data splitting scenarios, negative sample ratios, and negative sampling methods on both classification and ranking tasks to assess the models’ generalizability and robustness. Our benchmark analyzed performance differences among the models and emphasized the importance of data and real-world scenarios. Finally, we suggest future directions to improve machine learning methods for SL discovery in terms of predictive power and interpretability.

The advent of single-cell RNA sequencing (scRNA-seq) has offering unprecedented insights at the single-cell level. This groundbreaking technology has opened new pathways for understanding cellular diversity and revealing novel insights into disease mechanisms. However, the analysis of scRNA-seq data is challenging, primarily due to dropout events caused by technical noise. Developing effective imputation methods is crucial for the reliable and informative analysis of scRNA-seq data. While deep learning-based approaches have been proposed for scRNA-seq data imputation, they often fall short of optimal performance, especially in identifying rare cell types. Here we propose a novel self-supervised deep learning model named scGANCL for scRNA-seq data imputation. scGANCL combines bidirectional generative adversarial network (BiGAN) with contrastive learning (CL) to enhance imputation performance. To fully exploit gene expression profiles, a contrastive learning module is introduced to enhance the representation learning of cells by minimizing the discrepancy between the distributions of real and generated data. Comprehensive experiments have been conducted on ten simulated and seven real datasets to validate scGANCL's effectiveness. The results demonstrated scGANCL consistently outperformed seven state-of-the-art methods across various downstream tasks. Ablation studies further validated the contribution of each component to the overall performance of the model.

Recent advances in spatial transcriptomics have enabled subcellular-resolution profiling of gene expression, offering unprecedented opportunities to investigate intracellular architecture and local microenvironmental interactions. Graph neural networks (GNNs) have shown great promise in modeling spatial transcriptomics data. However, existing GNN-based methods primarily focus on low-frequency signals, overlooking high-frequency signals critical for resolving transcriptional differences across subcellular compartments and cell boundaries. This limits their ability to characterize fine-grained structural and functional heterogeneity within tissues, hindering accurate spatial domain identification. In this study, we propose HiFi-ST, a High-Frequency-Aware Graph Integration framework for subcellular spatial transcriptomics. HiFi-ST employs a high-pass filter to extract high-frequency transcriptional differences, which are then integrated with spatial contexts through a transformer-based architecture. A contrastive learning module is designed to enhance cell representation by aligning spatial organization with transcriptional heterogeneity. Comprehensive experiments on subcellular datasets demonstrated that HiFi-ST consistently outperformed six state-of-the-art methods in spatial clustering, gene expression enhancement, and niche identification.

Recent advancements in spatial transcriptomics (ST) technologies have greatly revolutionized our understanding of tissue heterogeneity and cellular functions. However, popular ST, such as 10x Visium, still fall short in achieving true single-cell resolution, underscoring an urgent need for in-silico methods that can accurately resolve cell type composition within ST data. While several methods have been proposed, most rely solely on gene expression profiles, often neglecting spatial context, which results in suboptimal performance. Additionally, many deconvolution methods dependent on scRNA-seq data fail to align the distribution of ST and scRNA-seq reference data, consequently affecting the accuracy of cell type mapping. In this study, we propose stGNN, a novel spatially-informed graph learning framework powered by statistical modeling for resolving fine-grained cell type compositions in ST. To capture comprehensive features, we develop a dual encoding module, utilizing both a graph convolutional network (GCN) and an auto-encoder to learn spatial and non-spatial representations respectively. Following that, we further design an adaptive attention mechanism to integrate these representations layer-by-layer, capturing multi-scale spatial structures from low to high order and thus improving representation learning. Additionally, for model training, we adopt a negative log-likelihood loss function that aligns the distribution of ST data with scRNA-seq (or snRNA-seq) reference data, enhancing the accuracy of cell type proportion prediction in ST. To assess the performance of stGNN, we applied our proposed model to six ST datasets from various platforms, including 10x Visium, Slide-seqV2, and Visium HD, for cell type proportion estimation. Our results demonstrate that stGNN consistently outperforms seven state-of-the-art methods. Notably, when applied to mouse brain tissues, stGNN successfully resolves clear cortical layers at a high resolution. Additionally, we show that stGNN is able to effectively resolve ST at different resolutions. In summary, stGNN provides a powerful framework for analyzing the spatial distribution of diverse cell populations in complex tissue structures. stGNN’s code is openly shared on https://github.com/LiangSDNULab/stGNN . Overview of stGNN. stGNN consists of three main modules: an auto-encoder module, a GCN-based encoder module, and an attentive integration module. First, the auto-encoder takes the preprocessed gene expressions as inputs to learn non-spatial representations, capturing the fundamental gene expression patterns. Next, the GCN-based encoder is to learn spatial representations by incorporating the representation learned by the auto-encoder and spatial context. To learn more informative representations, an attention mechanism is introduced to adaptively integrate the non-spatial and spatial representations layer by layer. The last layer of the GCN-based encoder is designed as a multiple classification layer (predictor) to predict cell type proportions. To align the distribution of ST data and scRNA-seq data and facilitate model training, a negative log-likelihood loss function is introduced, where the mean and dispersion parameter of ST data is estimated by the scvi-tools package from scRNA-seq reference data.

Spatial transcriptomics has revolutionized biological research but is constrained by high cost and limited resolution. We present SpaFoundation, an image-only histology foundational model pre-trained on 1.79 million patches to learn general-purpose representations for spatial transcriptomics. Leveraging histology image alone and evaluated on 117 samples, our method accurately infers spatial gene expression, enhances resolution to single-cell level, outperforming state-of-the-art lightweight models, and demonstrates strong transferability to tumor detection and spatial clustering. ### Competing Interest Statement The authors have declared no competing interest. National Natural Science Foundation of China under Grant, No.62225209

It is a vital step to identify the enzyme turnover number (kcat) in synthetic biology and early-stage drug discovery. Recently, deep learning methods have achieved inspiring process to predict kcat with the development of multi-species enzyme-substrate pairs turnover number data. However, the performance of existing approaches still heavily depends on the effectiveness of feature extraction for enzymes and substrates, as well as the optimal fusion of these two types of features. Furthermore, it is essential to identify the key molecular substructures that significantly impact kcat prediction. To address these issues, we develop a novel end-to-end dual-representation interpretability framework GELKcat by harnessing graph transformers for substrate molecular encoding and CNNs for enzyme word2vec embeddings. We further integrate substrate and enzyme features using the adaptive gate network, which assigns optimal weights to capture the most suitable feature combinations. The comparison with several state-of-the-art methods demonstrates the superiority of our GELKcat and the ablation studies further illuminate the invaluable roles of three main components. Furthermore, case studies illustrate the interpretability of GELKcat by identifying the key functional groups in a substrate, which are significantly associated with turnover number. It is anticipated that this work can bridge current gaps in enzyme-substrate representation, which can give some guidance for drug discovery and synthetic biology.

Advances in spatial omics technologies now allow multiple types of data to be acquired from the same tissue slice. To realize the full potential of such data, we need spatially informed methods for data integration. Here, we introduce SpatialGlue, a graph neural network model with a dual-attention mechanism that deciphers spatial domains by intra-omics integration of spatial location and omics measurement followed by cross-omics integration. We demonstrated SpatialGlue on data acquired from different tissue types using different technologies, including spatial epigenome-transcriptome and transcriptome-proteome modalities. Compared to other methods, SpatialGlue captured more anatomical details and more accurately resolved spatial domains such as the cortex layers of the brain. Our method also identified cell types like spleen macrophage subsets located at three different zones that were not available in the original data annotations. SpatialGlue scales well with data size and can be used to integrate three modalities. Our spatial multi-omics analysis tool combines the information from complementary omics modalities to obtain a holistic view of cellular and tissue properties.

MOTIVATION:Spatial transcriptomics (ST) technologies have revolutionized our ability to map gene expression patterns within native tissue context, providing unprecedented insights into tissue architecture and cellular heterogeneity. However, accurately deconvolving cell-type compositions from ST spots remains challenging due to the sparse and averaged nature of ST data, which is essential for accurately depicting tissue architecture. While numerous computational methods have been developed for cell-type deconvolution and spatial distribution reconstruction, most fail to capture tissue complexity at the single-cell level, thereby limiting their applicability in practical scenarios. RESULTS:To this end, we propose a novel cycle-consistent generative adversarial network named STCGAN for cellular deconvolution in spatial transcriptomic. STCGAN first employs a cycle-consistent generative adversarial network (CGAN) to pre-train on ST data, ensuring that both the mapping from ST data to latent space and its reverse mapping are consistent, capturing complex spatial gene expression patterns and learning robust latent representations. Based on the learned representation, STCGAN then optimizes a trainable cell-to-spot mapping matrix to integrate scRNA-seq data with ST data, accurately estimating cellular composition within each capture spot and effectively reconstructing the spatial distribution of cells across the tissue. To further enhance deconvolution accuracy, we incorporate spatial-aware regularization that ensures accurate cellular distribution reconstruction within the spatial context. Benchmarking against seven state-of-the-art methods on five simulated and real datasets from various tissues, STCGAN consistently delivers superior cell-type deconvolution performance. AVAILABILITY:The code of STCGAN can be downloaded from https://github.com/cs-wangbo/STCGAN and all the mentioned datasets are available on Zenodo at https://zenodo.org/doi/10.5281/zenodo.10799113.

Accurately predicting drug-target interactions (DTI) is a critical step in drug discovery. Existing methods of DTI prediction primarily employ Simplified Molecular-Input Line-Entry System (SMILES) sequences or molecular graphs to learn drug representations. However, the features learned by such single-view approach is prone to incomplete. While some multiview methods that consider the views of both SMILES sequences and molecular graphs have been developed, these methods often fall in short in capturing potential interactions between views. In this work, we propose a novel dual contrastive learning framework CSCL-DTI for DTI prediction. First, we design a contrastive-enhanced cross-view representation learning (CVRL) to learn representations for drugs. In this module, Transformer-based and graph convolutional network (GCN)-based encoders are separately adopted to learn view-specific representations, followed by contrastive learning to enrich the representations by accounting for the potential interplay between local chemical context and topological structure. Second, we combine Transformer with self-supervised contrastive learning (SSCL) to learn representations for targets by modelling protein amino acids sequences. The scheme allows to effectively preserve the intrinsic characteristics of the sequences. Finally, we introduce a bilinear attention network to obtain an integrated representation by adaptively incorporating drug and target representations. Benchmarking experiments on two datasets demonstrated that CSCL-DTI 1 outperforms six state-of-the-art methods.

Spatial transcriptomics enable us to dissect tissue heterogeneity and map out inter-cellular communications. Optimal integration of transcriptomics data and associated spatial information is essential towards fully exploiting the data. We present SEDR, an unsupervised spatially embedded deep representation of both transcript and spatial information. The SEDR pipeline uses a deep autoencoder to construct a low-dimensional latent representation of gene expression, which is then simultaneously embedded with the corresponding spatial information through a variational graph autoencoder. We applied SEDR on human dorsolateral prefrontal cortex data and achieved better clustering accuracy, and correctly retraced the prenatal cortex development order with trajectory analysis. We also found the SEDR representation to be eminently suited for batch integration. Applying SEDR to human breast cancer data, we discerned heterogeneous sub-regions within a visually homogenous tumor region, identifying a tumor core with pro-inflammatory microenvironment and an outer ring region enriched with tumor associated macrophages which drives an immune suppressive microenvironment.

ezSingleCell is an interactive and easy-to-use application for analysing various single-cell and spatial omics data types without requiring prior programing knowledge. It combines the best-performing publicly available methods for in-depth data analysis, integration, and interactive data visualization. ezSingleCell consists of five modules, each designed to be a comprehensive workflow for one data type or task. In addition, ezSingleCell allows crosstalk between different modules within a unified interface. Acceptable input data can be in a variety of formats while the output consists of publication ready figures and tables. In-depth manuals and video tutorials are available to guide users on the analysis workflows and parameter adjustments to suit their study aims. ezSingleCell's streamlined interface can analyse a standard scRNA-seq dataset of 3000 cells in less than five minutes. ezSingleCell is available in two forms: an installation-free web application (https://immunesinglecell.org/ezsc/) or a software package with a shinyApp interface (https://github.com/JinmiaoChenLab/ezSingleCell2) for offline analysis. ezSingleCell is an interactive and user-friendly application for the analysis of single-cell and spatial omics data, without the need for programming expertise. Here, authors integrated top-performing publicly available methods to enable comprehensive data analysis and interactive visualisation.

Abstract Integration of multiple data modalities in a spatially informed manner remains an unmet need for exploiting spatial multi-omics data. Here, we introduce SpatialGlue, a novel graph neural network with dual-attention mechanism, to decipher spatial domains by capturing the significance of each modality and neighbor graph in cross-omics and intra-omics integration. We demonstrate that SpatialGlue can accurately aggregate cell types into spatial domains at a higher resolution across different tissue types and technology platforms, as well as gain biological insights into cross-modality spatial correlations.

Integration of multiple data modalities in a spatially informed manner remains an unmet need for exploiting spatial multi-omics data. We introduce SpatialGlue, a graph neural network with dual-attention mechanism, to learn each modality’s significance at cross-omics and intra-omics integration. We demonstrate that SpatialGlue can accurately aggregate cell types into spatial domains at a higher resolution on different tissue types and technology platforms, as well as gain insights into cross-modality spatial correlations.

MOTIVATION:Metabolic stability plays a crucial role in the early stages of drug discovery and development. Accurately modeling and predicting molecular metabolic stability has great potential for the efficient screening of drug candidates as well as the optimization of lead compounds. Considering wet-lab experiment is time-consuming, laborious, and expensive, in silico prediction of metabolic stability is an alternative choice. However, few computational methods have been developed to address this task. In addition, it remains a significant challenge to explain key functional groups determining metabolic stability. RESULTS:To address these issues, we develop a novel cross-modality graph contrastive learning model named CMMS-GCL for predicting the metabolic stability of drug candidates. In our framework, we design deep learning methods to extract features for molecules from two modality data, i.e. SMILES sequence and molecule graph. In particular, for the sequence data, we design a multihead attention BiGRU-based encoder to preserve the context of symbols to learn sequence representations of molecules. For the graph data, we propose a graph contrastive learning-based encoder to learn structure representations by effectively capturing the consistencies between local and global structures. We further exploit fully connected neural networks to combine the sequence and structure representations for model training. Extensive experimental results on two datasets demonstrate that our CMMS-GCL consistently outperforms seven state-of-the-art methods. Furthermore, a collection of case studies on sequence data and statistical analyses of the graph structure module strengthens the validation of the interpretability of crucial functional groups recognized by CMMS-GCL. Overall, CMMS-GCL can serve as an effective and interpretable tool for predicting metabolic stability, identifying critical functional groups, and thus facilitating the drug discovery process and lead compound optimization. AVAILABILITY AND IMPLEMENTATION:The code and data underlying this article are freely available at https://github.com/dubingxue/CMMS-GCL.

Drug-drug interaction (DDI) plays an increasingly crucial role in drug discovery. Predicting potential DDI is also essential for clinical research. Given the high cost and risk of wet-lab experiments, in-silico DDI prediction is an alternative choice. Recently, deep learning methods have been developed for DDI prediction. However, most of existing methods focus on feature extraction from either molecular SMILES sequences or drug interactive networks, ignoring the valuable complementary information that can be derived from these two views. In this paper, we propose a novel interpretable Multi-View Attention network (MVA-DDI) for DDI prediction. MVA-DDI can effectively extracts drug representations from different perspectives to improve DDI prediction. Specifically, for a given drug, we design a transformer-based encoder and a graph convolutional networkbased encoder to learn sequence and graph representations from SMILES sequence and molecular graph, respectively. To fully exploit the complementary information between the sequence and molecular views, an attention mechanism is further adopted to adaptively aggregate the sequence and graph representations by taking the importance of different views into accounts, generating the final drug representations. Comparison experiments demonstrated that our MVA-DDI 1 model achieved superior performance to state-of-the-art models on DDI prediction.

Computational modeling and identification of the enzyme turnover number k cat are crucial for synthetic biology and early-stage lead optimization. Therefore, the accurate assessment of the k cat for enzyme-substrate pairs is essential. Considering wet-lab experiment is time-consuming, laborious, and expensive, in silico prediction of k cat is an alternative choice. However, few computational methods have been developed to address this task and other enzyme kinetics predictions. To address this, we develop a novel end-to-end dual-representation framework GELKcat by harnessing graph transformers for substrate molecular encoding and CNNs for enzyme word2vec embeddings. We further integrate substrate and enzyme features using the adaptive gate network, which assigns optimal weights to capture the most suitable feature combinations. The comparison with several state-of-the-art methods exhibits the superiority of our GELKcat. The Ablation studies further illuminate the invaluable roles of the word2vec embeddings of enzymes. It is anticipated that this work can bridge current gaps in enzyme-substrate representation, which can give some guidance for drug discovery and synthetic biology.

Motivation Recent advances in spatial transcriptomics technologies have enabled gene expression profiles while preserving spatial context. Accurately identifying spatial domains is crucial for downstream analysis and it requires the effective integration of gene expression profiles and spatial information. While increasingly computational methods have been developed for spatial domain detection, most of them cannot adaptively learn the complex relationship between gene expression and spatial information, leading to sub-optimal performance. Results To overcome these challenges, we propose a novel deep learning method named Spatial-MGCN for identifying spatial domains, which is a Multi-view Graph Convolutional Network (GCN) with attention mechanism. We first construct two neighbor graphs using gene expression profiles and spatial information, respectively. Then, a multi-view GCN encoder is designed to extract unique embeddings from both the feature and spatial graphs, as well as their shared embeddings by combining both graphs. Finally, a zero-inflated negative binomial decoder is used to reconstruct the original expression matrix by capturing the global probability distribution of gene expression profiles. Moreover, Spatial-MGCN incorporates a spatial regularization constraint into the features learning to preserve spatial neighbor information in an end-to-end manner. The experimental results show that Spatial-MGCN outperforms state-of-the-art methods consistently in several tasks, including spatial clustering and trajectory inference.