Deep learning models can accurately predict molecular properties and help making the search for potential drug candidates faster and more efficient. Many existing methods are purely data driven, focusing on exploiting the intrinsic topology and construction rules of molecules without any chemical prior information. The high data dependency makes them difficult to generalize to a wider chemical space and leads to a lack of interpretability of predictions. Here, to address this issue, we introduce a chemical element-oriented knowledge graph to summarize the basic knowledge of elements and their closely related functional groups. We further propose a method for knowledge graph-enhanced molecular contrastive learning with functional prompt (KANO), exploiting external fundamental domain knowledge in both pre-training and fine-tuning. Specifically, with element-oriented knowledge graph as a prior, we first design an element-guided graph augmentation in contrastive-based pre-training to explore microscopic atomic associations without violating molecular semantics. Then, we learn functional prompts in fine-tuning to evoke the downstream task-related knowledge acquired by the pre-trained model. Extensive experiments show that KANO outperforms state-of-the-art baselines on 14 molecular property prediction datasets and provides chemically sound explanations for its predictions. This work contributes to more efficient drug design by offering a high-quality knowledge prior, interpretable molecular representation and superior prediction performance.

Spatially resolved transcriptomics provides genetic information in space toward elucidation of the spatial architecture in intact organs and the spatially resolved cell-cell communications mediating tissue homeostasis, development, and disease. To facilitate inference of spatially resolved cell-cell communications, we here present SpaTalk, which relies on a graph network and knowledge graph to model and score the ligand-receptor-target signaling network between spatially proximal cells by dissecting cell-type composition through a non-negative linear model and spatial mapping between single-cell transcriptomic and spatially resolved transcriptomic data. The benchmarked performance of SpaTalk on public single-cell spatial transcriptomic datasets is superior to that of existing inference methods. Then we apply SpaTalk to STARmap, Slide-seq, and 10X Visium data, revealing the in-depth communicative mechanisms underlying normal and disease tissues with spatial structure. SpaTalk can uncover spatially resolved cell-cell communications for single-cell and spot-based spatially resolved transcriptomic data universally, providing valuable insights into spatial inter-cellular tissue dynamics.

Advances in single-cell RNA sequencing (scRNA-seq) have furthered the simultaneous classification of thousands of cells in a single assay based on transcriptome profiling. In most analysis protocols, single-cell type annotation relies on marker genes or RNA-seq profiles, resulting in poor extrapolation. Still, the accurate cell-type annotation for single-cell transcriptomic data remains a great challenge. Here, we introduce scDeepSort (https://github.com/ZJUFanLab/scDeepSort), a pre-trained cell-type annotation tool for single-cell transcriptomics that uses a deep learning model with a weighted graph neural network (GNN). Using human and mouse scRNA-seq data resources, we demonstrate the high performance and robustness of scDeepSort in labeling 764 741 cells involving 56 human and 32 mouse tissues. Significantly, scDeepSort outperformed other known methods in annotating 76 external test datasets, reaching an 83.79% accuracy across 265 489 cells in humans and mice. Moreover, we demonstrate the universality of scDeepSort using more challenging datasets and using references from different scRNA-seq technology. Above all, scDeepSort is the first attempt to annotate cell types of scRNA-seq data with a pre-trained GNN model, which can realize the accurate cell-type annotation without additional references, i.e. markers or RNA-seq profiles.

Uncovering the tissue molecular architecture at single-cell resolution could help better understand organisms’ biological and pathological processes. However, bulk RNA-seq can only measure gene expression in cell mixtures, without revealing the transcriptional heterogeneity and spatial patterns of single cells. Herein, we introduce Bulk2Space (https://github.com/ZJUFanLab/bulk2space), a deep learning framework-based spatial deconvolution algorithm that can simultaneously disclose the spatial and cellular heterogeneity of bulk RNA-seq data using existing single-cell and spatial transcriptomics references. The use of bulk transcriptomics to validate Bulk2Space unveils, in particular, the spatial variance of immune cells in different tumor regions, the molecular and spatial heterogeneity of tissues during inflammation-induced tumorigenesis, and spatial patterns of novel genes in different cell types. Moreover, Bulk2Space is utilized to perform spatial deconvolution analysis on bulk transcriptome data from two different mouse brain regions derived from our in-house developed sequencing approach termed Spatial-seq. We have not only reconstructed the hierarchical structure of the mouse isocortex but also further annotated cell types that were not identified by original methods in the mouse hypothalamus.

The utilization of ground-breaking large language models (LLMs) accompanied with dialogue system has been progressively prevalent in the medical domain. Nevertheless, the expertise of LLMs in Traditional Chinese Medicine (TCM) remains restricted despite several TCM LLMs proposed recently. Herein, we introduced TCMChat (https://xomics.com.cn/tcmchat), a generative LLM with pre-training (PT) and supervised fine-tuning (SFT) on large-scale curated TCM text knowledge and Chinese Question-Answering (QA) datasets. In detail, we first compiled a customized collection of six scenarios of Chinese medicine as the training set by text mining and manual verification, involving TCM knowledgebase, choice question, reading comprehension, entity extraction, medical case diagnosis, and herb or formula recommendation. Next, we subjected the model to PT and SFT, using the Baichuan2-7B-Chat as the foundation model. The benchmarking datasets and case studies further demonstrate the superior performance of TCMChat in comparison to existing models. Our code, data and model are publicly released on GitHub (https://github.com/ZJUFanLab/TCMChat) and HuggingFace (https://huggingface.co/ZJUFanLab), providing high-quality knowledgebase for the research of TCM modernization with a user-friendly dialogue web tool.

Technical limitations in spatial and single-cell omics sequencing pose challenges for capturing and describing multimodal information at the spatial scale. To address this, we develop SIMO, a computational method designed for the Spatial Integration of Multi-Omics datasets through probabilistic alignment. Unlike previous tools, SIMO not only integrates spatial transcriptomics with single-cell RNA-seq but expands beyond, enabling integration across multiple single-cell modalities, such as chromatin accessibility and DNA methylation, which have not been co-profiled spatially before. We benchmark SIMO on simulated datasets, demonstrating its high accuracy and robustness. Further application on biological datasets reveals SIMO’s ability to detect topological patterns of cells and their regulatory modes across multiple omics layers. Through comprehensive analysis of real-world data, SIMO uncovers multimodal spatial heterogeneity, offering deeper insights into the spatial organization and regulation of biological molecules. These findings position SIMO as a powerful tool for advancing spatial biology by revealing previously inaccessible multimodal insights.