在生命科学的浩瀚星海中， 空间转录组学 正像一束穿透迷雾的光，把"细胞在哪里、在做什 么、在与谁对话"从想象变成可测量的事实。它让我们第一次有机会把分子表达、组织结构与 细胞命运放在同一张坐标系里理解——而这也意味着，真正的挑战不再是"有没有数据"，而 是"能不能把数据变成机制、把机制变成主图"。 我们相信，未来的空间组学竞争不在"知道使用哪个软件"，而在于谁能更快构建可解释的空间 坐标系，谁能更稳建立"位置-互作-信号-表型"的证据链，谁能把结果做成审稿人一眼认可的 主图。愿这门课程成为你2026开年的第一份"科研加速器"：让复现不再靠运气，让机制不再 停留在猜想，让每一次分析都能沉淀为可复用、可扩展、可发表的能力与作品。 论文介绍 本次复现的论文如下 4.源码级系统学习： https://www.nature.com/articles/s41586-024-08466-x 01 课程特色 1.全流程系统教学： 从空间转录组数据结构、QC、注释到出图全流程，配套"保姆级"R/Python代码模板，快速补齐空间数据分析 底层能力 2.一对一指导 + 包教包会： 针对你自己的数据（Xenium/VisiumHD ...

编辑丨王多鱼 排版丨水成文 细胞的功能不仅取决于其自身的基因表达，更取决于其在三维 （3D） 空间中的位置以及与周围微环境的 3D 互作，理解组织结构的 3D 复杂性是解析生物功能的关键。然而，当前主流的空间转录组学 （Spatial Transcriptomics，ST） 技术大多局限于二维 （2D） 平面，这种观测维度的局限导致我们难以还原器官 内部基因表达的连续梯度、细胞细胞微环境的立体分布以及精细的细胞互作网络。 尽管研究者们试图在Z轴上堆叠多张 2D 切片以近似三维结构，但受限于高昂的实验成本与有限的实验通 量，不得不对 Z 轴方向的采样密度做出妥协。这种折衷方案导致切片之间往往存在不可忽视的物理间距 （例如100微米 ，相当于缺失了约 5-10 层细胞的信息 ） ，使得最终获取的数据在Z轴上呈现为高度稀疏 的采样状态。这种采样方式将原本连续的生物组织割裂为离散的稀疏的断层面，使得现有的三维图谱往往 缺乏连续性与完整性。 2025 年 12 月 31 日，复旦大学 冯建峰 / 原致远 团队联合中国科学院计算技术研究所 赵屹 团队，在 Nature Methods 期刊发表了题为： Bridging ...

Core Insights - A new Siamese clustering framework named scSiameseClu has been proposed by a research team from institutions including the Chinese Academy of Sciences and Northeast Agricultural University, aimed at interpreting single-cell RNA sequencing (scRNA-seq) data effectively [1][4][5] - The framework addresses challenges in scRNA-seq data analysis, particularly the issues of representation collapse and the need for clearer cell population classification [1][4][5] Summary by Sections Introduction to scRNA-seq - Traditional bulk RNA sequencing focuses on average gene expression at the population level, potentially masking the characteristics of rare cells [1] - Single-cell RNA sequencing (scRNA-seq) captures comprehensive genetic information from individual cells, revealing hidden complexities [1] Challenges in scRNA-seq Data - scRNA-seq data is characterized by high noise, sparsity, and dimensionality, leading to issues such as "insufficient graph construction" and "representation collapse" even in advanced methods like graph neural networks (GNNs) [2][4] scSiameseClu Framework - The scSiameseClu framework integrates three key modules: Dual Augmentation, Siamese Fusion, and Optimal Transport Clustering, designed to capture and refine complex intercellular information [4][5][9] - The framework has shown superior performance in clustering and other biological tasks compared to state-of-the-art methods [5] Performance Evaluation - The performance of scSiameseClu was evaluated on seven real scRNA-seq datasets, which included samples from both mice and humans, covering various cell types [7][8] - The framework demonstrated significant advantages in clustering metrics such as Accuracy (ACC), Normalized Mutual Information (NMI), and Adjusted Rand Index (ARI) [14][15] Key Modules of scSiameseClu - **Dual Augmentation Module**: Enhances robustness against noise by simulating natural fluctuations in gene expression and generating augmented adjacency matrices for cell graphs [11] - **Siamese Fusion Module**: Integrates refined gene expression and cell graph matrices to learn robust and meaningful representations, improving clustering performance [12] - **Optimal Transport Clustering**: Aligns and corrects predicted distributions to ensure balanced clustering and avoid collapse [13] Experimental Results - The framework's performance was validated through extensive experiments, including comparisons with nine advanced benchmark models, showing consistent superiority across various datasets [14][15] - In downstream tasks, scSiameseClu accurately identified cell types and their marker genes, achieving over 90% similarity with gold standard references [15][17] Conclusion - The introduction of scSiameseClu represents a significant advancement in computational biology, effectively addressing long-standing challenges in cell heterogeneity analysis [20] - The framework exemplifies the integration of computer science methodologies with life sciences, paving the way for future innovations in the field [20]

Group 1 - The article discusses various innovative technologies showcased at the Data Expo, highlighting advancements in artificial intelligence and user interaction [1][22] - Xiaomi's AI assistant, Xiao Ai, integrates with smart devices to provide personalized services through voice commands, enhancing user experience [4] - JD.com's 3D product display technology allows users to interact with products in a virtual environment, improving online shopping experiences [8] Group 2 - Flexible display technology, introduced at the expo, offers design freedom and superior image quality, applicable in various devices from smartphones to medical equipment [13] - The 360 Smart Firewall utilizes AI to enhance network security by detecting threats and providing a user-friendly management interface [17] - Super microcomputers, a cutting-edge concept, leverage tiny computing units for efficient energy use and potential applications in health monitoring and environmental sensing [21] Group 3 - Big data genetic prediction technology analyzes genomic data to forecast individual health risks and responses to medications, promoting personalized healthcare [27]

Core Insights - The presentation by Associate Professor Zhang Shugang from Ocean University of China focuses on the construction and application of an intelligent protein computing system, highlighting breakthroughs in traditional protein research challenges [1][3][4]. Group 1: Traditional Challenges in Protein Research - Proteins play a crucial role in biological functions but face challenges such as high structural analysis costs, delayed functional annotation, and low efficiency in novel protein design [1][3]. - The demand for understanding complex protein characteristics has increased, necessitating innovative approaches to overcome these challenges [1][3]. Group 2: AI-Driven Innovations - The introduction of AI technologies has revolutionized protein research, exemplified by the awarding of the 2024 Nobel Prize in Chemistry for breakthroughs in AI-driven protein structure prediction and design [3][4]. - The intelligent protein computing system enables significant advancements in large-scale functional annotation, interaction prediction, and 3D structure modeling, providing new technical pathways for drug discovery and life system simulation [1][3]. Group 3: Key Breakthroughs in Protein Computing - The core tasks of intelligent protein computing include: 1. **Protein Structure Prediction**: AlphaFold's models have achieved unprecedented accuracy in predicting protein structures, with AlphaFold2 providing atomic-level precision and AlphaFold3 extending capabilities to predict interactions with various biomolecules [4][5]. 2. **Functional Annotation**: The team has developed methods to automate protein function annotation using deep learning, significantly increasing the scale of data processed and improving prediction accuracy [6][7]. 3. **Interaction Prediction**: A self-developed model has been created to enhance the prediction of protein interactions, achieving over 95% accuracy in specific applications [16][20]. 4. **Protein Design**: The potential for designing new proteins has been demonstrated, with innovative approaches being explored for applications in vaccine development and cancer treatment [22]. Group 4: Multiscale Modeling in Life Systems - The research emphasizes the importance of multiscale modeling in understanding complex life systems, integrating various biological scales from molecular to cellular levels [23]. - The team has proposed a comprehensive modeling framework that encompasses multiple research points, aiming for a holistic simulation of life systems [23].

Core Insights - Chen Runsheng is a pioneer in non-coding gene research and a participant in the Human Genome Project, which is one of the largest life science projects globally [1][2] - He emphasizes that the future of AI in China lies not in the quantity of chips but in the density of intelligent computing [1] Group 1: Contributions to Genomics - Under Chen's leadership, China became the sixth country globally to possess large-scale genome sequencing capabilities [2][6] - In 1999, China joined the International Human Genome Project, taking on the task of sequencing approximately 30 million base pairs of the human chromosome 3 short arm, which represented 1% of the entire project [6] - Chen's team innovated sequencing methods, completing their tasks two years ahead of schedule, demonstrating significant advancements in genomic research [6] Group 2: Discoveries in Non-Coding DNA - Chen discovered that only 2%-3% of the human genome encodes proteins, while 97% consists of non-coding sequences previously deemed "junk DNA" [6][7] - His team focused on these non-coding regions, leading to the identification of new disease-related loci, particularly in cancer research [7] Group 3: Open Science and Collaboration - Since 1993, Chen's team has established a comprehensive database of 640,000 non-coding molecular information, which they chose to share openly with the global scientific community [7] - Chen believes that science is a collective human contribution and emphasizes the importance of sharing research findings for the advancement of knowledge [7] Group 4: AI and Future Innovations - Chen has been involved in AI research since the late 1980s, applying artificial neural networks to predict coding genes [8] - His current work involves integrating traditional Chinese medicine data into AI models, aiming to create a platform that merges different medical perspectives [8] - He advocates for viewing AI not merely as a tool but as a new center for innovation, which could lead to more creative possibilities in research and development [8]