PO.BCS01.12 · 生物信息与计算
MeDOC-KB：用于揭示肥胖相关癌症之间代谢关联的知识库
MeDOC-KB: Knowledge base for unraveling the metabolic links between obesity-related cancers
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
代谢失调与肥胖癌症风险协作组（MeDOC）是一个由 NCI 资助的项目，采用团队科学和跨学科方法，阐明肥胖、代谢失调与癌症风险之间的关联机制。肥胖和代谢失调都会导致脂肪细胞功能、生长因子、炎症、肠道微生物组、免疫功能、性激素、脂质和葡萄糖代谢等一系列紊乱，而这些又可扰乱与癌症发生和进展相关的多条下游信号通路。鉴于这种复杂性，整合多组学、小鼠模型和流行病学数据至关重要。MeDOC 知识库（MeDOC-KB）是一个综合性图谱，编目各种关联，以连接来自协作组研究和外部文献的肥胖、代谢失调与癌症风险。一个基于大语言模型的智能体协调了来自代谢组学、蛋白质组学和脂质组学平台（如 Nightingale、Olink 和 Metabolon）的生物标志物，规范命名并将跨平台同义词解析为统一的词汇表。MeDOC-KB 采用 Neo4j 图数据库架构，以高效遍历生物标志物、出版物、队列和癌症部位之间的复杂关系。文献数据被提取为四张核心表（引文、队列、方法和关联），并转换为图结构，其中节点代表出版物、队列、生物标志物和癌症部位，边则表示它们之间的关系。MeDOC-KB 可通过交互式 R Shiny 网络应用访问。当前版本包含 42,838 个节点和 189,709 条关系，其中 21,945 条为生物标志物-癌症关联，涵盖 1,645 个生物标志物和 15 种癌症结局，包括 13 种已知肥胖相关癌症中的 11 种。例如，MeDOC-KB 揭示了胰岛素样生长因子结合蛋白-1（IGFBP-1）在多个队列中与子宫内膜癌、结直肠癌和胰腺癌呈负相关，提示存在共同的代谢机制。该平台使研究人员能够识别与肥胖诱导的代谢失调相关的、跨癌症类型的生物标志物模式，跨队列比较研究结果，并发现研究不足的生物标志物-癌症关系。MeDOC-KB 为关于机制通路的假设生成、跨人群的生物标志物验证，以及为高风险代谢人群识别有前景的癌症预防策略靶点，提供了一项关键资源。该知识库公开可访问，并将随协作组研究成果和文献持续更新。
查看英文原文 English abstract
Metabolic Dysregulation and Obesity Cancer Risk Consortium (MeDOC) is an NCI-sponsored program using team science and transdisciplinary approach to elucidate mechanisms linking obesity, metabolic dysregulation, and cancer risk. Both obesity and metabolic dysregulation contribute to a cascade of derangements in adipocyte function, growth factors, inflammation, gut microbiome, immune function, sex hormones, lipid and glucose metabolism which, in turn, can disrupt several downstream signaling pathways related to cancer initiation and progression. Given this complexity, integrating multi-omic, mouse models and epidemiologic data is critical. The MeDOC-Knowledge Base (MeDOC-KB) is a comprehensive atlas cataloging associations to link obesity, metabolic dysregulation, and cancer risk from consortium studies and external literature. A large language model-based agent harmonized biomarkers derived from metabolomics, proteomics, and lipidomics platforms such as Nightingale, Olink, and Metabolon, standardizing nomenclature and resolving cross-platform synonyms into a unified vocabulary. MeDOC-KB uses a Neo4j graph database architecture to enable efficient traversal of complex relationships among the biomarkers, publications, cohorts, and cancer sites. Literature data are extracted into four core tables (Citation, Cohort, Methods, and Association) and transformed into a graph structure where nodes represent publication, cohorts, biomarkers, and cancer sites while edges denote their relationships. MeDOC-KB is accessible through an interactive R Shiny web application. The current version contains 42,838 nodes and 189,709 relationships with 21,945 being biomarker-cancer associations spanning 1,645 biomarkers and 15 cancer outcomes, including 11 of the 13 known obesity related cancers. For example, MeDOC-KB reveals that insulin-like growth factor binding protein-1 (IGFBP-1) shows inverse associations with endometrial, colorectal, and pancreatic cancers across multiple cohorts, suggesting shared metabolic mechanisms. The platform enables researchers to identify biomarker patterns associated with obesity-induced metabolic dysregulation across cancer types, compare findings across cohorts, and discover understudied biomarker-cancer relationships. MeDOC-KB provides a critical resource for hypothesis generation regarding mechanistic pathways, biomarker validation across populations, and identification of promising targets for cancer prevention strategies in high-risk metabolic populations. The knowledge base is publicly accessible and will be continuously updated with consortium findings and literature.
利益披露 Disclosure
M. Subramanian, None..
S. Rosin, None..
S. Hall, None..
N. Grover-Fairchild, None..
K. Robien, None..
L. DiPietro, None..
M. Temprosa, None.