急性髓系白血病（AML）是一种恶性克隆性造血系统疾病，预后较差。了解白血病细胞与肿瘤微环境（TME）之间的相互作用有助于预测白血病的预后并指导其治疗。使用基于 Python 的管道（包括基于机器学习的 scVI 工具）重新分析来自 CSC 和 G20 队列的 scRNA-seq 数据，概括了 AML 患者样本中独特的层次结构。进行加权相关网络分析（WGCNA）来构建加权基因共表达网络并识别主要关注造血干细胞（HSC）、多能祖细胞（MPP）和自然杀伤（NK）细胞的基因模块。分析揭示了与有氧呼吸和核糖体/细胞质翻译相关的基因模块的显着失调。 CellChat 软件包阐明了细胞间通讯，揭示了激活和抑制性免疫信号通路的不平衡。拦截白血病 HSC 中上调的基因

Acute myeloid leukemia (AML) is a malignant clonal hematopoietic disease with a poor prognosis. Understanding the interaction between leukemic cells and the tumor microenvironment (TME) can help predict the prognosis of leukemia and guide its treatment. Re-analyzing the scRNA-seq data from the CSC and G20 cohorts, using a Python-based pipeline including machine-learning-based scVI-tools, recapitulated the distinct hierarchical structure within the samples of AML patients. Weighted correlation network analysis (WGCNA) was conducted to construct a weighted gene co-expression network and to identify gene modules primarily focusing on hematopoietic stem cells (HSCs), multipotent progenitors (MPPs), and natural killer (NK) cells. The analysis revealed significant deregulation in gene modules associated with aerobic respiration and ribosomal/cytoplasmic translation. Cell-cell communications were elucidated by the CellChat package, revealing an imbalance of activating and inhibitory immune signaling pathways. Interception of genes upregulated in leukemic HSCs & MPPs as well as in NKG2A-high NK cells was used to construct prognostic models. Normal Cox and artificial neural network models based on 10 genes were developed. The study reveals the deregulation of mitochondrial and ribosomal genes in AML patients and suggests the co-occurrence of stimulatory and inhibitory factors in the AML TME.© 2024. The Author(s), under exclusive licence to Springer Nature America, Inc.