癌基因标记物的错综复杂关联对胃癌的准确诊断造成了负面影响，并导致死亡率的增加。由于多种细胞类型参与，分子异质性在确定胃癌的进展状态方面是不可避免的。识别致病基因签名对于理解该疾病的病因至关重要。本研究展示了一种系统的方法来识别与不同细胞类型相关的癌基因。采用“SingleR”算法，将邻近正常和胃癌样本的原始计数进行质量控制处理。利用主成分分析（PCA）和均匀流形逼近投影（UMAP）技术进行降维和多维聚类分析。利用人类原代细胞图谱数据库对邻近正常和胃癌样本的细胞群进行注释。利用轨迹分析描绘了不同细胞群之间的细胞状态转换。通过血管内皮生长因子（VEGF）与细胞群之间的配体-受体相互作用揭示了胃癌进展的关键分子途径。软骨细胞、平滑肌细胞和成纤维细胞细胞群中的基因根据生存分析的危险比与较低的存活率相关。通过将基因集与DisGeNET数据库进行比较，得到了与胃癌相关的癌基因标记物。进一步通过基因集富集分析验证了胃癌相关的12个生物标志物（SPARC, KLF5, HLA-DRB1, IGFBP3, TIMP3, LGALS1, IGFBP6, COL18A1, F3, COL4A1, PDGFRB, COL5A2）。药物-基因相互作用分析发现PDGFRB与多种抗癌药物有相互作用，可能是胃癌的潜在抑制剂。对这些分子标志物的进一步研究将有助于发展精准治疗，为胃癌患者带来更长的寿命。© 2023. The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.

The intricate association of oncogenic markers negatively impacts accurate gastric cancer diagnosis and leads to the proliferation of mortality rate. Molecular heterogeneity is inevitable in determining gastric cancer's progression state with multiple cell types involved. Identification of pathogenic gene signatures is imperative to understand the disease's etiology. This study demonstrates a systematic approach to identifying oncogenic gastric cancer genes linked with different cell types. The raw counts of adjacent normal and gastric cancer samples are subjected to a quality control step. The dimensionality reduction and multidimensional clustering are performed using Principal Component Analysis (PCA) and Uniform Manifold Approximation and Projection (UMAP) techniques. The adjacent normal and gastric cancer sample cell clusters are annotated with the Human Primary Cell Atlas database using the "SingleR." Cellular state transition between the distinct groups is characterized using trajectory analysis. The ligand-receptor interaction between Vascular Endothelial Growth Factor (VEGF) and cell clusters unveils crucial molecular pathways in gastric cancer progression. Chondrocytes, Smooth muscle cells, and fibroblast cell clusters contain genes contributing to poor survival rates based on hazard ratio during survival analysis. The GC-related oncogenic signatures are isolated by comparing the gene set with the DisGeNET database. Twelve gastric cancer biomarkers (SPARC, KLF5, HLA-DRB1, IGFBP3, TIMP3, LGALS1, IGFBP6, COL18A1, F3, COL4A1, PDGFRB, COL5A2) are linked with gastric cancer and further validated through gene set enrichment analysis. Drug-gene interaction found PDGFRB, interacting with various anti-cancer drugs, as a potential inhibitor for gastric cancer. Further investigations on these molecular signatures will assist the development of precision therapeutics, promising longevity among gastric cancer patients.© 2023. The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.