结直肠癌是常见的消化道恶性肿瘤之一，严重地危害人类身体健康。临床上缺乏有效的针对二期和三期病人的术后复发风险标志物。基于肿瘤组织基因表达值构建的风险模型容易受到实验条件等批次效应的影响，使得现有分子标志物的可重复性降低。此外，传统的转录组数据无法反映细胞亚群的基因表达水平，单细胞转录组技术为寻找更加精确的复发风险标志物提供了可能。申请人前期基于基因表达秩序关系识别了具有跨数据集性能的个体化药效预测标志物。本课题将在前期的研究工作基础上（1）构建单细胞转录组差异表达基因筛选的数据分析流程;（2）从结直肠癌的单细胞转录组中识别细胞亚群差异分子特征与复发风险基因对标志物;（3）利用多维组学数据，识别不同复发风险亚型的信号通路。本课题的研究结果对利用单细胞转录组数据识别结直肠癌复发风险标志及揭示预后亚型的多维度组学机制具有一定的理论与应用价值，为改善结直肠癌患者预后提供参考。

Colorectal cancer (CRC) is one of the most lethal gastrointestinal malignancies. There were few effective recurrence risk signatures for stage II/III CRC patients in clinical practice. Most of the risk-score-based signatures were based on the gene expression abundance, which could be sensitive to the experimental batch effects and had low reproducibility. In addition, bulk tumor-based gene expression profiles could not accurately characterize subpopulations of cancer cells. Single-cell sequencing provides a high-resolution platform for prognostic signatures detection from subpopulations of cells. We have identified an individualized prognostic signature for stage II/III CRC patients treated with 5-fluorouracil-based therapy. Based on the previous study, we will propose and evaluate data analysis strategies to detect differentially expressed genes (DEGs) from single-cell transcriptomes based on relative expression orderings. Next, with DEGs from subpopulations of cells, we will screen recurrence risk gene pair signature for stage II/III CRC patients. To this end, using the multi-omics data of CRC, we will further reveal the signaling pathways for patients with different recurrence risk. Results of this research would have theoretical and practical values for identification of recurrence risk signatures based on single-cell transcriptomes and reveal multidimensional omics mechanisms of prognostic of CRC.