基于马尔可夫随机场识别急性髓系白血病核心基因

Identifying Core Genes of Acute Myeloid Leukemia Based on Markov Random Field

摘要: 在大量的候选基因中筛选出关联急性髓系白血病(AML)的核心基因对于其治疗具有重要作用，鉴于肿瘤细胞高度异质性的特点，单细胞测序数据为AML生物标志物的筛选提供了可能。本研究基于马尔科夫随机场和高斯混合模型整合DNA甲基化和单细胞测序数据，将正常组和疾病组间基因互作网络的差异重连信息纳入马尔科夫随机场框架，通过重连网络和蛋白互作网络鉴定出CD86、TNF、GRAP2、FGFR1、IL18五个免疫相关的预后基因。基因GRAP2、FGFR1、TNF的高表达与预后较好有关，而DNA甲基化水平与预后较差有关，基因CD86、IL18的高表达与预后较差有关，而DNA甲基化水平与预后较好有关，CD86、TNF、GRAP2、FGFR1、IL18表达水平与DNA甲基化水平呈负相关；基因CD86表达水平与Monocytes、Macrophages M2免疫浸润水平显著正相关，与Mast cells resting、naïve B cell、naïve CD4 T cell、CD8 T cell、NK cell resting免疫浸润显著负相关，基因TNF、GRAP2、FGFR1反之。基因CD86、TNF、GRAP2、FGFR1、IL18可以作为AML发生、发展及免疫治疗中的预后生物标志物，为进一步研究提供依据。

Abstract: Screening out the core genes associated with acute myeloid leukemia (AML) from a large number of candidate genes plays an important role in its treatment. In view of the high heterogeneity of tumor cells, single-cell sequencing data provides the possibility to screen for AML biomarkers. The research is based on Markov random field and Gaussian mixture model to integrate DNA methylation and single-cell sequencing data, and incorporates the difference reconnection information of the gene interaction network between the normal group and the disease group into the Markov random field framework, through the reconnection network and protein interaction network identified five immune-related prognostic genes CD86, TNF, GRAP2, FGFR1, and IL18. High expression of genes GRAP2, FGFR1, TNF is related to better prognosis, while DNA methylation level is related to poor prognosis, high expression of genes CD86 and IL18 is related to poor prognosis, and DNA methylation level is related to better prognosis, CD86, TNF, GRAP2, FGFR1, IL18 expression levels are negatively correlated with DNA methylation levels; Gene CD86 expression level is significantly positively correlated with Monocytes and Macrophages M2 immune infiltration levels, and significantly negatively correlated with Mast cells resting, naïve B cell, naïve CD4 T cell, CD8 T cell, and NK cell resting immune infiltration levels. Genes TNF, GRAP2, FGFR1 are on the contrary. Genes CD86, TNF, GRAP2, FGFR1, IL18 can be used as prognostic biomarkers in the occurrence, development and immunotherapy of AML, providing a basis for further research.