乳腺癌相关的lncRNA-mRNA共表达扰动网络构建

Construction of a network of lncRNA-mRNA co-expression perturbation related to breast cancer

目的基于复杂生物网络和机器学习方法,识别乳腺癌相关的边缘生物标志物,构建乳腺癌生存预后模型,从而在系统水平解释乳腺癌的发生发展机制。方法首先基于TCGA数据库的RNA-seq数据识别乳腺癌相关的lncRNA-mRNA共表达扰动关系对,进一步构建乳腺癌相关的lncRNA-mRNA共表达扰动网络并对网络中的关键基因进行通路富集分析。然后,基于乳腺癌相关的lncRNA-mRNA关系对,构建乳腺癌预测的分类器模型。最后,通过Lasso回归筛选变量构建多因素Cox比例风险回归模型对乳腺癌患者进行生存预后分析。结果构建了乳腺癌相关的lncRNA-mRNA共表达扰动网络,其中的关键基因富集分析得到32条与乳腺癌相关的生物通路。分类预测模型的灵敏度、特异度和准确性分别为98.2%、85.2%、97.6%。Lasso回归共筛选出22个和乳腺癌生存预后显著相关的lncRNA-mRNA互作关系对,进而构建的生存预测模型把训练集和测试集的乳腺癌患者分为高风险组和低风险组,两组患者生存预后均存在明显差异。结论LncRNA-mRNA共表达互作网络中的关键基因以及乳腺癌相关的边缘生物标志物大多被证明与乳腺癌相关。同时基于边缘生物标志物的预后模型可以稳健地预测乳腺癌患者的生存预后状态,有利于从网络层面更好地理解乳腺癌的发生发展机制。

Objective Based on the complex biological networks and machine learning methods,the edge biomarkers associated with breast cancer were identified and the survival prognosis model of breast cancer was constructed to further explain the occurrence and development of breast cancer at a systematic level.Methods Firstly,based on the RNA-seq data of TCGA,we identified lncRNA-mRNA co-expression perturbation pairs and constructed the co-expression perturbation networks related to breast cancer.Further,we conducted the pathway enrichment analysis of key genes in the network.Then,a classifier model for breast cancer prediction was constructed based on the co-expression perturbation pairs.Finally,a multivariate Cox proportional risk regression model was established by screening variables using Lasso regression to analyze the survival prognosis of breast cancer patients.Results We constructed a lncRNA-mRNA co-expression perturbation network associated with breast cancer.And the key genes in the network were used to perform pathway enrichment analysis.A total of 32 biological pathways associated with breast cancer were obtained.The sensitivity,specificity,accuracy of the classifier model for breast cancer prediction were 98.2%,85.2%and 97.6%respectively.A total of 22 lncRNA-mRNA interaction pairs,which were significantly associated with breast cancer survival prognosis,were identified by Lasso regression.Based on the survival prediction model,breast cancer patients in the training set and the test set were divided into high-risk group and low-risk group,and the survival prognosis of patients in the two groups was significantly different.Conclusions Key genes in the lncRNA-mRNA co-expression perturbation network and breast cancer-related edge biomarkers have mostly been proved to be associated with breast cancer.Meanwhile,the prognostic model based on edge biomarkers can predict the survival and prognosis of breast cancer patients robustly.This paper is helpful to better understand the mechanism of the occurrence and development of breast cancer in network level.