基于降噪风险基因网络的生存风险基因筛选  

Survival risk gene screening based on noise reduction risk gene network

作　　者：陆震 刘艳[1] 李婧惟 Lu Zhen;Liu Yan;Li Jingwei(Department of Health Statistics, Harbin Medical University, Harbin 150081, China)

机构地区：[1]哈尔滨医科大学卫生统计学教研室,黑龙江哈尔滨150081

出　　处：《中国医院统计》2021年第2期126-130,共5页Chinese Journal of Hospital Statistics

基　　金：黑龙江省自然科学基金(LH2019H005);哈尔滨医科大学研究生科研和实践创新基金(YJSKYCX2019-74HYD)。

摘　　要：目的对基于降噪风险基因网络的生存风险基因筛选算法CoxLASSO-ISIS-N的表现作出评价,并与其他5种算法进行比较,分析算法的优劣,以期为高维基因表达数据的生存风险基因筛选提供一种新的思路。方法针对带有噪声的高维基因表达数据的生存风险基因筛选问题,分别利用模拟数据和真实数据,对6种算法(CoxLASSO、CoxLASSO-N、CoxLASSO-SIS、CoxLASSO-SIS-N、CoxLASSO-ISIS和CoxLASSO-ISIS-N)进行比较,分析算法的优劣。结果算法CoxLASSO-ISIS-N在模型的整体估计效果(LR和CS)、解释信息的比例(R2)以及一致性(CI)上均表现最优且最稳定。结论基于降噪风险基因网络的生存风险基因筛选算法CoxLASSO-ISIS-N可以对带有噪声的高维基因表达数据实现降噪,从而更精确地筛选生存风险基因,较好地反映死亡或其他结局发生与高维基因表达数据之间的关系,为临床诊断以及预后管理提供依据。Objective To evaluate the performance of the survival risk gene screening algorithm CoxLASSO-ISIS-N based on the noise-reduction risk gene network,and compare it with five other algorithms to analyze the pros and cons of the algorithm,with a view to providing a new idea of the survival risk gene screening of high-dimensional gene expression data.Methods Aiming at the survival risk gene screening problem of high-dimensional gene expression data with noise,simulated data and real data were used to analyze six algorithms(CoxLASSO,CoxLASSO-N,CoxLASSO-SIS,CoxLASSO-SIS-N,CoxLASSO-ISIS and CoxLASSO-ISIS-N)to compare and analyze the pros and cons of the algorithms.Results The algorithm CoxLASSO-ISIS-N has the best and most stable performance in the overall estimation effect of the model(LR and CS),the ratio of interpreted information(R2)and consistency(CI).Conclusion CoxLASSO-ISIS-N,a survival risk gene screening algorithm based on noise reduction risk gene network,can reduce noise on high-dimensional gene expression data with noise,so as to more accurately screen survival risk genes,better reflect the relationship between the occurrence of death or other outcomes and high-dimensional gene expression data and provide a basis for clinical diagnosis and prognosis management.

关 键 词：基因网络 基因表达数据 基因筛选 噪声 

分 类 号：R195.4[医药卫生—卫生统计学]