面向个体癌症病人驱动基因预测的网络控制模型研究

从个体癌症病人基因组学数据中发现个体特异癌症驱动基因，对理解癌症异质性和精准药物设计有着重要作用。由于个体癌症病人的有效样本数量通常非常有限，传统大样本统计方法获得统计指标的精确性以及可靠性会大大降低，很难有效预测个体癌症驱动基因。鉴于此，本项目将个体癌症病人看作一个动态网络系统，拟应用网络结构控制理论，建立适用于个体癌症病人组学数据分析的网络控制模型，预测癌症个体化驱动基因，挖掘驱动基因重要模式和规律：首先，考虑个体癌症病人基因间调控权重信息，设计加权网络控制模型，筛选重要个体癌症病人驱动基因；其次，考虑个体癌症病人基因调控关系的多时间尺度信息，设计时序网络控制模型，预测个体癌症驱动基因的动态变化规律；最后，利用计算生物学方法，设计个体癌症病人驱动基因预测模型评价框架，评估模型预测性能。本项目旨在从网络结构控制的角度，阐明癌症基因组的个体异质性，为癌症精准医疗提供理论依据。

The genomic profiles of cancer patients are diverse and heterogeneous, which is believed to be responsible for heterogeneous drug response.Moreover, the complicated and multiple driver genes in individual cancer patients may cause resistance to individual drugs.It is an important challenge for finding personalized driver genes to improve the outcome of cancer patients, especially for patients with drug resistance. Since the number of samples of an individual patient is usually limited, the accuracy and reliability of statistical methods based on large number of samples will greatly be reduced for predicting personalized driver genes. By using network structural controllability theory on genomic omics data of cancer patients, this project established network control models for predicting personalized driver genes: firstly, weight network control models were designed to screen important personalized driver genes by considering the weight information of gene regulations; then, by considering the multi-time scale information of gene regulations,temporal network control models were designed to mine the dynamic change of personalized driver genes; finally, by using computational biology methods, we designed the computational framework for evaluating the performance of personalized driver gene prediction model. This project aims to demonstrate the individual heterogeneity of cancer genomic omics data and provide theoretical foundation for cancer precision medicine from the perspective of network structural controllability theory.