化学计量学在基因表达定量解析及临床预测模型构建中的应用研究

项目名称： 化学计量学在基因表达定量解析及临床预测模型构建中的应用研究

项目编号： No.21205085

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 分析化学

项目作者： 文志宁

作者单位： 四川大学

项目金额： 25万元

中文摘要： 癌症的预后是针对癌症可能的病程及治疗结果的预测，是近年来生物信息学的研究热点。微阵列基因芯片技术的问世，为各类癌症分子水平的研究提供了有效的方法，因此，建立基于基因表达的癌症预后模型及提高模型的预测能力是亟待解决的关键问题。美国食品和药物管理局发起的MAQC-II项目，对基于基因表达谱的共30,000多个预测模型进行考察，发现对于一些复杂癌症，模型预测能力不足，例如对多发性骨髓瘤病人存活率预测结果的AUC值仅为0.615。 癌症样本的复杂性对基因表达谱的检测结果有很大的影响，进而干扰模型的预测能力。本项目拟采用多元分辨与校正的方法，对于不同类别癌症，分别从其复杂样本基因表达数据中对肿瘤细胞表达谱进行分离，降低非肿瘤细胞造成的干扰；进而结合生物统计方法与基因功能分析，筛选对应类别癌症的特征基因；最后利用模式识别的方法，重建预测模型，以期提高MAQC项目中三类复杂癌症的预后结果。

中文关键词： 基因表达；定量分析；临床样本；癌症预后；特征筛选

英文摘要： The prognosis for a cancer includes the expected duration and the likely outcomes of this kind of disease, and is a hotspot in bioinformatics researches. Since Brown and his colleagues developed microarray system in 1995, researchers can monitor the expression of thousands of genes simultaneously. Different biological events in cell are closely associated with gene expression. Therefore, Gene expression profiling (GEP) becomes a powerful technique for exploring global expression patterns of cells, and is widely applied to seek biomarkers of cancers for well understanding the pathogenetic mechanism at molecular level. Therefore, it is important to create the prediction models for cancer prognosis and improve the prediction accuracy of them based on gene expression data. MAQC-II project, which was launched by U.S. Food and Drug Administration, investigated more than 30,000 predictive models generated by 36 independent research groups and found that the prediction accuracy is not satisfied for some complex cancers. For example, the AUC of predicting the overall survival of multiple myeloma patients was only 0.615. The complexity of clinical cancer samples will greatly impact the detection results of microarray gene chip and will also lower the prediction ability of the models. In this project, we will separate the

英文关键词： Gene expression；Quantitative analysis；Clinical sample；Cancer prognosis；Feature selection