Objective Using mRNA/miRNA/protein expression profiles to identify the molecular biomarker for pancreatic ductal adenocarcinoma(PDAC). Method We used miRNA , mRNA and protein ex-pression data of PDAC to construct the molecular regulatory network by the interaction of"transcription factor (TF) target miRNA" and"miRNA target gene", and based on this network to infer some candidate genes and miRNAs which were related to the molecular mechanism of PDAC. Result We found a key network related to 6 miR let-7 family members (hsa-let-7i、hsa-let-7d、hsa-let-7b、hsa-let-7g、hsa-let-7c、hsa-let-7e) and 5 differently expressed genes(ALDH1A1、TGFB1、ANXA2、FN1、THBS1), all genes and miRNA were PDAC related molecular in this key network. Three large-scale expression profile of PDAC was used to test the 5 genes. The results shown that the prediction accuracy were 93%(91%~94%)、90%(89%~92%)、88%(87%~89%) in three expression datasets, respectively. Conclusion Five genes identified by our new methods with multidimensional data minning show a great classified power for PDAC, and should be new biomarkers for PDAC.%目的：整合高通量miRNA/mRNA/蛋白质表达数据，寻找胰腺导管腺癌（pancreatic ductal ade-nocarcinoma, PDAC）分子标志物（Biomarker）。方法本文提出利用PDAC样本的mRNA、miRNA与蛋白质表达谱数据，通过“转录因子（TF）调控miRNA”与“miRNA靶向调控基因”的关系构建在PDAC发病过程中的核心调控网络，并据此预测定义选候PDAC相关mRNA与miRNA，同时利用差异表达蛋白进一步定义调控网络。结果发现6个miR let-7家族miRNA成员（hsa-let-7i、hsa-let-7d、hsa-let-7b、hsa-let-7g、hsa-let-7c、hsa-let-7e）与5个在基因与蛋白层面同时差异表达的基因（ALDH1A1、TGFB1、ANX-A2、FN1、THBS1）形成的一个核心调控网络，其网络中涉及的5个基因全都为已知与PDAC发生发展明确相关的分子。利用5个基因对3套大规模的PDAC表达数据进行分型分析，结果发现其分类准确性分别达到了93%（91%～94%）、90%（89%～92%）与88%（87%～89%）。结论通过miRNA/mRNA/蛋白三种数据整合分析得到的5候选标记基因具有良好的分型能力，可作为新的PDAC分型Biomarker。
展开▼
