肺腺癌致癌关键分子的靶向干预是治疗原发性肺腺癌的关键，但现有的关键分子存在特异性和敏感性不足等问题，致使治疗效果不甚理想。测序技术的迅猛发展为肺腺癌研究开辟了新途径，但由于技术平台、试剂仪器的差异,导致数据存在偏好性，使得常规的整合手段难以从海量的测序数据中提取出真正的关键分子。有鉴于此，本项目拟建立基于相对/全局多维度显著性评测结合新型wrapper feature筛选的integrative分析新策略，该策略具有高度鲁棒性和良好的收敛性，对海量测序数据综合评测，而后结合多个数据库的底层信息确立致癌关键分子及其信号通路，并对其进行信息学和生物实验验证（细胞、动物、临床样本三个层次）。以上新策略已在75对肺腺癌/癌旁临床测序数据中进行预实验，其筛选出的关键分子在生物实验验证中表现优越，后期该模型将应用到大规模数据中。本策略不仅能获得肺腺癌致癌关键分子，更为多中心数据处理提供了新思路。

The carcinogenic molecules of lung adenocarcinoma plays an important role in treatment of primary carcinoma. The specificity and sensitivity of carcinogenic molecules are insufficient to achieve satisfactory treatment in clinic. The sequencing technique provides new approaches for researching of lung adenocarcinoma, however, data preference still exists in the approach which are caused by the differences of platforms, reagents and instruments. These differences will interfere in common meta-analysis methods to cover the meaningful lung adenocarcinoma biomarker when analyze mass sequencing data. To deal with these issues, this project will apply the novel integrative analysis which has robustness, high fault tolerances and strong convergences. The strategy is: applying re-contribute “genome-wide relative significance” and “genome-wide global significance” to mass sequencing data, and selecting carcinogenic molecules via “wrapper feature selection” model. Then, the combining results of several database will be applied to analyze the carcinogenic molecules and the corresponding networks. The evaluation of the molecules and the signaling network by computational simulation as well as biological experiments (three levels: cell lines, animal and clinic samples) will enable us to obtain reliable and clinically valuable molecules. The preliminary experiment has been done in 75 pairs sequencing data of lung adenocarcinoma tissues and para-cancer tissues, and the three carcinogenic molecules were found and validated that they can be used for distinguishing cancer and para-cancer. Next work of this project will improve the strategy to fit big amount sequencing data, and validate the analysis results with dry and wet lab experiments based on cell lines, mouse and clinical samples. In addition, the project not only demonstrates the key carcinogenic molecules in the signaling network regarding the lung adenocarcinoma mechanisms, but also provides new approaches to gain insights and theoretical basis into mass data processing.