蛋白质异常定位组是指在疾病，药物干预或环境压力下亚细胞定位发生变化的蛋白质。这类蛋白质对于复杂疾病早期诊断及治疗具有重要意义，但却无法被现有的蛋白质亚细胞定位预测方法有效识别。本项目研究蛋白质异常定位组的计算预测方法：通过融合基因表达谱，基因突变，蛋白质翻译后修饰改变，蛋白质相互作用网络等多种异质数据，构造不同细胞条件下具有结构动态性的蛋白质相互作用网络模型；综合利用网络模型的局部拓扑特征和基因表达模式特征，建立不同细胞条件下的蛋白质亚细胞定位预测器；通过比较不同条件下的蛋白质亚细胞定位预测结果，预测蛋白质的异常定位组，再利用文献中已有报道或直接实验对预测结果进行验证。本项目在原理设计，计算模型，预测技术等方面均打破了现有研究工作中的常见框架，致力于建立一个通用的预测蛋白质异常定位组的计算方法，并希望通过其应用为生命科学研究，复杂疾病诊断及药物发现提供可借鉴的经验。

The protein mislocalizome refers to the set of proteins, which have altered subcellular localizations or subcellular distributions in different cellular conditions, such as diseases, drug therapies or environmental stresses. Understanding the functions of these proteins are helpful in disease diagnostics and drug discovery. However, few methods can be applied in predicting protein mislocalizome. In this project, we will utilize heterogeneous data resources, including the gene expression patterns, gene mutations, protein-protein interaction networks and post-translational modifications, to establish structural flexible protein-protein interaction network models for different cellular conditions. Based on this model, we will predict protein subcellular locations proteome-wide under various conditions. By comparing the predicted subcellular locations under different conditions, we can systematically identify the mislocalized proteins, which will be verified by either the reports in other literatures or direct experiments. In general, we will try to develop a universal method that can identify protein mislocalizome in various conditions. By applying this methods on several given diseases and drug therapies, the predicted mislocalizome are expected to be useful in both theoretical and practical studies.