蛋白质亚核定位及其特征信息的理论研究

On the basis of protein subcellular localization database, the subnuclear protein of the Nuclear Protein Database was integrated. Further, the nuclear and non-nuclear protein sub-database was built and a dataset consisting of proteins with multi-localizations was constructed. After analysis of feature motif, chemical shift and protein interaction, the protein feature parameters which are related to protein subnuclear localization were selected and the extracting method of gene ontology was perfected. Combined the classical amino acids composition, physicochemical properties information with these feature parameters, all the feature parameters were integrated by using increment of diversity and covariant discriminant function. On this basis, the predictive model was proposed. First, the nuclear and non-nuclear protein was identified. Second, the protein subnuclear localization was predicted further. Finally, the valid predictive algorithm and theory were proposed to predict the multi-localization proteins. The selecting of feature parameter and predictive method were improved constantly according to predictive results, and the protein subnuclear localization predictive ability was enhanced. These provide theoretical guidance for next experimental research and help to study the protein biological function.

本项目在已有的蛋白质亚细胞定位数据库基础上，整合核定位蛋白数据库（The Nuclear Protein Database ）中的核定位蛋白，进一步构建和完善核定位蛋白和非核定位蛋白数据子库及多定位蛋白质亚核定位数据子库。深入分析不同亚核定位蛋白所包含的特征模体、化学位移、蛋白质相互作用信息，并提取与蛋白质亚核定位相关的蛋白质特征信息参数，完善基因本体（GO）信息的提取方法，将传统的蛋白质氨基酸组份、物理化学性质信息与这些信息参数相结合，利用离散增量和协变判别式进行特征信息参数的组合，寻找最佳组合参数，提出预测模型，首先识别核定位蛋白和非核定位蛋白，然后再对核定位蛋白的亚核定位进一步预测，并给出预测多定位蛋白亚核定位的有效模型和理论。根据预测结果不断改进和完善特征信息参数的选取和预测算法，提高蛋白质亚核定位的预测能力。为进一步的实验研究提供理论指导，为蛋白质生物学功能的研究提供帮助。

本项目构建了核定位蛋白和非核定位蛋白数据库以及多定位蛋白质亚核定位数据库并对数据库进行不断地更新、完善。深入分析了核定位蛋白的模体、化学位移、结构域和蛋白质相互作用特征，完善了基因本体（GO）信息的提取方法，在此基础上把这些特征信息参数与传统的氨基酸组份和氨基酸物理化学性质信息进行参数组合，寻找到最佳组合参数，提出了理论预测模型，对核定位蛋白和非核定位蛋白以及核定位蛋白的亚核定位进行了识别，取得了较高的预测成功率。对多定位亚核蛋白质序列的特征信息进行了分析，给出了多定位蛋白质亚核定位的预测模型。