Tsuji Toshiyuki, Yoda Takao, Shirai Tsuyoshi
Nagahama Institute of Bio-Science and Technology, and Japan Science and Technology Agency, Bioinformatics Research Division, Nagahama, Shiga 526-0829, Japan.
Sci Rep. 2015 Nov 9;5:16341. doi: 10.1038/srep16341.
Many biological molecules are assembled into supramolecules that are essential to perform complicated functions in the cell. However, experimental information about the structures of supramolecules is not sufficient at this point. We developed a method of predicting and modeling the structures of supramolecules in a biological network by combining structural data of the Protein Data Bank (PDB) and interaction data in IntAct databases. Templates for binary complexes in IntAct were extracted from PDB. Modeling was attempted by assembling binary complexes with superposed shared subunits. A total of 3,197 models were constructed, and 1,306 (41% of the total) contained at least one subunit absent from experimental structures. The models also suggested 970 (25% of the total) experimentally undetected subunit interfaces, and 41 human disease-related amino acid variants were mapped onto these model-suggested interfaces. The models demonstrated that protein-protein interaction network modeling is useful to fill the information gap between biological networks and structures.
许多生物分子组装成超分子，这些超分子对于在细胞中执行复杂功能至关重要。然而，目前关于超分子结构的实验信息并不充分。我们通过结合蛋白质数据库（PDB）的结构数据和IntAct数据库中的相互作用数据，开发了一种预测和模拟生物网络中超分子结构的方法。IntAct中二元复合物的模板是从PDB中提取的。通过组装具有叠加共享亚基的二元复合物来尝试进行建模。总共构建了3197个模型，其中1306个（占总数的41%）包含至少一个在实验结构中不存在的亚基。这些模型还提示了970个（占总数的25%）实验未检测到的亚基界面，并且41个人类疾病相关的氨基酸变体被映射到这些模型提示的界面上。这些模型表明，蛋白质-蛋白质相互作用网络建模有助于填补生物网络和结构之间的信息空白。
