RicePPINet - A computational Interactome for Rice
RicePPINet covers 16,895 non-transposable-element (non-TE) related proteins with 708,819 high-confidence interactions (download riceppinet full dataset) at a 71% true positive rate. The power of the network for discovering novel protein interactions is demonstrated by a comparison with other publicly available PPI prediction methods as well as by experimentally determined PPI datasets. Furthermore, global analysis of domain-mediated interactions reveals RicePPINet well reflects PPIs at the domain level. Our studies have demonstrated the efficiency of RicePPINet-based methods in prioritizing candidate genes involved in complex agronomic traits, is approximately 2-11 times better than random prediction. RicePPINet provides an expanded landscape of predicted interactome, which will certainly facilitate genetic dissection of agronomically important traits in rice.
Search the database!
search against RicePPINet
Features used in machine learning to predict rice protein-protein interactions
Coverage(1): the number of protein pairs with an available given feature was divided by the total number of possible pairs.
||Predicted protein interaction
|Preserved interface size
|Fraction of preserved interface
|Biological process ontology
|Molecular function ontology
|Cellular component ontology
Coverage(2): the number of predicted protein-protein interactions with an available given feature was divided by the total number of predicted interactions.
Please cite: Shiwei Liu, Yihui Liu, Jiawei Zhao, Shitao Cai, Hongmei Qian, Kaijing Zuo, Lingxia Zhao, and Lida Zhang. A computational interactome for prioritizing genes associated with complex agronomic traits in rice. Plant Journal. 2017. DOI:10.1111/tpj.13475. Article