WorkshopsHuman Protein-Protein Interaction Prediction by A Novel Sequence-Based Coevolution Method: Coevolutionary Divergence Method
81
reads
Protein-protein interaction (PPI) plays an important role in understanding gene functions and many computational PPI prediction methods are proposed in recent years. Despite the extensive efforts, the prediction of PPI still has much room to improve. Sequence-based coevolution methods include the substitution rate method and the mirror tree method, which compare sequence substitution rates and topological similarity of phylogenetic trees respectively. Although they have been used to predict PPI in small genomes like Escherichia coli, such methods have not been tested in large scale genome like Homo sapiens. In this study, we propose a novel sequence-based coevolution method, the coevolutionary divergence method, to predict PPI in human genome. Our method assumes that protein pairs with similar substitution rates are likely to interact with each other. We integrate the evolutionary information from 14 species of vertebrates by a naïve Bayes approach to infer PPI. The result suggests that the coevolutionary divergence method outperforms the mirror tree method in three independent PPI datasets of human. The coevolutionary divergence method has the potential to further improve PPI prediction with the increasingly more species genome information generated by next generation sequencing.
reads
Shin-Sheng Yuan
2012-07-24
14:20:00 - 15:10:00
308 , Mathematics Research Center Building (ori. New Math. Bldg.)
Protein-protein interaction (PPI) plays an important role in understanding gene functions and many computational PPI prediction methods are proposed in recent years. Despite the extensive efforts, the prediction of PPI still has much room to improve. Sequence-based coevolution methods include the substitution rate method and the mirror tree method, which compare sequence substitution rates and topological similarity of phylogenetic trees respectively. Although they have been used to predict PPI in small genomes like Escherichia coli, such methods have not been tested in large scale genome like Homo sapiens. In this study, we propose a novel sequence-based coevolution method, the coevolutionary divergence method, to predict PPI in human genome. Our method assumes that protein pairs with similar substitution rates are likely to interact with each other. We integrate the evolutionary information from 14 species of vertebrates by a naïve Bayes approach to infer PPI. The result suggests that the coevolutionary divergence method outperforms the mirror tree method in three independent PPI datasets of human. The coevolutionary divergence method has the potential to further improve PPI prediction with the increasingly more species genome information generated by next generation sequencing.