Prediction Of The PPI In The Human Integrin Adhesom Based On A Domain Approach

The interactions of integrin adhesome are vital to a number of biological processes. At present, with the development of bioinformatics, there have been a large number of data on the interactions of integrin adhesome and methods of predicting Protein–Protein Interactions(PPIs). In particular, the prediction of PPIs based on domain information is becoming one of the hot research fields in recent years. However, because of the first step of these methods mostly was to get the possible interacting protein pairs at the protein level, and then, they built predictive models with the relationship between protein pairs and domain information, so there would be some high false negative predictions in these methods. In order to solve the problem of these high false negative prediction results, we proposed a method for the prediction of PPIs, which obtain possibly interactional protein pairs at the domain level. Then, we used this method to predict the interactions of integrin adhesome. We also built the Protein–Protein Interaction(PPI) networks, and predicted the signal transduction pathways of integrin adhesome. So we can accomplish the purpose of understanding the mechanisms of PPIs based on a domain method, at the molecular level.Firstly, with the Domain–Domain Interaction(DDI) information, a model for prediction of PPIs was established via using the interactions between the proteins and the domains. We took a set of certified PPI data as the benchmark to optimize the parameters of our prediction model. In addition, we also contrasted our optimized model with other PPI predicting methods, to prove the predicting accuracy and sensitivity of our model.Secondly, we predicted the PPIs of a set of source data, which contains 147 integrin adhesome, and we got a total of 736 PPIs and their corresponding confidence probabilities by using our predictive model. Then, based on these predicted PPIs, we constructed a protein interaction network. Furthermore, according to the directions of these PPIs, we constructed the undirected graph and the directed network respectively, and then we analyzed these integrin adhesome interactions in the form of network graphs.Thirdly, by using the 736 predicted interactions of integrin adhesome, we constructed a weighted PPI network, in which the confidence probabilities of PPIs were taken as the edge weights. Then, the pathways of the minimum weight were identified via a dynamic programming algorithm. Further, according to certain rules, all pathways that the algorithm calculated were aggregated into some probable pathway networks. In such a way, we can predict signal transduction pathways from the interaction network of integrin adhesome. At last, seven signal transduction pathways were obtained.In summary, this study proposed a method for predicting protein interactions based on DDI information. The result of this method had a lower false negative rate and a higher reliability. A set of interactions of integrin adhesome was successfully predicted via using this method. Moreover, our study also constructed a PPI network, and predicted the signal transduction pathways of the integrin adhesome from the network. This could provide references not only for experimental study of the mechanism of integrin-related signal transduction pathways, but also for exploring disease mechanism in the research field of basic medical sciences.