A new method combining quantitative mass spectrometry and Bayesian inference for the analysis of protein-protein-interactions
Knowledge on the function of individual proteins is fundamental to the understanding of biological systems. A fruitful approach to this end is the identification of proteins interacting with the unknown protein of interest, as according to the ‘guilt-by-association’ principle, they may reveal important hints towards its function. A frequently used method for the identification of protein-protein interactions are pull-down assays. Here, a bait protein is precipitated together with its interaction partners from cell lysates and precipitates are then analyzed by mass spectrometry (MS). The high sensitivity of today’s mass spectrometers usually results in lists of hundreds of proteins. Of these only a small fraction are true interaction partners, while the majority represents contaminants, which were co-precipitated because they unspecifically bind to the beads or cross-react with antibodies used. Hence, discriminating between true interaction partners and contaminants is a major challenge which to address we intend to establish a new method based on quantitative mass spectrometry. Here, combining an intelligent mixing strategy in pull-down assays with stable isotope labeling will allow a robust contamination estimation based on Bayesian probabilistic modelling. With this model, target probabilities can be inferred from the data leading to a stochastic classiﬁcation of true interaction partners and contaminants. In this project we will apply the new method to analyse the interactome of five selected proteins that are involved in processes associated with thylakoid membranes.