Uncovering Arabidopsis Membrane Protein Interactome Enriched in Transporters Using Mating-Based Split Ubiquitin Assays and Classification Models

High-throughput data are a double-edged sword; for the benefit of large amount of data, there is an associated cost of noise. To increase reliability and scalability of high-throughput protein interaction data generation, we tested the efficacy of classification to enrich potential protein–protein interactions. We applied this method to identify interactions among Arabidopsis membrane proteins enriched in transporters. We validated our method with multiple retests. Classification improved the quality of the ensuing interaction network and was effective in reducing the search space and increasing true positive rate. The final network of 541 interactions among 239 proteins (of which 179 are transporters) is the first protein interaction network enriched in membrane transporters reported for any organism. This network has similar topological attributes to other published protein interaction networks. It also extends and fills gaps in currently available biological networks in plants and allows building a number of hypotheses about processes and mechanisms involving signal-transduction and transport systems.This work was supported in part by grants from Junta de Castilla y León (GR147) and Ministerio de Ciencia y Innovación (BIO2008-00194) to Jose Luis Revuelta and grants from the National Science Foundation (Arabidopsis2010-0618402 and MCB-1052348) to Seung Y. Rhee and Wolf B. Frommer.Peer reviewedPeer Reviewe