Network inference from temporal phosphoproteomics informed by protein-protein interactions

Abstract

Network inference from time-course data holds the promise to overcome challenges associated with other methods for deciphering cell signaling networks. Integration of protein-protein interactions in this process is frequently employed to limit wiring options. In this study, a graph approach for the analysis of data of high temporal resolution is introduced and applied to a 5 s resolution phosphoproteomics dataset. Steiner trees informed by protein-protein interactions are constructed on individual time slices, which are then stitched together into a temporal signaling network. Systematic benchmarking against existing knowledge indicates that the approach enriches signaling-relevant edges. The workflow is compatible with future extensions for reliably extracting extended signaling paths.