Mapping‐based temporal pattern mining algorithm (MTPMA) identifies unique clusters of phosphopeptides regulated by vasopressin in collecting duct

Abstract

We developed a new algorithm, MTPMA, for clustering time‐course patterns following step‐inputs in biological systems. We tested the algorithm with data from large‐scale quantitative phosphoproteomics experiments done as follows: Inner medullary collecting duct (IMCD) samples were incubated in the presence or absence of 1nM dDAVP (vasopressin) for 0.5, 2, 5, and 15 minutes (N=3 pairs) followed by LC‐MS/MS‐based phoshoproteomic analysis. Quantification used 8‐plex iTRAQ and commercially available software. Of the 12,533 phosphopeptides identified, 3,298 were found in at least 2 out of 3 time courses and had quantifiable iTRAQ ratios. These phosphopeptides were analyzed with MTPMA in order to identify groups that changed in abundance with similar temporal responses after vasopressin addition. The algorithm maps the data from a Cartesian plane to a discrete binary plane and uses an efficient dynamic programming technique to mine similar patterns after mapping. The mapping allows clustering of similar time‐courses that are temporally closer to each other than to other time‐courses. The algorithm identified 30 clusters of phosphopeptides with distinct temporal profiles, including peptides that significantly increased or decreased in abundance in response to vasopressin. These time‐course clusters provide a starting point for modeling the signaling network involved.