Bridging neuropeptidomics and genomics with bioinformatics: Prediction of mammalian neuropeptide prohormone processing

Neuropeptides are an important class of cell to cell signaling molecules that are difficult to predict from genetic information because of their large number of post-translational modifications. The transition from prohormone genetic sequence information to the determination of the biologically active neuropeptides requires the identification of the cleaved basic sites, among the many possible cleavage sites, that exist in the prohormone. We report a binary logistic regression model trained on mammalian prohormones that is more sensitive than existing methods in predicting these processing sites, and demonstrate the application of this method to mammalian neuropeptidomic studies. By comparing the predictive abilities of a binary logistic model trained on molluscan prohormone cleavages with the reported model, we establish the need for phyla-specific models.