Detecting Secreted Peptides in Plants Using SPADA (Secreted Peptide Alignment Discovery Algorithm)

Motivation: Small, secreted peptides have recently been shown to affect many numerous aspects of plant development, reproduction and defense responses. As one of the most highly represented clas-ses of secreted peptides, the cysteine-rich peptides (CRPs) are also suggested to play multiple roles as signaling factors in developmen-tal patterning as well as pathogen responses and symbiosis. Due to the inherent limitations of general-purpose gene finding algorithms, however, both the prediction and proper annotation of these small, secreted peptides in plants remain difficult tasks, and various cases of mis-annotation exist in popular plant genome databases.
Results: We present a homology-based gene prediction program to accurately predict secreted peptides in plants. Given a high-quality profile alignment for each CRP class, SPADA identifies and anno-tates nearly all CRPs in tested plant genomes, with a better perfor-mance than all general-purpose gene prediction programs surveyed. We also show that SPADA is readily generalizable to other classes of secreted peptides (e.g., Self-incompatibility protein homologues) simply by substituting the appropriate sets of HMMs. Predictions made by SPADA should constitute the most complete set of plant cysteine-rich peptides, and in this regard, will provide an invaluable resource for the research of small, secreted peptides in plants.
Availability and Implementation: SPADA is a perl-based pipeline that internally runs HMMer, Augustus, Genewise, SplicePredictor and a number of custom scripts. The program is freely available through Google code under a GPL license. Instructions for down-loading and installation are available at https://code.google.com/p/medicago/wiki/SPADA_Readme.