Abstract: Many secreted proteins of plant pathogens have shown to be the elicitor and the pathogenetic factors for interacting with the plant receptors. Analysis of the secretome by utilizing genomic database information and computer prediction algorithms would facilitate to clarify the molecular mechanism on the interaction between plant and plant pathogens. To investigate the function of secreted proteins in Magnaporthe grisea, we used a set of prediction algorithms to predict the secreted proteins encoded M.grisea genome. First, the presences or absences of a N-terminal signal peptide of all 11108 ORFs from M.grisea were predicted by the SignalP program and 2468 ORFs were found to encode proteins with N- terminal signal peptide. Second, Protcomp program were used to predict the sub-cellular localization of these 2468 ORFs, and 1858 proteins were extracellularly localized. Then, the 1858 proteins were further predicted by using TMHMM program to move the proteins with more than one transmembrance domains, big-PI Predictor to move the GPI-anchored proteins and TargetP to eliminate proteins with mitochondrial targeting signals step by step. Finally, 1235 ORFs were predicted to be secreted proteins from M.grisea genome.

Key words: Magnaporthe grisea, secreted proteins, signal peptide, prediction algorithm