Performance of different classification algorithms for the prediction of TTSS effectors.

The performance of different classification algorithms in a tenfold cross-validation on the animal pathogen and plant symbiont training set is shown. The cross-validation has been repeated five times with different negative sets that were randomly chosen from the respective organisms. Sensitivity (defined as TP/[TP+FP]), selectivity (defined as TN/[TN+FP]), and the AUC value are given with their standard-deviation (sd) computed from the five runs.