Figure 2

VirulentPred predictions in the various ranges of SVM scores. For generating the plot, 367 sequences of both the independent datasets (consisting of 181 virulent and 186 non-virulent proteins) were classified using VirulentPred. The number of false positive prediction was very high at a threshold SVM scores in the range 0.8 to 1.1. Hence, to reduce false positive prediction, a stringent criterion of threshold value of ≥1 was used for the annotation of complete proteomes of pathogens.