Fig. 4

Accuracy distribution of 100 models learned independently from the two-class and three-class MIV (A and B, respectively) and IV (C and D, respectively) datasets using OneR (1R), JRip (JR) and PART (PT). The datasets contain either the concatenated alignments of all IAV proteins or individual alignment of PB2, PB1, PA, HA, NP, NA, M1, NS1, PB1-F2, PA-X, M2 or NS2 proteins. The red dashed horizontal line indicates the accuracy of zero rule learner, while the blue horizontal lines indicate significant difference (Tukey’s HSD adjusted p-value < 0.05) between two virulence models generated from the same protein alignment. Information about significant differences between all possible pairs of virulence models in each plot can be found in Figure S3 (Additional file 3)