Table 1.

Characteristics of all four interaction datasets used

	B.anthracis	F.tularensis	Y.pestis	S.typhi
Total no. of bacterial proteins (‘reviewed’ protein set from UniprotKB)	2321	1086	4600	3592
Total no. of human–bacteria protein pairs	59.4 M	27.8 M	117.7 M	87.7 M
No. of known interactions	3073	1383	4059	62
No. of interactions with no missing features	655	491	839	62
Size of training data with 1:100 class ratio	66 155	49 591	84 739	6262
No. of unique features in the training data	69 4715	468 955	886 480	349 155

Note: Total no. of human proteins: 25 596; M, million. For each host–pathogen PPI dataset, the number of pathogen proteins, the size of the dataset and other such statistics are shown.