. 2013 Mar 5;8(3):e58173. doi: 10.1371/journal.pone.0058173

Table 3. Performance comparison between T3_MM, BPBAac and Effective T3 on new datasets.

Dataset	Software	Recall (%)^a	Selectivity (%)^b	A (%)
Mukaihara2010	T3_MM	32/35 (91.43)	64/70(91.43)	91.43
	BPBAac	21/35 (60.00)	67/70(94.87)	83.81
	Effective T3	20/35(57.14)	70/78(92.86)	80.95
Baltrus 2011	T3_MM	275/291 (94.50)	539/582 (92.61)	93.24
	BPBAac	234/291 (80.41)	558/582 (95.88)	90.72
	Effective T3	223/291 (76.63)	533/582 (91.58)	86.60

Note: ^a‘Recall’ was adopted here instead of sensitivity to describe the number of validated T3S proteins correctly predicted from the total number of T3S proteins. The recall percentage was noted within parentheses after recall value, which was identical to the sensitivity.

‘Selectivity’ was adopted here instead of specificity, to describe the number of non-T3S proteins correctly predicted from the total number of non-T3S proteins. The selectivity percentage was noted within parentheses after selectivity value, which was identical to the specificity.