Table 2. Models of human TFs.

%ID range	Reliable models	Reliable models (cumulative)	TF2DNA performance	Correct targets	Correct targets (cumulative)
≥90	90	90	74.8	67	67
80–90	54	144	77.8	42	109
70–80	60	204	81.2	49	158
60–70	246	450	78	192	350
50–60	312	762	76.8	240	590
40–50	200	962	81.8	164	753
30–40	188	1150	49	92	845
<30	171	1248	45.2	78	923

The total number of targeted TF sequences is 1825 (obtained from list of Ensembl identifiers provided by Vaquerizas et al. (43)). First column shows the percent sequence identity between human TF target and their structural template. The fourth column shows the fraction of successful predictions by TF2DNA (at 95% confidence level) (from Supplementary Table S4). The fifth column shows the estimated numbers of correctly predicted targets, which is obtained by multiplying the number of reliably modeled targets (second column) and the calculated TF2DNA performance (fourth column).