Fig. 4.

TM-scores of the predicted structure models by using the homologs (i.e. structure templates) found by HHsearch, RaptorX and ENTS versus using those found by CMsearch: the black points represent the cases where both methods can predict high-quality structures (with TM-scores above 0.6), which should be considered as the easy cases; the gray points represent the opposite cases where both methods cannot predict high-quality models; the red points represent the cases where the competing method cannot predict any high-quality models but CMsearch can; and the blue points represent the opposite cases where CMsearch cannot predict any high-quality models while the competing method can (statistics related to these figures are shown in Table 2)