Detecting distant-homology protein structures by aligning deep neural-network based contact maps
Table 2
Threading results obtained by different methods for the 539 proteins from Benchmark Set-II.
P-values were calculated between the CEthreader alignment TM-scores and other methods’ TM-scores using one-sided Wilcoxon signed-rank tests; NST represents the number of targets with a first identified template whose TM-score was >0.5; the success rate is equal to the fraction of targets whose Folds were correctly recognized, i.e., the first identified template had the same Fold as the query.