Table 1. Performance of CLADE, CLADE_BEv, HMMScan and HHblits on a SCOP dataset of sequences.

ASTRAL10		Mono-source	Multi-source
		HMMScan	CLADE	CLADEBEv
	TP	24	217	165
	FP	10	89	141
	FN	272	0	0
	F-measure	0.15	0.83	0.70
	PPV	0.71	0.71	0.54
	Sen	0.08	1	1
		HHblits	CLADE_HHblits	CLADEBEvHHblits
	TP	29	300	289
	FP	144	5	16
	FN	133	1	1
	F-measure	0.17	0.99	0.97
	PPV	0.17	0.98	0.95
	Sen	0.18	1	1
ASTRAL30		Mono-source	Multi-source
		HMMScan	CLADE	CLADEBEv
	TP	430	994	741
	FP	127	229	471
	FN	694	28	39
	F-measure	0.51	0.89	0.74
	PPV	0.77	0.81	0.61
	Sen	0.38	0.97	0.95
		HHblits	CLADE_HHblits	CLADEBEvHHblits
	TP	476	1241	1150
	FP	502	7	97
	FN	318	3	4
	F-measure	0.54	1	0.96
	PPV	0.49	0.99	0.92
	Sen	0.6	1	1
ASTRAL95		Mono-source	Multi-source
		HMMScan	CLADE	CLADEBEv
	TP	7000	8512	7772
	FP	710	117	857
	FN	923	4	4
	F-measure	0.9	0.99	0.95
	PPV	0.91	0.99	0.90
	Sen	0.88	1	1
		HHblits	CLADE_HHblits	CLADEBEvHHblits
	TP	3934	8537	7947
	FP	1679	96	686
	FN	3020	0	0
	F-measure	0.63	0.99	0.96
	PPV	0.7	0.99	0.92
	Sen	0.57	1	1

SCOP datasets ASTRAL95, ASTRAL30, ASTRAL10 contain sequences with at most 95%, 30%, 10% of sequence identity.