Application of amino acid occurrence for discriminating different folding types of globular proteins

BMC Bioinformatics

Table 3 Performances with independent dataset Predictive ability [%]of our method to the independent dataset of proteins used in Ding and Dubchak [18]. wo: without re-weighting, w: with re-weighting

	Number	Ratio	Sensitivity		Precision		F1
Fold Description		[%]	wo	w	wo	w	wo	w
Cytochrome C	16	3	56	94	64	47	60	63
DNA/RNA binding 3-helical bundle	32	6	75	56	41	47	53	51
Four helical up and down bundle	15	3	33	33	71	42	45	37
EF hand-like fold	15	3	53	53	57	42	55	47
Immunoglobulin-like β-sandwich	74	14	66	31	44	68	53	43
Cupredoxin-like	21	4	29	38	50	33	36	36
Concanavalin A-like lectins/glucanases	13	2	38	38	42	33	40	36
SH3-like barrel	16	3	0	50	-	44	-	47
OB-fold	32	6	16	28	26	31	20	30
TIM barrel	77	14	40	25	66	70	50	37
FAD/NAD: (P)-binding domain	23	4	22	30	114	50	37	38
Flavodoxin-like	24	5	8	13	28	35	13	18
NAD: (P)-binding Rossmann-fold domains	40	8	40	35	5	8	8	13
P-loop containing nucleoside triphosphate hydrolases	22	4	23	18	38	50	29	27
Thioredoxin fold	17	3	18	35	33	25	23	29
Ribonuclease H-like motif	22	4	5	18	14	22	7	20
α/β-Hydrolases	18	3	33	39	43	41	38	40
β-Grasp, ubiquitin-like	15	3	0	33	0	20	-	25
Ferredoxin-like	40	8	23	3	11	10	15	4
Total/Mean	532		31	35	42	38	34	34
Accuracy
without reweighting	36
with reweighting	32

ISSN: 1471-2105