An effective approach for annotation of protein families with low sequence similarity and conserved motifs: identifying GDSL hydrolases across the plant kingdom

Table 1 Comparison of VD and PD/VD performance based on visual inspection of 300 of the top ranked sequences for each proteome

Organism	Method	TP	FP	FN	FDP (%)	S (%)
A. thaliana	VD	116	8	0	6.5	100.0
A. thaliana	PD/VD	116	4	0	3.3	100.0
S. bicolor	VD	105	6	0	5.4	100.0
S. bicolor	PD/VD	104	2	1	1.9	99.0
O. sativa	VD	90	11	1	10.9	98.9
O. sativa	PD/VD	90	6	1	6.3	98.9
P. patens	VD	40	0	3	0.0	93.0
P. patens	PD/VD	37	0	6	0.0	86.0
P. trichocarpa	VD	95	14	3	12.8	96.9
P. trichocarpa	PD/VD	95	11	3	10.4	96.9
V. vinifera	VD	80	10	0	11.1	100.0
V. vinifera	PD/VD	80	6	0	7.0	100.0
Σ208,298 sequences	VD	526	49	7	8.5	98.7
Σ208,298 sequences	PD/VD	522	29	11	5.3	97.9

The total number of entries in six proteomes and the numbers of GDSL identified by VD and PD/VD as well as respective statistical measures are shown in bold. TP true positive, FP false positive, FN false negative, S sensitivity [S = TP/(TP + FN)], FDP false discovery proportion [FDP = FP/(FP + TP)]; TPs that were eliminated by PD/VD protocol were added to the FN group

ISSN: 1471-2105