Table 5

Validation of selection strategy using a variety of prediction tools

Group

Resulting proteins

Hypothetical proteins

Characterized proteins

Removed proteins

Characterized mitochondrial proteins


TargetP + SMART

(Grp I)

20

30%

(6 of 20)

65%

(13 of 20)

5%

(1 of 20)

85%

(11 of 13)

TargetP + SMART

(Grp III)

100

36%

(36 of 100)

53%

(53 of 100)

11%

(11 of 100)

45%

(24 of 53)

MITOPRED + SMART

198

34%

(68 of 198)

57%

(113 of 198)

8%

(16 of 198)

25%

(28 of 113)

WoLF PSORT + SMART

154

31%

(48 of 154)

61%

(94 of 154)

7%

(11 of 154)

28%

(26 of 94)

TargetP + Prosite

(Reliability class A)

9

11%

(1 of 9)

89%

(8 of 9)

0%

100%

(8 of 8)


2006 dataset

-

25%

(1455 of 5860)

21%

(1215 of 5860)

54%

(3190 of 5860)

6%

(67 of 1215)


Different combinations of prediction tools were used on either the whole 2006 dataset or parts of it, to demonstrate that our selection strategy can use a variety of prediction tools and is neither dependent on TargetP nor the SMART program.

Desler et al. BMC Bioinformatics 2009 10:289   doi:10.1186/1471-2105-10-289

Open Data