Table 1

False positive predictions. The table shows the rate of false positive predictions on Gram-positive, and Gram-negative proteins which are secreted via the Sec-dependent pathway. Furthermore is shows cytoplasmic proteins and transmembrane helix proteins, both carrying two consecutive arginines within the initial 30 amino acids. Used methods for prediction are regular expression (Regex), TATFIND and TatP (regular expressions in combination with a neural network (NN)).

Method

Gram+

Gram-

Cyt. (RR)

Transmem (RR)


Regex

4%

7%

15%

23%

TATFIND

1%

6%

1.4%

7.0%

TatP (Regex+NN)

3%

5%

0.4%

3.5%


Bendtsen et al. BMC Bioinformatics 2005 6:167   doi:10.1186/1471-2105-6-167

Open Data