Skip to main content
. 2012 Sep 12;7(9):e45152. doi: 10.1371/journal.pone.0045152

Table 4. Comparison among the tri-peptide subsequence models with or without propensity.

Kernels Sn (%) P (%) F-measure
Tri-peptide Propensity only N.A. 56.5±12.5 61.0±6.3 0.584±0.085
Tri-peptide w./o. Propensity Blosum62 54.5±6.5 60.5±1.5 0.573±0.035
PAM160 55.0±7.2 61.1±1.8 0.578±0.040
w./Propensity Blosum 62 * 80.1± 2.1 55.2± 1.0 0.693± 0.060
PAM160 69.3±10.0 58.5±3.5 0.633±0.050
AA_A pattern w./o. Propensity Blosum62 54.8±6.8 60.5±1.5 0.579±0.040
PAM160 55.2±7.1 61.3±2.0 0.577±0.045
w./Propensity Blosum62 60.5±5.5 57.5±2.5 0.589±0.040
PAM160 59.5±5.5 57.5±1.5 0.585±0.035
A_AA pattern w./o. Propensity Blosum62 55.5±8.5 60.6±2.2 0.581±0.050
PAM160 55.2±8.1 60.5±1.5 0.577±0.055
w./Propensity Blosum62 60.5±6.5 57.5±1.5 0.590±0.040
PAM160 59.5±5.5 57.5±1.5 0.585±0.025
*

The corresponding model is defined as SVMTriP.