Table 4

Comparison of predictive performance of SVM based on different encoding input information. More details for prediction accuracy measurement are given in the Methods section. The results were obtained by 5-fold cross-validation.
Methods
 Prediction accuracy (%)


Q2
 MCC
 Sensitivity
 Specificity
LSa
 62.8
 0.26
 56.6
 68.7
AAb
 61.6
 0.23
 59.8
 63.2
MSc
 69.8
 0.40
 70.5
 68.7
SSd
 63.6
 0.27
 57.8
 69.3
MS+SSe
 71.5
 0.43
 70.7
 72.2

aLS: prediction performance for the local sequence encoding scheme;

bAA: prediction performance for the amino acid composition encoding scheme of local sequence;

cMS: prediction performance for the multiple sequence alignment encoding scheme in the form of PSI-BLAST profile;

dSS: prediction performance for the predicted secondary structure encoding scheme by PSIPRED;

eMS+SS: prediction performance for the multiple sequence alignment plus secondary structure encoding scheme.

Song et al. BMC Bioinformatics 2006 7:124   doi:10.1186/1471-2105-7-124