IRESPred

Validation:

The performance of IRESPred was validated using receiver operating characteristic (ROC) analysis.

The sequences in test data sets were used to assess the predictive performance of IRESPred. The summary of ROC analysis is presented in following table.


Model

Optimum parameters used in model building*

Performance measures used in model evaluation§

s

t 

d

g

c

CV (%)

Acc (%)

Sn (%)

Sp (%)

Pr (%)

MCC

1

2

0

1

3.1192

0.0347

63.54

75.51

75.75

75.25

75.75

0.51

2

1

1

1

1.8022

0.0347

68.75

63.44

63.44

63.44

63.44

0.26

3

0

1

2

1.0050

1.0397

67.19

62.36

62.36

62.36

62.36

0.24

4

2

0

1

3.1192

0.0347

69.79

65.05

61.30

68.82

66.28

0.30

5

1

1

1

2.2181

0.0347

67.71

61.83

60.22

63.44

62.22

0.23


* s: SVM type, t: kernel type, d: degree, g: gamma, c: cost and CV: 10-fold cross validation accuracy. Parameters as specified by svm-train program in LibSVM3.12 package.
§ Acc: accuracy, Sn: sensitivity, Sp: specificity, Pr: precision and MCC: Matthews correlation coefficient.

The model highlighted in blue is implemented in IRESPred.