Skip to main content

Table 2 Statistics for PLS-DA, OPLS-DA model, details for predictive and orthogonal parts; ROC analysis

From: Platelet protein biomarker panel for ovarian cancer diagnosis

A) PLS-DA-based analysis of protein spots expression in 2D

 Component

Latent variables

R2X (cum)

Q2 (cum)

CV-ANOVA, p-value

permutation test, p-value

  

Sensitivity

Specificity

 Model

3

0.318

0.72

8,5 * 10–9

<0.001

 

calibration set

0.96

0.88

       

validation set

1.00

0.44

B) OPLS-DA-based analysis of protein expression in western blot.

 Component

Latent variables

R2X (cum)

R2 (cum)

Q2 (cum)

CV-ANOVA, p-value

permutation test, p-value

 

Sensitivity

Specificity

 Model

1 + 1

0.203

0.632

0.477

4.41E-14

<0.001

calibration set

0.83

0.89

 Predictive

1

0.0717

0.632

0.477

  

validation set

0.88

not tested

 Orthogonal

1

0.131

0

      

C) ROC analysis of protein expression in western blot.

 Compared groups

AUC

standart deviation

95% confidence interval

z statistics

p-value

  

Sensitivity

Specificity

 ROC 1

0.777

0.0418

0,695 to 0,859

6.639

<0,0001

 

ROC1

60

83.33

 ROC 2

0.831

0.0501

0,733 to 0,930

6.615

<0,0001

 

ROC2

83.33

76.19

D) OPLS-DA-based analysis of protein expression in Digi west.

 Component

Latent variables

R2X (cum)

R2 (cum)

Q2 (cum)

CV-ANOVA, p-value

permutation test, p-value

 

Sensitivity

Specificity

 Model

1 + 2

0.24

0.785

0.345

4.50E-03

<0.001

test set

0.7

0.83

 Predictive

1

0.037

0.785

0.345

     

 Orthogonal

2

0.203

       
  1. R2X cumulative percentage of X variance explained, R2 cumulative percentage of Y variance explained, Q2 cumulative percentage of variance of Y predicted, CV-ANOVA p-value p-value of cross-validation ANOVA, permutaion test p-value p-value of (1000 iterations) permutation test