Table 4

Validation of AuDIT.

Dataset

Annotation

TN

TP

FN

FP

Overall Accuracy (%)

Sensitivity (%)

Specificity (%)


10 Peptide Standard Curve, 3 transitions MultiQuant

Site 1

Global

89

11

119

29

33

77

80

73

Focused

7

144

1

8

97

99

94


Site 2

Global

9

217

14

30

84

94

23

Focused

23

247

0

0

100

100

100


Site 3

Global

19

200

33

18

81

86

51

Focused

50

218

2

0

99

99

100


Site 4

Global

21

162

74

13

68

69

62

Focused

81

174

14

1

94

93

99


10 Peptide Standard Curve, 3 transitions, Skyline

Site 1

Global

29

163

35

43

71

82

40

Focused

56

206

8

0

97

96

100


Site 2

Global

1

210

15

44

78

93

2

Focused

15

254

1

0

100

100

100


Site 5

Global

35

34

2

199

26

94

15

Focused

37

232

0

1

100

100

97


10 Peptide Standard Curve, 5 transitions, MultiQuant

Site 6

Global

46

16

277

122

23

69

69

67

Focused

8

294

0

6

99

100

97


Clinical Samples, 3 transitions, MultiQuant

Cardio-vascular Peptides

Global

4

33

5

9

73

87

31

Focused

9

40

0

2

96

100

82


For each dataset, two contingency matrices are calculated. The 'pre-test' evaluation by the expert identifies overall data problems like poor chromatography, inaccurate peak integration, etc. Comparison of this global annotation with the algorithm calls results in one set of contingency matrices (shown under Annotation = Global). The second 'post-test' re-evaluation is based on the algorithm outcome, and accounts for the fact that the global annotation could be overly conservative (i.e., mark too many transitions as BAD). This focused annotation is compared with the algorithm-derived decisions to derive a second, algorithm-guided set of contingency matrices, shown under Annotation = Focused. TN: True Negative, TP: True Positive, FN: False Negative, and FP: False Positive. Overall Accuracy = (TP + TN)/(TP + TN + FN + FP). Sensitivity = TP/(TP + FN). Specificity = TN/(TN + FP). A transition is BAD if it has some form of interference, i.e., it is imprecise of inaccurate. If not, the transition is labeled as GOOD. Adapted from Abbatiello, Mani, et. al., Clinical Chemistry, 56, 291-305 [17].

Mani et al. BMC Bioinformatics 2012 13(Suppl 16):S9   doi:10.1186/1471-2105-13-S16-S9

Open Data