Skip to main content

Table 3 Effect of different cutoff values for RF on RATE results. A cutoff of RF ≥ 1.3 gave the best results with MCC = 0.314

From: Experimental validation of the RATE tool for inferring HLA restrictions of T cell epitopes

Cutoffs

TP

FP

FN

TN

Total

Accuracy

Sensitivity

Specificity

Precision

False Positive Rate

Matthews Correlation Coefficient

RF ≥ 1.00

34

44

6

18

102

0.510

0.850

0.290

0.436

0.710

0.162

RF ≥ 1.20

33

36

7

26

102

0.578

0.825

0.419

0.478

0.581

0.255

RF ≥ 1.25

33

35

7

27

102

0.588

0.825

0.435

0.485

0.565

0.270

RF ≥ 1.30

33

32

7

30

102

0.618

0.825

0.484

0.508

0.516

0.314

RF ≥ 1.35

31

31

9

31

102

0.608

0.775

0.500

0.500

0.500

0.275

RF ≥ 1.40

31

30

9

32

102

0.618

0.775

0.516

0.508

0.484

0.290

RF ≥ 1.45

30

29

10

33

102

0.618

0.750

0.532

0.508

0.468

0.279

RF ≥ 1.50

29

27

11

35

102

0.627

0.725

0.565

0.518

0.435

0.284

RF ≥ 1.75

26

24

14

38

102

0.627

0.650

0.613

0.520

0.387

0.257

RF ≥ 2.00

21

22

19

40

102

0.598

0.525

0.645

0.488

0.355

0.168

RF ≥ 2.50

16

15

24

47

102

0.618

0.400

0.758

0.516

0.242

0.168