Skip to main content

Table 2 Ratio between the empirical type I error and the given significance level estimated over 105 simulations under common variants

From: A comprehensive comparison of multilocus association methods with summary statistics in genome-wide association studies

Method

Significance level α

Performance of type I error control

0.05

0.01

0.001

Average

Inflated

Well-controlled

Conservative

MLR

0.00

0.00

0.00

0.00

  

√

FLM

0.00

0.00

0.00

0.00

  

√

HC

1.33

1.82

2.33

1.83

√

  

GHC

1.26

1.65

1.94

1.62

√

  

BJ

1.29

1.64

1.97

1.63

√

  

GBJ

0.85

1.32

1.71

1.29

√

  

DOT

0.00

0.00

0.00

0.00

  

√

BT

1.04

1.07

1.10

1.07

 

√

 

SKAT-O

1.08

1.18

1.11

1.12

 

√

 

SKAT

1.02

1.08

1.08

1.06

 

√

 

Simes

0.82

0.82

0.82

0.82

 

√

 

FCP

5.29

21.88

174.81

67.33

√

  

TPM

2.45

10.39

86.81

33.22

√

  

RTP

3.76

14.71

110.07

42.85

√

  

minP

0.88

0.82

0.77

0.82

 

√

 

ART

4.15

16.51

126.91

49.19

√

  

ART-A

1.17

3.05

12.97

5.73

√

  

GM

2.01

7.43

52.03

20.49

√

  

SimpleM

0.39

0.41

0.41

0.40

  

√

GATES

1.47

1.53

1.51

1.50

√

  

HMP

0.87

1.01

1.06

0.98

 

√

 

ACAT

1.04

1.08

1.07

1.06

 

√

 
  1. Determine whether a SNP-set method was inflated, well-controlled or conservative according to the average ratio between the empirical type I error and the given significance level over 105 simulations. inflated: ratio > 1.2; well-controlled: 0.8 ≤ ratio ≤ 1.2; conservative: ratio < 0.8