Skip to main content

Table 2 Distributions of probes per cluster in the TAD and sliding window clusters

From: Using LDpred2 to adapt polygenic risk score techniques for methylation score creation

Window size

Probes clustered*

Min

Q1

Q2

Mean

Q3

Max

p-value

Nagel-

kerke R2

AIC

5kb

74%

1

1

2

2.31

3

91

1.88 × 10− 8

0.0361

1596.5

10kb

78%

1

1

2

2.50

3

119

1.83 × 10− 8

0.0361

1596.4

20kb

81%

1

1

2

2.71

3

119

1.85 × 10− 8

0.0362

1596.3

100kb

85%

1

1

2

3.03

4

119

1.85 × 10− 8

0.0362

1596.3

500kb

86%

1

1

2

3.09

4

119

1.84 × 10− 8

0.0361

1596.3

1Mb

86%

1

1

2

3.09

4

119

1.86 ×10− 8

0.0361

1596.3

TAD clusters

99.99%

1

45

82

118.46

138

2690

1.68 × 10− 8

0.0363

1596.2

  1. *The percentage of probes included in non-singleton clusters
  2. TAD: topologically associating domain
  3. The increase in probes clustered diminishes with larger window sizes and the percentage is constant above 500kb. The distributions change very little after 10kb. The differences in probe distributions between the sliding window and TAD clusters did not have any significant effect on the methylation scores’ performance